313 videos so far; the round is still running and this page updates every few minutes. Last updated Nov 4, 8:06 PM PST. Grouped by topic; each section says what it is testing and each clip says what it is trying to show. Thumbnails are small previews; click any item to open the full-size original (untouched file, same resolution). In the viewer, "1:1" shows actual pixels. ← gallery home
Text to video: styles and motion V1V1-H 83 videos
What we're testing: Can a model make a convincing clip from a text prompt alone? V1 is the fast grid: 17 prompts (walking, hip-hop dance, gymnastics, breakdancing, a sword duel and a hand-to-hand fight, plus looks like anime, 35 mm cinema, 1950s newsreel, 1980s sitcom, 1970s sci-fi TV and VHS) on Wan 2.2 14B Lightning and HunyuanVideo 1.5. V1-H reruns the best 10 prompts at full quality: Wan 2.2 14B with full steps (2 seeds) and HunyuanVideo 1.5 (1 seed). Each group below is one prompt, so the fast and full-quality versions sit side by side. · V1 notesV1-H notes
Reviewed 20 temporal samples plus full-resolution 1.75-second frame. Clearly anime-rendered fighters retain distinct faces, hair and outfits, unlike the photographic Wan result for this same prompt. A short blade exchange/parry is followed by separation and a long guard hold. Motion and footwork are limited, and lower legs/feet are cropped rather than full-body wide framing. Two swords remain identifiable; one free hand touches a blade during the exchange, so combat mechanics are approximate. No evident doubled facial features or major weapon duplication at sampled times.
Twenty sampled frames show two stable adult fighters performing a readable high parry and reset, with limited footwork and ambiguous blade contact. The requested anime rendering is absent: the result is photographic/live-action throughout. No doubled facial features seen in these samples.
Reviewed nine evenly spaced whole-clip frames. Two distinct adult anime swordsmen raise weapons, make short blade contacts/parries and step apart in a plain courtyard. Full bodies remain visible, with coherent cel-shaded styling, clothing and background. Motion is restrained and pose-held, with limited footwork and weak sense of impact; blade curvature/length and small hand alignment vary. Identity stays readable and no conspicuous doubled face is apparent. This delivers anime style, but the action is less vigorous than the stronger Wan sword samples.
Reviewed nine evenly spaced frames and full-resolution2s. Two distinct anime adult fighters in green and orange remain in full-body framing through lunges, raised guards and repeated sword crossings/parries. Coherent single faces and stable courtyard; anime rendering is maintained. Blade length/curvature changes and hand-to-hilt contact becomes ambiguous during a raised guard; foot contacts slide somewhat. No doubled face or catastrophic instability. First720p Sage2 hero passes workflow review; retained imperfections are model-quality findings.
Reviewed nine evenly spaced frames. Two anime fighters remain distinct in a full-body courtyard shot and perform a short sword exchange with lunges/parries before separating into a prolonged pause. A small hop and some sliding support weaken physical realism; blade shape and grip alignment vary. Clothing, main faces and architecture remain coherent at sampled times, with no apparent doubled foreground face. Anime style holds, but action is less sustained than seed24601.
Reviewed 20 temporal samples. Two distinct adult fighters remain in full-body view through a brief controlled sword exchange, parry and separation. Faces, clothing, two weapons and stone courtyard remain coherent at sampled times. Footwork is small and stiff, with a long static guard hold in the latter half, limiting motion score. Camera travels gently backward/sideways without an evident tile seam or doubled facial features. Better full-body framing than the matching Hunyuan anime motion-ladder duel.
Twenty sampled frames show stable realistic adult fighters and courtyard, a high parry, follow-through and reset. Sword geometry/contact is occasionally ambiguous and footwork remains stiff; no doubled facial features seen in the samples.
Realistic sword duel, HunyuanVideo 1.5 (full-quality hero), seed 24603: can it keep two fighters and their blades coherent through strikes and parries?
Reviewed nine evenly spaced whole-clip frames. Two realistic adult swordsmen raise and cross their weapons during a brief controlled exchange, shift position and settle into facing poses in a courtyard. Full bodies, identities, dark uniforms and setting stay coherent. Much of the latter clip is held stance, with limited dynamic footwork; blade/hand alignment and wrist geometry fluctuate during the raised parry. Separation is modest. No conspicuous duplicated main face at sample scale, but profiles and distance limit fine facial assessment.
Realistic sword duel, Wan 2.2 14B full (full-quality hero), seed 24601: can it keep two fighters and their blades coherent through strikes and parries?
Reviewed nine evenly spaced frames. Two distinct realistic adults exchange controlled sword strikes and raised parries, lunge, reset and separate in a plain stone courtyard. Both bodies and feet remain visible. Clothing and faces stay coherent at sampled times, with readable contact and more sustained action than the earlier paused anime seed. Some blade edges/lengths and hand-to-hilt alignment vary; stance transitions are slightly stiff. No visible doubled face, overlay text or watermark.
Realistic sword duel, Wan 2.2 14B full (full-quality hero), seed 24602: can it keep two fighters and their blades coherent through strikes and parries?
Reviewed nine evenly spaced frames. Two realistic adults in teal and black exchange sword gestures/parries and move around a plain courtyard, remaining in full-body view. Identity, clothing and stonework are coherent. Right fighter makes an unmotivated hopping/skipping motion, and the exchange gives way to approaching/resetting rather than consistently controlled combat. Blade length/alignment and foot support vary. The back-facing left fighter limits facial assessment; no apparent doubled main face in visible views.
V1-H_F1_real_wan22-t2v14-full-sage2-fp8_24602
Anime-style sword duel with speed lines (style test) 3
Anime-style sword duel with speed lines (style test), HunyuanVideo 1.5 (fast grid), seed 24601: does it nail the cel-shaded action look while two fighters clash?
Reviewed20 temporal samples plus full-resolution0.5s/2s/4s frames for tiled-decoder gate. No visible spatial tile boundaries or abrupt window break in these samples; horizontal bright edge streaks are consistent with prompted speed-line styling, not an evident tile grid. Coherent cel-shaded adult-intended fighters and courtyard; a brief high parry occurs around1-2s, then long planted poses, with weak footwork and limited exchange. Faces remain singular/coherent. Decoder512px overlap64 temporal64 overlap8 is disclosed;928.17s wall927.469exec38.815GiBpeak. This is not a controlled untiled-vs-tiled speed test.
Anime-style sword duel with speed lines (style test), Wan 2.2 14B Lightning (fast grid), seed 24601: does it nail the cel-shaded action look while two fighters clash?
Reviewed 20 temporal samples across five seconds. Two cel-shaded adult fighters perform a visible blade exchange and separate; courtyard and identities remain stable. Footwork is limited and sword-contact mechanics are simplified. No doubled facial features, added text, or watermark observed in samples. Mild sword/hand ambiguity; promising pipeline result, not a perfect combat simulation.
Reviewed20 temporal samples plus full-resolution1.75sframe. Two stable realistic adult fighters perform an overhead strike, crossed-blade parry, lower swords and reset; full-body courtyard framing and faces stay coherent. Footwork is stiff and much of the final portion is static. Full-resolution inspection shows one broad bright blade rather than a clear extra weapon, but blade-edge/guard detail is artificial/ambiguous across motion; fine contact physics remain uncertain. No doubled facial features seen. Retain this measured single-seed result.
20 sampled frames: two distinct adult fighters perform a readable overhead parry and return to guard. Stable courtyard and clothing; limited shuffle-like footwork and stiff body/contact mechanics. Blades/hands briefly ambiguous at crossing. No doubled face or visible watermark/text at contact scale. Photorealistic style retained.
Reviewed nine evenly spaced whole-clip frames. Two realistic adult swordsmen make a brief raised strike/parry, lower their blades and separate in a naturally lit courtyard. Full bodies and the broad sequence remain readable, with coherent identities, shirts and architecture. Footwork is limited and much of the later clip is a held stance. Blade alignment/length, grip and wrist geometry vary around the contact. No conspicuous duplicated main face at review scale; the result is less dynamic than the stronger Wan sword samples.
Reviewed nine whole-clip frames and60samples at12fps. Two realistic adults perform visible sword contacts/parries, lunges and an added side kick in a sunlit courtyard, then separate/reset. Full bodies remain in view; identities, costumes and courtyard geometry are coherent. The exchange is active and readable, but grip/blade alignment varies during fast movement and the low crouch-to-kick transition has uncertain support mechanics. Blades are briefly hard to distinguish against the background; no sustained third weapon or duplicated main face is established at review scale.
Reviewed nine evenly spaced whole-clip frames. Two realistic adult swordsmen exchange visible strikes/parries in a courtyard, then lower their weapons and separate. Full-body wide framing and readable stepping are maintained, with stable outfits, identities and architecture. Minor blade-length/alignment, grip and foot-support inconsistencies remain, but the requested action sequence is clear. Faces are relatively small and one is frequently in profile/back view. The wall plaque is too small to evaluate its lettering reliably; no overlay watermark or conspicuous doubled main face.
Anime hand-to-hand sparring: kick, block, throw, HunyuanVideo 1.5 (fast grid), seed 24601: can it handle body contact and a throw without merging limbs?
Reviewed 20 whole-clip samples, an additional 10fps contact sheet over seconds3-5, and full-resolution4.75s. Proper anime style and two distinct faces/outfits persist. Both fighters kick and make blocking/contact gestures; the ending shows a leg catch followed by the black-clad fighter dropping into a crouch, not a clearly readable controlled throw. Lower legs and landing area are cropped. Contact mechanics and support are approximate, with held poses. A pale human-like patch emerges in the background panel late in the clip, reducing temporal consistency; no doubled foreground facial features.
Anime hand-to-hand sparring: kick, block, throw, Wan 2.2 14B Lightning (fast grid), seed 24601: can it handle body contact and a throw without merging limbs?
Twenty sampled frames show coherent alternating front kicks and guards with stable adult identities, but no controlled throw. Requested anime rendering is absent: photographic martial-arts training throughout. Hands and kick contact remain approximate at contact-sheet resolution.
Anime hand-to-hand sparring: kick, block, throw, HunyuanVideo 1.5 (full-quality hero), seed 24603: can it handle body contact and a throw without merging limbs?
Reviewed nine whole-clip samples and60samples at12fps. Two distinct anime martial artists perform kicks, guards and punches on a mat; the final action is close grappling rather than a completed controlled throw. Anime styling is retained, unlike both Wan F2anime seeds. Feet/lower legs are cropped for much of the clip, failing the full-body-wide requirement and limiting support assessment. Held poses, sliding turns and unclear hand/limb contacts in the final clinch reduce motion realism. Clothing, identities and background are otherwise stable, with no conspicuous doubled main face at sample scale.
Anime hand-to-hand sparring: kick, block, throw, Wan 2.2 14B full (full-quality hero), seed 24601: can it handle body contact and a throw without merging limbs?
Reviewed nine whole-clip frames plus60samples at12fps over the first5s. Two distinct adult martial artists perform kicks, blocks/leg catches and resets in a training gym, staying in full-body view. No controlled throw occurs in the dense sequence. The output is photographic/live-action rather than the requested anime, a clear style failure. Identities and bodies stay coherent; extended kick holds, minor sliding/hopping and hand/foot contact ambiguities remain. This valid render is retained without reroll.
Anime hand-to-hand sparring: kick, block, throw, Wan 2.2 14B full (full-quality hero), seed 24602: can it handle body contact and a throw without merging limbs?
Reviewed nine whole-clip samples,60samples at12fps and full-resolution first/4.3s frames. Photographic sparring again misses requested anime style. Two fighters kick and block; the red-gloved fighter falls near the other feet, but a controlled throw mechanism is not clearly readable. Crucially, the red-gloved fighter changes from a long ponytail and clean face to short hair and a beard after the fall, a major identity-continuity failure. Contact/occluded-limb geometry is ambiguous. Gym wall lettering is malformed. Retained quality failure without reroll.
Realistic hand-to-hand sparring: kick, block, throw, HunyuanVideo 1.5 (fast grid), seed 24601: can it handle body contact and a throw without merging limbs?
Reviewed 20 temporal samples plus full-resolution2.5s and3s frames. Two distinct adult men retain coherent faces and clothing through repeated high kicks, blocking/guard gestures and a brief clinch, then reset. No controlled throw is visible. Much of the clip crops lower legs and feet, limiting assessment of support and missing full-body wide framing. Kick/contact mechanics remain approximate. Red/yellow lettering-like clothing graphics are malformed or unreadable, especially the back patch; no identifiable overlay watermark. Background and identities are otherwise stable at sampled times.
Realistic hand-to-hand sparring: kick, block, throw, Wan 2.2 14B Lightning (fast grid), seed 24601: can it handle body contact and a throw without merging limbs?
Twenty sampled frames show a readable front kick toward pads, return to guard, and small stance shifts, with stable realistic adult identities. The requested controlled throw never appears; this reads as pad practice rather than the full kick-block-throw sequence. Fine foot/contact physics remain approximate.
Realistic hand-to-hand sparring: kick, block, throw, HunyuanVideo 1.5 (full-quality hero), seed 24603: can it handle body contact and a throw without merging limbs?
Reviewed nine whole-clip samples and60samples at12fps. Two realistic adult martial artists kick, guard/block and exchange hand contacts on a training mat. The ending resembles a leg catch/clinch; no completed controlled throw occurs. Lower legs/feet are cropped in several views, so sustained full-body framing is missed. Identities, dark outfits and background remain coherent. Extended held kicks, sliding resets, unclear wrist contacts and the final intertwined limbs weaken physical realism. No conspicuous doubled main face at review scale; contact anatomy remains partly ambiguous.
Realistic hand-to-hand sparring: kick, block, throw, Wan 2.2 14B full (full-quality hero), seed 24601: can it handle body contact and a throw without merging limbs?
Reviewed nine whole-clip samples and60samples at12fps. Two realistic adult martial artists kick, block and reset on a mat, staying in full-body view with stable clothing and identities. No controlled throw occurs. Several extended leg holds, hopping resets and ambiguous hand-to-foot contact reduce physical realism, but the exchange is readable and no conspicuous duplicate face or extra limbs are apparent at sample scale. Retained without reroll.
Realistic hand-to-hand sparring: kick, block, throw, Wan 2.2 14B full (full-quality hero), seed 24602: can it handle body contact and a throw without merging limbs?
Reviewed nine whole-clip samples and60samples at12fps. Two adult martial artists in white uniforms exchange high kicks, punches and blocks in a realistic training gym. Full bodies remain visible and the distinct identities, uniforms and setting are coherent. No controlled throw occurs. Some extended kick holds, hopping/sliding resets and uncertain limb contact remain; no conspicuous doubled main face at review scale. Background spectators are small and soft. Retained without reroll.
Reviewed 20 temporal samples. A single adult dancer performs coherent knee lifts, forward kicks, small hops and lateral steps with stable clothing and limbs. The framing repeatedly cuts off the head in the first part and feet later, missing the requested full-body wide view. Face is coherent where visible, but cannot be assessed during head crops. Studio background and neutral cinematic rendering remain stable; no evident doubled features or tile seams. Wan M2 is a stronger full-body source candidate despite this result having readable dance motion.
20 sampled frames: clear full-body hip-hop-like stepping, knee lifts and varied arm gestures on a spacious studio floor. More movement variety than the two-person social-dance cases; face, gray shirt and room stable. Glossy floor reflections visible but fine reflection/hand physics not fully resolved at contact scale. No conspicuous doubled features or watermark.
Reviewed nine evenly spaced whole-clip frames. A single adult dancer performs hopping steps, knee bends, hip shifts and side lunges on a spacious studio floor. Full-body wide framing is maintained, with stable black shirt, beige trousers, face and curtain/mirror background. Choreography reads as ordinary hip-hop stepping rather than the acrobatic interpretation of Wan seed24602. Some foot sliding/weight-transfer ambiguity and small-hand softness remain. No conspicuous doubled face or added limbs at review scale.
Reviewed nine evenly spaced whole-clip frames. A single adult dancer performs varied crossing steps, arm swings, bends and low transitions on a spacious studio floor, in full-body wide framing. Identity, white overshirt, dark trousers and studio geometry remain coherent. Movement reads as hip-hop footwork; minor foot sliding/support ambiguities and small soft hands remain. No conspicuous duplicate face or extra limb at this review scale. Native H264 output was recovered by faststart remux after redundant re-encoding stalled; inference was not repeated.
Reviewed nine whole-clip frames and60samples at12fps. The hip-hop-step prompt becomes sustained acrobatic breakdancing: one-hand support, leg sweeps, inversion and head/shoulder-supported rotations. Full body and spacious sunlit studio are retained. Clothing and background remain stable, but repeated inverted spins, hand/head support transitions and leg geometry are approximate rather than convincingly weight-bearing. Face is frequently hidden or tiny, limiting facial assessment. No conspicuous added limbs at sample scale. The first seed is the clearer example of ordinary stepping choreography.
Reviewed 20 temporal samples: two distinct adults perform small synchronized social-dance steps, weight shifts and arm gestures in a bright spacious studio. Full bodies remain readable, with stable faces, clothes and separation. Movement is modest but more clearly stepped than the Wan style-grid partner dance. A later gray-shirted figure at left is visible within the wall mirror, consistent with a reflection rather than an extra dancer. Fine reflection geometry and hand detail are not fully resolved by the contact sheet; no obvious doubled facial features or tile boundaries at sampled times.
20 sampled frames: two adults in a naturally lit studio retain face/outfit/full-body framing. Mostly mirrored extended-arm gestures with limited planted-foot weight shifts and small hops; synchronized but shallow social-dance choreography. Stable room/reflections and realistic lighting, no conspicuous corruption at contact scale.
Reviewed nine evenly spaced whole-clip frames. The scene contains three clearly separate adults rather than the requested two. A pair on the left makes small steps/turns while the man on the right mainly claps/gestures, reading like a loose dance lesson rather than synchronized social partner dancing. Full-body wide framing and natural-light photographic styling are retained. Individual identities, outfits and studio background remain coherent. Motion is modest with hand/contact ambiguity, but no conspicuous duplicated main face; the extra person is a distinct additional subject, not a reflection.
Reviewed nine evenly spaced whole-clip frames. Two realistic adult dancers remain fully visible in a spacious bright studio, performing broadly coordinated arm extensions, raised knees, side kicks and lunges. Their faces, clothing and the room remain recognizable. Choreography reads more as contemporary/rehearsal dance than a social partner dance, and the two poses are not consistently synchronized. Minor foot-support and hand-shape ambiguities remain. Background mirror/reflection figures are present, so they are not counted as additional foreground dancers; reflection accuracy is not established. No conspicuous doubled foreground face.
Reviewed nine evenly spaced whole-clip frames. Two adult partners perform coordinated side steps, hand-held approaches/separations and an underarm turn in a spacious naturally lit studio. Full bodies remain visible, with stable clothing, silhouettes, room geometry and lighting. This seed reads clearly as social partner dance, unlike the contemporary choreography of seed24601. Minor hand-contact/support ambiguities remain. Faces are small and backlit, so fine facial anatomy cannot be fully assessed; no conspicuous doubled foreground face is apparent.
Twenty temporal samples preserve full-body4:3 kitchen-set framing, stable separate adult faces and broad classic-sitcom staging. The prompt asks for two synchronized dancers, but there are three adults: the left man and center woman perform largely independent steps/gestures while the right man mostly stands with hands at his hips. Motion is coherent but choreography weak. Warm wood cabinets/period-ish staging support the style; analog bloom remains subtle. Extra third person is a prompt-count miss, not duplicated facial features. No obvious technical artifact seen in samples.
20 sampled frames: two stable adult performers in full-body4:3 kitchen staging, mostly planted feet with arm waving and body bobbing; limited synchronized social-dance footwork. Bright studio staging reads as sitcom, but modern-looking appliances and crisp HD texture weaken the1980s analog-bloom target. Faces/outfits/background stable; no conspicuous visual corruption at contact scale.
Reviewed 20 temporal samples plus enlarged top-edge crop at 2.5 s. Stable adult dancer with coherent shuffle, hops and knee lifts in a 4:3 studio frame. Image remains clean and digitally sharp, with little convincing VHS chroma bleed, tracking damage or interlaced texture. Small timestamp-like white glyphs along the upper edge are malformed and unreadable; no identifiable watermark. Face/body remain coherent at sampled times.
VHS home-video hip-hop dance (style test), Wan 2.2 14B Lightning (fast grid), seed 24601: does it fake the VHS look (tracking lines, bleed) while dancing?
20 sampled frames: coherent full-body shuffle/hip-hop-like steps and stable adult performer/studio.4:3 output correct, but image remains crisp modern HD with little tracking noise or chroma bleed and no readable timestamp; a few corner marks do not establish VHS look. Requested analog degradation largely absent.
Reviewed 20 whole-clip samples,36 samples at12fps over the first3seconds, and full-resolution3.5s handstand. The gymnast performs repeated backward aerial rotations and settles into a prolonged handstand, ending with legs opening rather than a clear standing landing. The requested back-handspring plus full-twist sequence is not established. Takeoff/rotation transitions are floaty and support mechanics approximate; hands/wrists merge or become difficult to distinguish in the handstand, and facial detail is often hidden or blurred. Gym apparatus/background detail changes during camera travel. Neutral cinematic gym styling remains recognizable. This is a retained quality failure, not a reroll candidate.
Gymnastics: back handspring and full twist, Wan 2.2 14B Lightning (fast grid), seed 24601: hard acrobatics: does the body hold together through the flip?
Twenty sampled frames show a coherent sideways cartwheel into a standing balance, rather than the requested backward handspring and full aerial twist. Body remains mostly coherent through inversion, with ambiguous hand support and floor reflections. Wide padded-floor setting and cinematic style preserved; no doubled facial features seen. Sampled review does not certify precise gymnastics mechanics.
Reviewed 20 temporal samples. Adult dancer performs standing kicks, turns and a tucked aerial flip, then resets to upright arm gestures and a crouch. Requested windmills, headspin and floor freeze are not established. Full body is mostly visible; face and clothing remain coherent, with approximate aerial physics and motion blur. Stable bright studio background; no evident doubled face or watermark.
Breakdance: windmill, headspin, freeze, Wan 2.2 14B Lightning (fast grid), seed 24601: hard acrobatics: can it spin a body upside down without breaking it?
Twenty sampled frames show substantial floor rotation, hand-supported inversion, and a head-supported final pose, closer to the requested breakdance sequence than canonical Kite S1_M6. Leg/body configurations and support transitions become strained; exact windmill/headspin mechanics are not certified from samples. Clothing rides up consistently, while audience details drift slightly. No clearly doubled facial features, but inverted face is poorly resolved.
Reviewed20 temporal samples plus full-resolution2.25sframe. Silver hair and navy jacket with orange trim broadly match canonical Kite, but model invents an angular kite-like handheld prop, performs standing kicks/airborne rotation, and ends holding it in a planted pose. No windmill or headspin; some early framing crops the head and support/flight transitions are strained. Brief motion trails and body/prop transitions reduce temporal clarity. Cel-shaded style retained; no clear doubled facial features seen. The prop may relate to the ambiguous name Kite, but causation is untested. This quality miss is retained without reroll.
20-frame temporal contact review: corrected silver-haired/navy-orange adult Kite identity retained. First half is mostly a planted pose and arm flourish, followed by cartwheel/handstand and one-hand freeze. No complete windmill or headspin demonstrated; background arena lettering is garbled. Mild hand-ground/contact ambiguity; fixed arena and cel style stable.
Anime breakdance by Kite (style test), Wan 2.2 14B Lightning (fast grid), seed 24601 (earlier attempt): can it keep an anime character intact through spins?
20-frame temporal contact review: corrected silver-haired/navy-orange adult Kite identity retained. First half is mostly a planted pose and arm flourish, followed by cartwheel/handstand and one-hand freeze. No complete windmill or headspin demonstrated; background arena lettering is garbled. Mild hand-ground/contact ambiguity; fixed arena and cel style stable.
Reviewed 20 temporal samples. Stable adult woman in a gray jacket, with coherent face, hair and city setting during a gentle tracking/orbiting camera move. Framing stays at head and shoulders rather than the requested full-body view; legs and foot contacts are never visible, so walking gait cannot be established. Motion is mostly subtle upper-body/head movement and camera travel. Neutral cinematic rendering is coherent; no obvious doubled facial features, severe flicker or tile boundary at sampled times.
20 sampled frames: stable fictional adult woman walks naturally toward camera on a coherent street. Early full-body footfalls readable; framing tightens and feet approach lower edge near end instead of sustained full-body tracking. Face/outfit stable, neutral realistic styling. No conspicuous visual corruption at contact scale.
Reviewed nine evenly spaced whole-clip frames. An adult woman walks away along a city sidewalk in a stable rear tracking view. Hair, tan jacket, dark trousers and street architecture remain coherent. Framing stays around head-to-hip/thigh throughout, cropping legs and feet and failing the explicit full-body requirement. Small arm/shoulder movements suggest walking, but gait, foot support and facial integrity cannot be assessed from this view. No conspicuous duplication in the visible upper body; background signage is too small to evaluate.
Reviewed nine evenly spaced whole-clip frames. A single main adult woman walks away down a city street in sustained full-body tracking/wide framing. Her silhouette, blonde hair, black jacket and pale skirt remain coherent, as do the paving, lamp posts and buildings. Gait and foot placement are plausible at the small subject scale, with slight sliding uncertainty. The main face is not visible, so facial integrity is unassessed rather than proven. Background people and storefront lettering are small; no conspicuous overlay watermark or main-body duplication is apparent.
Reviewed nine evenly spaced whole-clip frames. An adult woman in a brown jacket and dark skirt walks toward camera along a busy city street, carrying a shoulder bag and glancing aside. Main identity, clothes, bag and city geometry remain coherent, with plausible alternating steps and arm motion. Camera framing grows progressively tighter; feet are cut off near the end, so the full-body requirement is only partly met. Small/distant background pedestrians limit fine anatomy assessment. No conspicuous doubled main face; storefront text is too small to evaluate reliably.
35 mm cinema: Mara walks and turns (style test), HunyuanVideo 1.5 (fast grid), seed 24601: does it keep Mara's described face and outfit while she walks and turns?
Twenty temporal samples show a coherent rear-view walk, turn toward camera and stable face through a closing camera move. Olive-toned skin, short dark bob, prominent cheek mark, teal outerwear, gray trousers and brown boots broadly follow the description, but the field jacket becomes a tailored blazer and full-body tracking becomes a close-up for most of the latter clip. Eye/hair nuances and exact canonical identity are not guaranteed by T2V description alone. Natural sunlight and city remain coherent; no doubled facial features or clear tile boundary seen. Potential detail-upscale source because the face/hair are large and stable, subject to final grid selection.
35 mm cinema: Mara walks and turns (style test), Wan 2.2 14B Lightning (fast grid), seed 24601: does it keep Mara's described face and outfit while she walks and turns?
20 sampled frames: believable forward walk, stable adult face/teal jacket and sunlit street. Starts full-body but camera does not maintain full-body tracking as she approaches; no distinct turn from profile toward camera. Natural-light realistic style strong, readable early footfalls; tiny face/finger detail limited by contact scale. No obvious doubled features, watermark or conspicuous artifact.
35 mm cinema: Mara walks and turns (style test), HunyuanVideo 1.5 (full-quality hero), seed 24603: does it keep Mara's described face and outfit while she walks and turns?
Reviewed nine evenly spaced whole-clip frames. A woman with a short wavy dark bob, teal jacket, charcoal trousers and brown boots walks away along a sunlit street, then turns toward camera. Photographic face, clothing and environment remain coherent. The camera rapidly pushes from full-body tracking to chest/head close-up, failing sustained full-body framing and limiting later gait assessment. The specific cheek freckle is not established and auburn hair color is weak. Visible walking and turn are plausible; no conspicuous doubled face.
35 mm cinema: Mara walks and turns (style test), Wan 2.2 14B full (full-quality hero), seed 24601: does it keep Mara's described face and outfit while she walks and turns?
Reviewed nine evenly spaced whole-clip frames. A woman with short auburn bob, teal jacket, dark trousers and brown boots walks away along a sunlit street, then turns toward camera. Natural-light photographic styling and visible identity are coherent. The tracking camera progressively pushes from full-body to waist/chest/head framing, violating sustained full-body framing and limiting late gait assessment. Small freckle and eye-color details cannot be confirmed at this scale. Initial walking/support is plausible with minor foot-placement uncertainty; no apparent doubled face.
35 mm cinema: Mara walks and turns (style test), Wan 2.2 14B full (full-quality hero), seed 24602: does it keep Mara's described face and outfit while she walks and turns?
Reviewed nine evenly spaced whole-clip frames. Short auburn hair, teal belted jacket, dark trousers and brown boots match the broad description, with a stable photographic face and sunlit street. The woman is already facing camera and walks forward throughout, so no requested turn is established. Framing progressively crops feet and then lower legs, missing sustained full-body tracking. Walking and arm swing appear plausible where visible; background pedestrians are small and soft. Freckle/eye-color details are unconfirmed at this scale; no conspicuous doubled main face.
Twenty temporal samples show a coherent waist-up tracking walk with a stable adult face, period hair/clothing and vintage traffic. Near-monochrome cool-gray rendering is a stronger period/newsreel cue than Wan S3, but film grain and gate weave are weak and the requested full-body view is absent throughout, so leg/foot mechanics cannot be assessed. Foreground/background motion appears coherent in samples; no doubled facial features or obvious tile boundary. Clean classic-film look rather than a complete distressed-newsreel treatment.
20 sampled frames: coherent adult woman walking on a crowded street with period-looking clothing and cars. Full body visible initially; foreground passerby crosses naturally. Image is desaturated rather than strictly black-and-white (colored lights/buildings/skin remain), with little visible film grain or gate weave; looks like a clean modern period reconstruction rather than newsreel. Main face/body stable at contact scale.
V1_S3_M1_wan22_lightning_24601
1970s sci-fi TV bridge walk-and-turn (style test) 6
1970s sci-fi TV bridge walk-and-turn (style test), HunyuanVideo 1.5 (fast grid), seed 24601: does it get the saturated 70s TV look while she walks and turns?
Twenty temporal samples show a coherent adult officer in a white retro uniform, plastic-looking bridge console and saturated red/blue space backdrop in4:3. She mainly pivots away and back with a few small steps, rather than clearly walking across the bridge; latter half is a posed hold with hands clasped. Face/body remain coherent through camera movement. Broad retro-space-television styling succeeds, but the colorful nebula is digitally crisp rather than clearly painted, and soft-focus film texture is weak. No obvious doubled features or tile boundary seen.
1970s sci-fi TV bridge walk-and-turn (style test), Wan 2.2 14B Lightning (fast grid), seed 24601: does it get the saturated 70s TV look while she walks and turns?
20 sampled frames: adult officer walks across raised bridge area, turns toward camera and approaches. Stable ochre uniform/face; another crew member crosses behind her (occlusion, not duplicated facial features).4:3 retro staging and saturated colors read well, although space panorama is crisper/more digital than a painted backdrop. No conspicuous corruption at contact scale.
1970s sci-fi TV bridge walk-and-turn (style test), HunyuanVideo 1.5 (full-quality hero), seed 24603: does it get the saturated 70s TV look while she walks and turns?
Reviewed nine evenly spaced whole-clip frames. An animated officer in a blue uniform walks straight toward camera on a colorful bridge, with no clear stop-and-turn sequence. The correct4:3 frame progressively tightens from wide/full-body to a large face close-up. Cel-shaded animation, sharp outlines and vivid space imagery replace the requested soft-focus1970s film/television look. Face, uniform and bridge stay mostly coherent, but gait is simple and then hidden by cropping. No conspicuous doubled face; retained style/staging failure without reroll.
1970s sci-fi TV bridge walk-and-turn (style test), Wan 2.2 14B full (full-quality hero), seed 24601: does it get the saturated 70s TV look while she walks and turns?
Reviewed nine evenly spaced whole-clip frames. A coherent adult female officer in a teal uniform moves across/around a bridge console with a stylized star-and-planet backdrop. She looks aside and leans toward the foreground; a deliberate walk-stop-turn-to-camera sequence is weak. The 4:3 frame is correct, but the subject becomes medium/close and the console obscures her lower body. Clean modern lighting, muted uniform and crisp detail only partly evoke the requested soft-focus1970s saturated-primary television look. Face and outfit remain stable, with some hand uncertainty near the console; no conspicuous doubled face.
1970s sci-fi TV bridge walk-and-turn (style test), Wan 2.2 14B full (full-quality hero), seed 24602: does it get the saturated 70s TV look while she walks and turns?
Reviewed nine evenly spaced whole-clip frames. An adult female officer walks laterally across a starship bridge, stops, turns toward camera and resumes a small forward step. Correct4:3 wide framing retains the full body much better than seed24601. Character silhouette, uniform and bridge geometry remain stable. The scene is very dim, with muted olive clothing and modern-looking console lighting rather than saturated-primary soft-focus1970s television. Darkness limits fine face/hand assessment; gait is plausible at sample scale and no conspicuous body duplication is apparent.
Image to video and character consistency V2V3V4 55 videos
What we're testing:V2 animates 8 style keyframe stills (made in the K suite) with Wan 2.2 I2V 14B vs HunyuanVideo 1.5 I2V: does the clip keep the still's look, face and outfit once it moves? V3V4 gives the model fixed images to hit. First + last frame: it must fill in the motion between a given start and end image (pose change, day to night, match cut, seamless loop) on Wan 2.2 first-last-frame. Reference: keep two original characters, Mara (realistic) and Kite (anime), recognizable across walk, talk and sword-duel shots on Wan VACE 14B, HuMo 1.7 and SkyReels V3. · V2 notesV3V4 notes
Keyframe S1: Anime swordsman swings and returns to guard 3
Anime swordsman swings and returns to guard, HunyuanVideo 1.5 I2V, seed 24601: can it animate the still and keep the cel-shaded style, face and outfit?
Reviewed 20-frame temporal contact, reference/start/middle/end panel and full-resolution3.5s frame. Face, silver hair and navy/orange outfit remain recognizable, with cel-shaded styling. The single sword/scabbard configuration morphs into two unsheathed swords; extra curved blade shapes appear during the draw/swing. Feet remain broadly planted while camera pushes in and crops the legs, so forward step and steady full-body framing are weak. Weapon continuity is a clear failure despite stable facial anatomy.
Reviewed20 temporal samples, reference/start/middle/end panel and full-resolution2.75s/3.25s frames. Kite retains silver hair, coherent face and navy/orange costume in anime style. Mostly planted stance with arm/sword raising and return to front guard; little clear forward stepping or decisive swing. Black sheath-like material remains attached around the outer blade and changes shape during the motion, producing an implausible partially sheathed weapon. Hands and sword grip are approximate; no doubled facial features. Background remains stable; feet partly crop as framing tightens.
V2_S1_wan22_i2v_fp8_24601
Keyframe S2: Woman walks a city street and turns 3
Reviewed 20-frame contact and reference/start/middle/end panel. Natural forward walk with small arm swing; face, short dark hair, teal jacket and olive top remain consistent with the reference. Realistic city lighting and background layout remain coherent. She faces the camera throughout with little distinct requested turn; advancing framing crops the feet and eventually lower body. No conspicuous doubled facial features or extra limbs in sampled frames.
Reviewed20 temporal samples and reference/start/middle/end panel. Mara walks forward, makes a small hand gesture and turns her head/body while retaining the short auburn bob, teal jacket, dark trousers and recognizable face. Realistic city, natural light and reference composition are maintained before the camera closes in; later foot contact cannot be assessed because legs crop. Background cars move consistently at sampled times. No evident doubled face or extra limbs; fine hand/traffic detail is limited by sampled review.
Reviewed 20-frame contact and reference/start/middle/end panel. Coherent forward sidewalk walk, stable face, short hair and jacket, with period cars and monochrome appearance preserved. The image remains relatively clean and stable rather than visibly grainy with newsreel gate weave; those requested style elements are weak. Framing gradually crops the legs as she approaches. No conspicuous doubled features or extra limbs in sampled frames.
Reviewed20 temporal samples and reference/start/middle/end panel. Initial held stance transitions into a coherent forward walk with small head turns. Reference face, short hair, jacket/trousers and period street/cars remain recognizable. Monochrome is preserved, but the image stays relatively clean and stable with little evident film grain or gate weave. Boots/feet crop later as the camera closes. No conspicuous doubled features or new large scene artifacts at sampled times.
Reviewed 20-frame contact and reference/start/middle/end panel. Full-body playful side-steps and raised-arm dance preserve face, teal jacket, boots and kitchen composition. Warm lighting remains more cinematic than flat multicamera sitcom lighting. Fine horizontal scanline/banding texture becomes more apparent over the later frames, including the cabinets and figure; not a clear clean slight-bloom treatment. No conspicuous doubled face or extra limbs in the sampled dance.
Reviewed20 temporal samples and reference/start/middle/end panel. Single recognizable Mara performs playful small dance steps, knee bends, raised-arm gestures and a turn, mostly full-body within the4:3 kitchen composition. Face, teal jacket and brown boots remain coherent. Cabinets, appliances and floor hold stable; no extra person or doubled face. Warm clean cinematic lighting remains stronger than the requested flat studio/video-bloom aesthetic, although period domestic styling is recognizable.
V2_S4_wan22_i2v_fp8_24601
Keyframe S5: 1970s sci-fi TV officer walks and turns 3
Reviewed temporal contact and reference/start/middle/end panel. Canonical K reference already depicts anime Kite on a bridge; this is not a newly introduced conversion from live action. The figure advances toward the camera with limited early leg motion, then grows into a poorly framed close-up that crops the top of his head and eyes. No clear two-step stop-and-turn sequence. Silver hair and navy/orange outfit remain recognizable, but jacket/collar proportions and folds shift. Saturated space background persists; requested1970s soft-focus television look is weak.
Reviewed20 temporal samples and reference/start/middle/end panel. Canonical S5 reference is an anime Kite on a saturated space bridge; this anime appearance is inherited from the supplied keyframe, not newly introduced by the video model. The clip gives a small orientation change and apparent forward drift with largely planted legs, not two clearly readable steps followed by a turn. Face and silver hair remain coherent, but orange collar/waist trim and jacket opening simplify or shift. Starfield/planet composition remains stable. Clean anime graphics lack the requested1970s television softness; distinguish reference styling limitation from motion failure.
Reviewed temporal contact and reference/start/middle/end panel. Varied hip-hop-like steps, knee lifts and upper-body gestures keep the broad face/hair and teal-jacket identity. Strong colored tracking lines in the canonical reference and first frame largely disappear almost immediately, leaving a clean studio image instead of the requested persistent VHS/chroma-bleed/480i appearance. Some transient face/hand smearing during faster movement and slight garment-proportion drift; no clear extra limb in sampled frames.
Reviewed20 temporal samples and reference/start/middle/end panel. Mara progresses from a short held stance into hops, side steps, knee bends and broad arm gestures with full body largely visible. Teal jacket, boots, short hair and broad facial identity persist. The shirt hem rises/shortens to expose the midriff as motion develops, an outfit-consistency miss. Inherited horizontal colored tracking lines vary in strength and motion blur grows; these are partly requested VHS effects, not all new defects. No clear duplicated face or extra limb at sampled times; hand/facial fine detail is blurred during faster movement.
Reviewed temporal contact and reference/start/middle/end panel. Upright kicks, turns and arm gestures end in a wide standing pose. No requested windmill, headspin or floor freeze appears in sampled sequence. Several fast turns/kicks show conspicuous translucent duplicate limb silhouettes and smeared body edges. Face, silver hair, navy/orange outfit and cel-shaded look remain broadly recognizable, but body scale/framing shifts during motion.
Reviewed20 temporal samples and reference/start/middle/end panel. Kite retains silver hair, navy/orange outfit and cel-shaded face. Standing arm poses and a turn lead into a one-hand cartwheel/brief inverted split, then a crouch and upright reset. Requested windmills, headspin and final freeze are not established. Full body and clean background remain visible; transitions and support mechanics are approximate, but no conspicuous added limbs or duplicate face are seen at sample scale. Face is hidden during the inverted/back-facing action.
V2_S7_wan22_i2v_fp8_24601
Keyframe S8: Realistic woman raises a sword and parries 3
Reviewed temporal contact and reference/start/middle/end panel. Mara raises/swings the sword, turns and finishes in a wide guard, with broad face/outfit identity and photographic studio appearance preserved. A second blonde opponent enters prominently despite the off-screen-opponent request. Fast turns and the encounter show severe transparent body/arm trails; blade length/curvature and contact geometry are unstable. Framing crops the head during the middle action. This is not a clean controlled parry sequence.
Reviewed20 temporal samples and reference/start/middle/end panel. Mara raises the sword, advances toward camera, makes an overhead strike/parry-like motion and returns to guard. Face, hair and clothing remain recognizable where visible. An opponent enters the foreground despite the off-screen instruction, obscuring action; the camera also crops Mara head/legs during the exchange. Blade shape changes from straight to visibly curved and contact mechanics are unclear. The second person is a newly visible opponent rather than a duplicated Mara face. Studio backdrop and photographic look are retained; no clear overlay text/watermark.
Reviewed nine evenly spaced frames including both endpoints: Mara raises her right arm naturally and holds the greeting pose. Identity, outfit, studio and framing stay stable; endpoint guidance is strong. Slight finger softness at intermediate poses. Full-step execution 825.1 s; sampling review rather than frame-by-frame hand validation.
Nine samples including endpoints show a clear daylight-to-night transition with practical lights coming on. Mara stays stable and nearly still, as expected for the timelapse framing; background traffic changes. No doubled features or obvious text/watermark. Strong endpoint control.
Nine samples including endpoints: ceramic cup becomes a glowing orb, then diner dissolves into spaceship bridge. Endpoint identities and composition mostly retained. This is a morph/cross-dissolve, not a crisp editorial match cut; overlapping sets ghost during the transition. Little body movement. Garbled diner sign is inherited from the start keyframe.
Nine samples show stable Mara/garden with very subtle movement. Identical supplied endpoints did not produce a pixel-seamless loop: decoded first/last RGB MAE5.50 versus average neighboring-frame MAE0.358 at320x180. Visible pose is close, but residual background/face drift makes the seam stronger than ordinary frame changes. No postprocessing used to conceal the result.
Reviewed nine evenly spaced frames. Clear forward walk along an outdoor brick-walled street, with alternating arm/leg positions and a coherent single face. Teal jacket, charcoal trousers, brown boots and auburn bob remain. Slight facial/hair drift; feet move outside the frame as she approaches, limiting late foot-contact assessment. Corrected zero-audio conditioning works; silent output.
Reviewed nine evenly spaced frames against matching SDPA baseline. Coherent street walk, single face, bob and outfit identity are comparable; similarly harsh contrast and blue clothing drift remain. Scene geometry differs and a new background sign has garbled letters. This is not bit-identical or artifact-free. Total1645.321s versus2525.147s SDPA, approximately70.8 versus114.6s per sampling step. Adopted for the six primary SkyReels cases with the text artifact explicitly retained.
Reviewed nine evenly spaced frames. Coherent single Mara walks toward camera along a real-looking city sidewalk with alternating steps and arm swing. Face, auburn bob and broad outfit identity remain, but jacket/trousers shift toward blue and the image has strongly crushed shadows and clipped highlights. Foot contacts are readable in most samples, with some crop near the end. SDPA baseline;2525.147s including cold loading, roughly114.6s per sampling step.
Reviewed nine evenly spaced frames. Coherent forward walk with natural arm swing and a single face; teal jacket, auburn bob and boots remain. The requested city street is replaced by the reference studio. Face changes mildly near the end; tighter crop follows the approach.
Nine samples: Mara identity/outfit strong, but only a few small steps and studio backdrop persists instead of city street. Studio geometry shifts behind her. Reference conditioning overpowers scene/action instructions. Cold load dominates586.54s execution; retain this low-adherence result.
Nine samples: coherent single-person talking close-up, natural mouth changes, stable framing and plain studio background. The duplicated full-body overlays from the original four-step configuration disappear. Hair changes to a shorter bob with fringe; facial details/identity less exact than the reference. Same prompt/reference/seed;20steps,cfg3.5,noLoRA versus4steps,cfg1,LoRA. Execution975.28s. Silent reference test, so no phoneme lip-sync score.
Reviewed nine evenly spaced frames. The large face has coherent speech-like mouth changes and recognizable Mara features, but an unwanted second full-body copy of the reference person remains at the right throughout. This is a major duplicated-character/composite failure, not an acceptable single close-up. Individual faces are coherent; the composition is wrong. No audio or phoneme lip-sync claim. Retained as a negative benchmark result under the same configuration as the successful walk.
Mara talk close-up from a reference image, Wan VACE 14B Lightning (official) [official fast control], seed 44101: is Mara still recognizable, same face and outfit?
Nine samples contain colored noise with no usable subject. The published VideoX-Fun sampler settings did not transfer successfully to this native-node split-sampler adaptation. This is a local adaptation failure, not evidence that the original custom workflow fails. No further fast-reference fixes after the two controls; adopt the validated20-step baseline.
Reviewed nine evenly spaced frames. A single close-up has clear silent speaking mouth shapes, blinks and head tilts; no inset or duplicate body. The realistic reference becomes strongly stylized/plastic-looking, with heavy red lipstick, saturated smooth skin and a straighter darker bob. Teal jacket and broad facial traits remain, but realism and identity fidelity weaken. Gaze often falls slightly off camera. No audio; phoneme lip-sync not assessed.
Nine samples reveal a severe composition failure: a talking close-up is overlaid with multiple full-body copies of the reference. Duplicated subjects/face composition is a fail despite recognizable Mara traits. This four-step T2V-LoRA/VACE configuration needs a full-step control on the same prompt before attributing the defect to base VACE.
Reviewed nine evenly spaced frames. Coherent single Mara in a wide sword-ready stance, with clear arm swings and torso turns; teal jacket, trousers and boots remain. Sword stays recognizable across the sequence, with blade blur and occasional uncertain grip during swings. Footwork is limited and the lower feet are partly cropped. Background becomes an indoor practice space. No duplicated person.
Walks and turns away across a studio with sword lowered, missing the requested ready stance; feet intermittently cropped. Coherent body and face with mild identity drift, harsh contrast and a changing/darkening sword blade. Studio equipment remains visible.
Reviewed nine evenly spaced frames. Unlike the nearly static fast baseline, the full-step output raises a sword and steps into a low ready stance. Body, outfit and single face remain coherent. The long blade visibly shortens to a dagger around the middle and remains short; weapon continuity is poor. Reference studio persists.
Nine samples: reference identity and clothes stay stable, but requested sword-ready stance, weapon and controlled footwork are absent. Almost static studio pose. Poor action adherence under the four-step configuration; retain rather than counting pipeline execution as creative success.
Reviewed nine evenly spaced frames. A recognizable single Kite with silver hair and navy/orange jacket advances along a textured city road under a high tracking camera. Arm/body changes and moving ground imply walking, but the waist-up crop prevents assessment of leg movement or foot contacts. Face and outfit remain coherent; background road textures are unusually rough.
Recognizable silver hair, amber eyes and navy/orange clothing on a city street. Upper-body bobbing and camera movement suggest walking, but legs remain outside frame so gait/foot contact cannot be assessed. Slight face and collar drift, strongly contrasted anime shading, small garbled shop signs, unprompted mouth opening late. Single coherent character without duplicate face.
Reviewed nine evenly spaced frames. Silver hair, navy/orange jacket and one coherent anime face persist; a city-street setting appears. Motion is restrained, with head turns and slight body bobbing while arms stay near the jacket; the thigh-up crop hides foot contacts, so natural walking physics cannot be confirmed. Detailed near-photographic background mixes with the anime figure.
Reviewed nine evenly spaced frames. A single chest-up anime Kite speaks silently with clear mouth changes, a blink and subtle head/shoulder motion. Silver hair, amber eyes and navy/orange jacket remain consistent; no extra person or inset. Clean facial structure and stable linework. Background changes to a warm paneled interior. No audio, so phoneme lip-sync is not assessed.
Reviewed nine evenly spaced frames. Coherent single anime adult with silver hair, amber eyes and navy/orange jacket. Blinks, mouth movements and a gradual camera push-in support the silent talk prompt. Framing starts medium rather than close-up; background has angular studio-like panels. Minor face/hair contour drift; no duplicate face. Lip synchronization cannot be evaluated without audio.
Reviewed nine evenly spaced frames. One stable anime character retains the reference outfit, hair, body proportions and studio. Small mouth/head changes suggest silent speaking, but the requested close-up is ignored: the entire body remains visible throughout. Face detail is too small to judge fine mouth anatomy; no audio or lip-sync claim.
Reviewed nine evenly spaced frames. Single recognizable Kite takes a wide two-handed sword stance, lowers the weapon and holds it; full body and both feet remain visible. Silver hair and navy/orange outfit are stable. Blade curvature/length varies mildly during the lowering motion, and hand-grip detail is limited at this scale. A high camera angle and busy gravel texture dominate the setting.
Reviewed nine evenly spaced frames. Recognizable silver-haired anime Kite in navy/orange clothing holds a sword lowered beside his leg, with small head turns, blinking and slight upper-body shifts. No clear transition into a guarded duel stance or controlled footwork; lower legs and feet are cropped. Strong radial speed lines dominate an angular corridor backdrop despite restrained subject motion. Face and body remain coherent, with limited hand/blade detail and mild clothing drift.
Reviewed nine evenly spaced frames. Kite turns away and swings a sword into a wide ready stance, with clear foot movement. Outfit and silver hair persist. The blade curvature/length changes, and legs become visibly soft/blurred during movement; grip details are ambiguous. The back-facing final pose limits face assessment.
What we're testing:V5 starts from an existing clip (walk, dance, two-person talk, breakdance) and swaps the person for Mara or Kite, or transfers the source dance onto them, with Wan-Animate 2, while keeping the motion and camera. The mask clips are helper passes (SAM2 / BiRefNet) that cut out the person, background or a detail (cup, jacket, sunglasses) for swaps. V6 restyles a whole clip (make it rain, day to night, live action to anime) with VACE 14B, next to a plain ffmpeg filter as the control. · V5 notesV6 notes
Reviewed nine output frames against the same walking source and Kite reference. Silver hair, amber-looking eyes, navy/orange jacket, black shirt, belt and boots are strongly retained in coherent anime styling. Forward walking/approach broadly follows the source. The source street is entirely replaced by the gray reference backdrop, and the framing progressively crops feet and the top of the head. Small hand/foot-support ambiguities remain, but identity is stable where visible. Valid native motion-regeneration output; fails localized source-surroundings preservation.
Clear Kite silver hair, navy/orange jacket and anime face composited into preserved photographic street. Forward walk and camera approach follow source. Late leg crop inherited from source framing; slight stylized subject/environment lighting mismatch but no obvious doubled features or strong halo in nine samples.
Reviewed nine output frames alongside the source strip and full-resolution Mara reference. Reference face, auburn bob, teal bomber, olive top, charcoal trousers and brown boots are strongly retained. The output transfers a forward walk/approach and is temporally coherent, with small hand/support ambiguities. It entirely replaces the source street with the reference studio backdrop, so the preserve-surroundings instruction fails. Framing also becomes tighter than the source and crops the top of the head late. This native Animate2 workflow supplies motion conditioning but no spatial edit mask; it is a functioning motion-regeneration baseline, not a successful localized swap. Source motion timing was assessed broadly, not by joint tracking.
Corrected masked-source blanking adopts reference auburn bob, teal zip-style jacket, olive top, charcoal trousers and brown boots while preserving street and forward walk. Fine face likeness not exact; source framing crops legs late. No obvious doubled face or strong edit halo in nine temporal samples.
Diagnostic: supplied editable source pixels without official inpainting blanking. Street/camera/walk preserved, but source buttoned field jacket persists instead of Mara bomber and olive top. Retained as preprocessing failure; not canonical matrix result.
Reviewed nine output frames against the dance-source strip and Kite reference. Silver hair, face, navy/orange jacket, belt and boots remain strongly recognizable in anime styling. Side steps, knee lifts, raised-hand gestures and crossed-leg movement broadly track the source choreography. The source brick/window studio is replaced by the gray reference backdrop, and several poses crop the top of the head or feet despite the source staying wide. Small hand and support ambiguities remain; identity is coherent in visible views. Native motion regeneration succeeds more clearly than selective scene preservation.
Silver hair and navy/orange outfit adopted; source studio, reflections, stepping and arm gestures preserved. Subject is back-facing and source-like in build; facial identity unassessable. Body rendering more photographic than cel-anime reference, so partial style/identity transfer. No clear limb duplication. QA correction after direct source comparison: source dancer faces camera, but replacement faces away throughout. This is an orientation/pose-preservation failure, not a limitation inherited from the source.
Reviewed nine output frames against the dance-source strip and Mara reference. Mara retains the reference face, auburn bob, bomber jacket, olive top, trousers and boots while reproducing broadly corresponding side steps, knee lifts and raised-hand gestures. The source brick/window studio is replaced by the reference backdrop and lighting, failing source-surroundings preservation. The character grows/tightens in frame, with head and feet cropped during several dance poses despite the source staying wide. Motion remains readable with small hand and boot/support ambiguities. Reference identity is strong where the face is visible.
Preserves brick-window studio, floor reflections, steps and arm gestures. Teal jacket, charcoal trousers and tall brown boots adopted, but broad source silhouette persists and hair short/indistinct. Face remains away so exact facial identity is unassessable; identity score reflects partial outfit/hair transfer only. No clear limb duplication. QA correction after direct source comparison: source dancer faces camera, but replacement faces away throughout. This is an orientation/pose-preservation failure, not a limitation inherited from the source.
Reviewed nine output frames against the dance-source strip and Kite reference. Knee lifts, side steps, hand raises and crossed-leg movement broadly reproduce the driving choreography, with strong silver-haired navy/orange reference identity and consistent anime styling. Changing to a motion-only prompt does not solve repeated head/foot cropping, so the full-body-visible instruction remains unmet. Hand placements differ somewhat from the source and support/contact details are approximate. Reference backdrop is acceptable for this motion-transfer task; no source-background preservation claim is made.
Reviewed nine output frames against the dance-source strip and Mara reference. Side steps, knee lifts, raised-hand gestures and crossed-leg movement broadly follow the driving choreography while the reference appearance remains strong. Repeated head/foot cropping fails the explicit full-body-visible requirement and limits anatomy assessment during raised poses. Small hand and boot-support ambiguities remain. The reference studio background is acceptable for this motion-transfer prompt, which does not require preserving source surroundings; it is not penalized here as a localized-edit failure. Exact timing was judged visually rather than by joint tracking.
Reviewed nine evenly spaced frames and full-resolution first frame. Two distinct fictional adults remain on separate sides of a coherent kitchen, with small mouth/head movements and gestures. Woman on left wears blue jacket and holds a clearly visible ceramic cup; cup is white rather than requested blue. Stable framing and identities, minor hand detail ambiguity around cup. Suitable auxiliary source for masked person/cup edits; silent, no phoneme lip-sync claim.
Reviewed nine output frames against the two-person kitchen source and Kite reference. A recognizable anime Kite stands almost still with small head/body changes against the gray reference backdrop. The man on the right and the entire kitchen disappear, and the source woman cup-holding pose/cup are not reproduced. This fails both selective two-person replacement and much of the source motion/interaction. Reference identity and visible anatomy remain coherent; no conspicuous doubled face. No lip-sync claim is made for the silent source.
Failed Kite replacement: long brown-haired photographic woman remains, with navy clothing but no silver hair, anime face or recognizable reference identity. Man, kitchen, cup and subtle conversation movement preserved. Retained fixed-seed quality failure; no phoneme lip-sync claim.
Reviewed nine output frames against the two-person kitchen source and Mara reference. The source woman cup-holding pose and small conversational head/hand motions transfer to a recognizable Mara in the reference outfit. The man on the right disappears entirely, and the kitchen is replaced by the studio reference backdrop. Framing changes from a two-person medium shot to a single full-body figure, with inferred lower-body stance. Thus the instruction to replace only the woman while preserving the other person/surroundings fails. Visible anatomy and identity are coherent, though the face is small/in profile and cup/hand detail is limited. No lip-sync claim is made for this silent source.
Man, kitchen, cup-holding interaction and subtle mouth/head motion preserved. Woman has teal clothing but retains long brown hair/source-like profile instead of reference short auburn bob and clear Mara identity. Partial clothing recolor rather than successful character replacement; no phoneme lip-sync claim from silent output.
Reviewed nine output frames against the breakdance-source strip and Kite reference. The arms-out held stance, bend, inversion and leg sweep broadly follow the source, with strong silver-haired navy/orange anime identity. Hand/limb outlines are cleaner than the realistic Mara transfer at sample scale, but inverted weight support and transition mechanics remain approximate. A raised leg/boot leaves the top edge, failing full-body visibility. Source long holds are retained; the driving clip itself does not establish a complete windmill/headspin sequence. The plain reference background is acceptable for this motion-only task.
Reviewed nine output frames, the breakdance-source strip and a full-resolution inverted frame at3.8s. Mara copies the source broad arms-out held stance followed by a two-hand/one-hand-supported inversion and leg sweep. Outfit and general reference identity are retained, with face assessment limited during inversion. Raised boots/legs leave the top of frame, failing full-body visibility. Supporting fingers/wrists are elongated and awkward in the full-resolution frame; inversion transitions and weight support are approximate. The source itself has long held poses and does not establish a complete windmill/headspin routine, so those are not attributed solely to this transfer output. Studio reference background is acceptable for motion-only transfer.
Auxiliary mask review: source/mask/red overlays at frames0,27,54,80; all81frame area diagnostics have no blanks. Dancer excluded while room and floor reflection are selected. Consistent silhouette with soft edge treatment.
Auxiliary mask review: source/mask/red overlays at frames0,27,54,80; all81frame area diagnostics have no blanks. Both adults and held cup excluded; kitchen selected, with soft boundaries around people. Stable throughout reviewed samples.
Auxiliary mask review: source/mask/red overlays at frames0,27,54,80; all81frame area diagnostics have no blanks. Background selected while walking subject is excluded; soft/dilated boundary may edit fringe pixels. Distant pedestrians are treated as background.
Auxiliary mask correction1 reviewed as source/red overlays at frames0,27,54,80. Revised positive/negative points and reduced dilation still select the hands while excluding the cup. Retained quality failure, not approved; second correction uses constrained CPU color tracking.
Auxiliary correction2 reviewed at frames0,27,54,80 plus enlarged full-resolution first-frame overlay. Mask selects exposed white ceramic body and rim, leaving surrounding fingers mostly untouched; tiny shaded edge portions are excluded and two-pixel dilation reaches boundary pixels. All81frames nonblank,maskarea0.247-0.303percent atreviewresolution. Approved only for this nearly stationary white-cup source; constrained color-component method is not a general learned segmentation result.
Auxiliary mask review: source/mask/red overlays at frames0,27,54,80; all81frame area diagnostics have no blanks. QUALITY FAILURE: stable mask selects hands and fingers around the cup while excluding the ceramic cup. Not approved for editing; corrective segmentation required.
Auxiliary mask review: source/mask/red overlays at frames0,27,54,80; all81frame area diagnostics have no blanks. Jacket torso and sleeves follow the approach; face, hair and trousers remain excluded. Small dilation reaches garment boundary.
Auxiliary mask review: source/mask/red overlays at frames0,27,54,80; all81frame area diagnostics have no blanks. Face/head region tracks the dancer without selecting the body. It is broader than an eye-only mask, so face preservation must be judged in the edit result; minor isolated specks.
Auxiliary mask review: source/mask/red overlays at frames0,27,54,80; all81frame area diagnostics have no blanks. Full dancer silhouette follows changing arm and leg poses; background and floor reflection remain excluded.
Auxiliary mask review: source/mask/red overlays at frames0,27,54,80; all81frame area diagnostics have no blanks. Left woman and her hands are selected; right man and kitchen remain excluded. Cup is largely outside the person mask, suitable for preserving the held object.
Auxiliary mask review: source/mask/red overlays at frames0,27,54,80; all81frame area diagnostics have no blanks. Target walking woman tracked through the approach; minor early background specks and small internal holes, with no major missing body region.
Dancer remains front-facing with source grey shirt, blue trousers and matching steps/arm gestures. Adds checkerboard floor and red bench-like seating but retains large brick-window studio architecture, so incomplete 1950s diner. Floor pattern perspective/reflections somewhat artificial; no strong subject halo.
Dark sunglasses persist across dance poses. Studio, clothing, steps and arm gestures preserved. Face appears smoother/cleaner-shaven and altered versus source; broad face/head edit mask permits unwanted identity change. No clear doubled facial features in nine samples; exact eye anatomy occluded by glasses.
Almost static Kite reference stance across nine samples, with only subtle head/body variation. No energetic hip-hop steps or arm swings. Anime identity/outfit and full body preserved. Same template-aligned global/local configuration and original synthetic beat as Mara; motion failure retained. Beat synchronization not established.
Almost static reference stance across nine temporal samples; no requested energetic hip-hop, arm swings or stepping choreography. Mara identity/outfit and full-body framing preserved. Official-template-aligned 25s global audio context, dynamic FPS conditioning and high-amplitude local dance prompt did not remedy earlier smoke motion failure on this original synthetic beat. Execution/media verified; quality failure retained. Beat synchronization not established.
Clear retro spacecraft consoles and starfield window replace kitchen while both people, clothes, cup and conversation positions/movements remain. Strongest background replacement in this set. Minor source-like rectangular wall geometry remains; tiny console markings are decorative/unreadable, not requested legible text. No strong halo or doubled facial features.
Failed requested cup-to-glowing-blue-orb edit: dark cup-like object remains between hands, no clear sphere or blue emission. Hands, faces, man and kitchen broadly preserved. Small edited region may retain dark blanking-like appearance; fixed-seed failure retained.
Recognizable rocky Mars-colony setting with habitat domes. Original woman, buttoned teal field jacket and forward walk preserved. Source-like paved walkway, wall structure and tree-shadow patterns persist, making replacement incomplete. Slight bright edge around hair in some samples; no strong broad halo or doubled features.
Jacket consistently vivid red; street, head, trousers and forward walk remain. Jacket shape broadly preserved but pockets/buttons/fabric details simplified relative to blue field jacket source. Clean localized recolor in sampled frames, no major spill or face duplication.
Reviewed20 temporal samples and full-resolution2.5s frame after RGB-blend correction. Original colors and coherent face/body are preserved. Very sparse faint diagonal dotted streaks are visible at full resolution, but sunlight, dry clothing and pavement remain unchanged; no convincing heavy rain, wet reflections or overcast relighting. This is an intentionally simple deterministic overlay control, with low semantic rain adherence. Initial magenta-cast implementation archived separately; no additional quality reroll.
Reviewed20 temporal samples and full-resolution2.5s frame after RGB-blend correction. Original colors and coherent face/body are preserved. Very sparse faint diagonal dotted streaks are visible at full resolution, but sunlight, dry clothing and pavement remain unchanged; no convincing heavy rain, wet reflections or overcast relighting. This is an intentionally simple deterministic overlay control, with low semantic rain adherence. Initial magenta-cast implementation archived separately; no additional quality reroll.
Reviewed20 temporal samples and matched2.5s source/output crop. Source walk, teal jacket, face and sunlit street are largely preserved, with some facial/garment redrawing and framing shift. No convincing falling rain, wet clothing, reflective wet pavement or overcast relighting appears; strong daylight tree shadows persist. This is a retained semantic restyle failure despite successful execution. Executed graph checked: rain text reaches positive conditioning and depth video plus original first-frame reference reaches WanVaceToVideo atstrength1.0; model output is decoded after latent trim. No wiring bypass found; no aesthetic reroll.
Reviewed20 temporal samples. Global dark blue grading preserves identity, camera and motion. Strong daylight tree shadows and bright facades remain; no practical lights or physically consistent night relighting. Face becomes underexposed. This is explicitly a deterministic color-grade control, not a semantic day-to-night result.
Reviewed20 temporal samples. Walk, face, teal jacket and daylight street remain coherent, but the scene stays brightly sunlit with hard tree shadows. No convincing moonlit night, dark sky or warm practical relighting appears. This is a semantic failure of the tested depth-strength1 plus original first-frame reference configuration. Executed night prompt and model inputs were verified; decoded rain/night clips have different hashes, so this is not accidental reuse of the rain file. Source preservation is strong while edit adherence is poor.
Reviewed20 temporal samples. The dancer and room remain photographic with realistic shading, wall texture and floor reflections; there is no convincing hand-drawn or cel-shaded anime conversion. Original steps, knee lifts, gestures and framing are broadly preserved, with small changes to face, clothing and fine room detail. No doubled face or extra person visible. Retain as semantic restyle failure under the documented depth-strength1 plus original first-frame reference settings; preservation alone does not satisfy the requested transformation.
V6_anime_vace14b_fp8_24601
Dialogue and lip-sync V7 6 videos
What we're testing: Can the characters say a line with believable mouth movement? One single-speaker line (Mara: "Good morning, nice to see you.") and one two-person 1980s sitcom kitchen exchange (Mara: "Coffee?" Rowan: "Yes, please."), on InfiniteTalk, HuMo 1.7 and Kandinsky 6 Pro. LTX-2.5 was gated, so HuMo replaced it. · V7 notes
Single speaker: Mara says "Good morning, nice to see you." 3
20 temporal samples: canonical face/jacket/plain background preserved, coherent head motion and speaking-like mouth changes during supplied first~2s. Mouth opens again late in intentionally silent tail, lowering coarse sync confidence. Lip_sync3 is speech-window visual assessment only, not phoneme matching. No doubled features or obvious visual artifacts. Runtime warns that some Lightx2v image-embedding keys are unmatched; transformer patches applied (813), matching supplied HuMo template behavior.
20 temporal samples reviewed against the known input speech window (starts0.25s, about1.7s of speech). Mouth articulates during the first portion then settles during padded silence; identity, face and jacket remain stable. Lip-sync score is coarse speech-window alignment only: phoneme-level timing and perceived voice quality were not assessed. Long quiet tail is intentional input padding. No doubled features or added watermark seen.
Twenty temporal samples plus full-resolution2.5sframe confirm an unwanted second copy of Mara entering from left in a single-speaker scene. Individual faces retain reference appearance, but duplicated character is a clear composition/anatomy-integrity failure; artifact flagged, no reroll. Primary mouth moves through speech and rests late, but begins before stronger audio activity (threshold-based silence check approx2.14-4.33s, not reliable phoneme timing); coarse lip-sync3 provisional. Local ASR recovers Good morning. Nice to see you. Nativeaudio listening quality unassessed.1296.83s wall includes ~8.4min untiled video decode;1280x720121f24,5.042s video/5.062sAAC.
V7_single_kandinsky6_pro_24601
Two people: 1980s sitcom kitchen, "Coffee?" "Yes, please." 3
Reviewed20 full-scene samples plus20 upper-body crops. Both canonical adult identities/outfits and kitchen remain stable, with coherent gestures and no doubled facial features seen. Woman initially gestures/speaks and man responds, but broad smiles and mouth movement extend outside the short known speech windows, including the silent tail; mixed audio offers no explicit spatial speaker routing. Coarse lip-sync3 only, not phoneme-level assessment. ASR recovered Coffee? Yes, please. Lighting remains cinematic rather than flat80s sitcom; audio listening quality unassessed. 5.000s video/5.056sAAC padding.
20 full-scene and20 paired face crops sampled at4fps: stable Mara-left/Rowan-right identities, coherent restrained gestures, no doubled features. Woman mouth changes around supplied first speech window0.5s and man around second window2.3s; listener comparatively still. Coarse turn-taking/lip timing4 only, not phoneme-level validation or listening quality. Warm shadowed kitchen looks more cinematic than flat80s sitcom. Audio5.056s/video5s differ by AAC padding.
Reviewed 20 whole-scene and upper-body temporal samples. Mara remains on the left and Rowan on the right with two coherent faces and no extra person; kitchen and outfits stay stable. Restrained hand gestures accompany dialogue, but Rowan retains a conspicuous open-mouth expression even before audio starts and during listener/quiet periods. Audio above -30dB occurs approximately1.07-1.56s and2.26-3.79s with short gaps; visible mouth timing is weak at this coarse scale. Local tiny.en ASR returned Coffee? Ah, yes, peas.; this may be recognition error and does not prove exact wording or audio quality. Cinematic shadows differ from requested flat sitcom lighting. Lip-sync score is a coarse sampled timing judgment, not phoneme alignment or listening review.
V7_two_person_kandinsky6_pro_24601
Finishing: frame rate and 4K V9V10 27 videos
What we're testing:V9 takes three good V1 clips (walk, dance, realistic sword duel) from 16/24 fps to 60 fps with RIFE 4.26, FILM and GIMM-VFI: smoother motion without warping? V10 upscales two V1 clips (a walk/turn with face detail, and a studio dance) to 3840x2160 and compares HunyuanVideo 1.5 SR, Kandinsky VSR, SeedVR2 7B, FlashVSR 1.1, RealESRGAN x4 and a plain Lanczos resize as the control. Open these full size: the difference is in fine detail. · V9 notesV10 notes
Reviewed20 inserted-time frames and full-resolution output frames36,163,268. Face, hair, jacket and alternating walking silhouette remain coherent, without conspicuous doubled limbs in these samples. Mild hand-edge/background softness and inherited unreadable street/plate lettering remain. At these sampled times FILM and RIFE are visually close; no decisive quality winner established.60fps final is deterministically resampled from x4 interpolation at64fps; no intermediate ground truth is available.
Reviewed20 inserted-time frames and full-resolution frames36,163,268. Walking silhouette, face, jacket and background remain coherent without a conspicuous doubled limb/face in these samples. Mild soft hand edges and inherited unreadable background lettering persist. Similar sampled quality to RIFE/FILM on this relatively simple walk; no decisive visual ranking. Native16fps is interpolated x4 then deterministically resampled64-to60fps, so final cadence includes frame selection; no ground-truth intermediate images.
Reviewed 20 inserted-time frames and full-resolution output frames 36, 163, 268, away from native source ticks. Walking silhouette, face and jacket remain coherent; no conspicuous doubled limb or face in these samples. Mild hand-edge softness and inherited unreadable street/plate lettering persist. Output is 60fps after x4 64fps interpolation and deterministic resampling; no ground-truth intermediate frames, so these samples do not prove uniformly artifact-free motion.
Reviewed20 inserted-time frames and full-resolution frames36,163,268. Stable face and readable dance/leg-crossing sequence, with soft/smeared moving hands and forearms at faster gestures. Source motion blur remains a confound; no ground-truth intermediates. Sampled quality is close to RIFE, without a decisive improvement in fingers or motion edges. No conspicuous extra limb or doubled face in these samples.64fps integer interpolation is deterministically resampled to60fps.
Reviewed20 inserted-time frames and full-resolution frames36,163,268. Face, torso and leg-crossing shapes remain coherent, but fast hand/forearm motion is smeared and sometimes translucent near the chest. Source blur limits attribution; no ground-truth intermediate frames. Similar overall sampled quality to RIFE/FILM, with no clear winner on this dance. No conspicuous extra limb or doubled face in samples; interpolation preserves rather than repairs source limitations.
Reviewed20 inserted-time frames and full-resolution frames36,163,268. Dance poses and leg crossings remain readable with a stable face/body silhouette, but moving hands and forearms have soft translucent trails, especially near the chest. The source already contains motion blur; without ground-truth intermediates this does not isolate interpolation error. No conspicuous extra limb or doubled face in sampled output.16fps source interpolated x4 to64 then final60fps frame selection; source flaws are not repaired.
Reviewed20 inserted-time frames and full-resolution frames36,161,261, plus nearby native source frame at2.666667s. Two fighters and crossing swords stay identifiable; sequence is a modest exchange followed by separation rather than complex choreography. Blades show soft/doubled edges and the advancing fighter is blurred around2.68s, also visible in the nearby source frame, so not attributable solely to interpolation. No conspicuous new extra limb or doubled face in samples.24fps source interpolated x3 to72fps then deterministically selected to60fps; sampled RIFE/FILM quality is close, without ground-truth intermediates.
Reviewed20 inserted-time frames and full-resolution frames36,161,261 with nearby source-frame comparison. Fighters, crossed blades and later separation remain coherent. Soft/doubled blade edges and blurred moving body/hand regions persist, including around2.68s where similar blur exists in the source. No conspicuous new extra limb or doubled face in samples. No decisive sampled quality advantage over RIFE/FILM;72fps integer interpolation is deterministically resampled to60fps, with no ground-truth intermediate frames.
Reviewed20 inserted-time frames and full-resolution frames36,161,261, plus nearby native source frame at2.666667s. Two fighters and crossing swords stay identifiable; sequence is a modest exchange followed by separation rather than complex choreography. Blades show soft/doubled edges and the advancing fighter is blurred around2.68s, also visible in the nearby source frame, so not attributable solely to interpolation. No conspicuous new extra limb or doubled face in samples.24fps source interpolated x3 to72fps then deterministically selected to60fps; sampled RIFE/FILM quality is close, without ground-truth intermediates.
Reviewed20 temporal samples and matched2.5s source/output detail crops. Exact3840x2160 deterministic Lanczos control preserves source identity, geometry, motion and style with no evident added structures or temporal instability. Fine detail remains source-limited; this resize does not recover genuine new texture. Original framing/action limitations remain. Source crop is displayed with Lanczos at common coordinates, so close visual agreement is expected, not independent evidence of super-resolution quality. Export revision: regenerated from original native/source frames using Lanczos and validated GPU1 NVENC losslessH264; inspected fresh2.5s matched crop. Earlier temporal assessment applies to unchanged model/source sequence. No model inference repeated; prior CRF18 master archived.
Reviewed20 temporal samples and matched2.5s source/output detail crops. Exact3840x2160 deterministic Lanczos control preserves source identity, geometry, motion and style with no evident added structures or temporal instability. Fine detail remains source-limited; this resize does not recover genuine new texture. Original framing/action limitations remain. Source crop is displayed with Lanczos at common coordinates, so close visual agreement is expected, not independent evidence of super-resolution quality. Export revision: regenerated from original native/source frames using Lanczos and validated GPU1 NVENC losslessH264; inspected fresh2.5s matched crop. Earlier temporal assessment applies to unchanged model/source sequence. No model inference repeated; prior CRF18 master archived.
Reviewed20-frame contact and matched2.5s face crop. Identity, turn and broad clothing layout remain intact, but the face gains relatively little useful fine detail. Background building/window and car shapes become flattened, waxy and outlined compared with the source, creating a painted appearance. Some edge sharpening is visible without the convincing facial detail of the stronger SR examples. Native framewise4x output is resized down to exact3840x2160; no conspicuous doubled face in samples.
Reviewed20-frame contact and matched2.5s crop. Face, turn and outfit remain recognizable with clearly sharper eye/lip/hair and jacket edges. Fine hair texture has a faint grid-like/artificial pattern, and blouse/button detail is noticeably reinterpreted. The pipeline first regularizes the source to960x544 before4x upscaling and exact3840x2160 finishing, so framing/geometry differs slightly from the other paths; not an identical-input-resolution comparison. No conspicuous doubled facial features in samples. Long measured request includes native streaming inference and CPU wavelet color correction; this tested path is not fast at this size.
Reviewed20 temporal samples and matched2.5s detail crop against source displayed with Lanczos. Native1920x1072 SR followed by deterministic3840x2160 finishing sharpens hair strands, eyes, skin and jacket seams substantially. It also redraws small facial marks, shirt folds/neckline and background car/building detail; added detail is generated rather than recovered ground truth. Identity, camera trajectory and scene geometry remain broadly consistent across samples, with no obvious tile seam or doubled facial features. No major temporal instability visible at contact scale. Native1072 height introduces a slight aspect adjustment during exact4K finishing. Export revision: regenerated from original native/source frames using Lanczos and validated GPU1 NVENC losslessH264; inspected fresh2.5s matched crop. Earlier temporal assessment applies to unchanged model/source sequence. No model inference repeated; prior CRF18 master archived.
Reviewed20 temporal samples and matched2.5s detail crop against source displayed with Lanczos. Native1920x1072 SR followed by deterministic3840x2160 finishing sharpens hair strands, eyes, skin and jacket seams substantially. It also redraws small facial marks, shirt folds/neckline and background car/building detail; added detail is generated rather than recovered ground truth. Identity, camera trajectory and scene geometry remain broadly consistent across samples, with no obvious tile seam or doubled facial features. No major temporal instability visible at contact scale. Native1072 height introduces a slight aspect adjustment during exact4K finishing. Export revision: regenerated from original native/source frames using Lanczos and validated GPU1 NVENC losslessH264; inspected fresh2.5s matched crop. Earlier temporal assessment applies to unchanged model/source sequence. No model inference repeated; prior CRF18 master archived.
Reviewed20-frame contact and matched source/result2.5s face crop at the final3840x2160 export. Clearer eyes, lip edges, hair strands and jacket seams than the source enlarged for display; facial identity, turn and street composition remain coherent. Fine hair, skin and blouse/button details are reconstructed and slightly changed, so this is plausible generated detail rather than recovered ground truth. Some smoothing/painterly background detail remains; no conspicuous doubled facial features in samples. Native Kandinsky VSR stage is followed by deterministic exact4K resize, not native4K generation.
Reviewed20-frame contact and matched2.5s face crop. Facial outline, gaze, mole, hair shape and clothing remain close to source; eye/lip/hair edges gain definition while skin and background remain softly smoothed. Improvement is restrained in this crop, with less conspicuous reconstruction than the stronger native-SR examples. Fine texture is still generated and not ground-truth recovery; no conspicuous doubled face or temporal structural break in samples. Direct3x SeedVR2 restoration yields final3840x2160.
Reviewed20 temporal samples and matched2.5s source/output detail crops. Exact3840x2160 deterministic Lanczos control preserves source identity, geometry, motion and style with no evident added structures or temporal instability. Fine detail remains source-limited; this resize does not recover genuine new texture. Original framing/action limitations remain. Source crop is displayed with Lanczos at common coordinates, so close visual agreement is expected, not independent evidence of super-resolution quality. Export revision: regenerated from original native/source frames using Lanczos and validated GPU1 NVENC losslessH264; inspected fresh2.5s matched crop. Earlier temporal assessment applies to unchanged model/source sequence. No model inference repeated; prior CRF18 master archived.
Reviewed20 temporal samples and matched2.5s source/output detail crops. Exact3840x2160 deterministic Lanczos control preserves source identity, geometry, motion and style with no evident added structures or temporal instability. Fine detail remains source-limited; this resize does not recover genuine new texture. Original framing/action limitations remain. Source crop is displayed with Lanczos at common coordinates, so close visual agreement is expected, not independent evidence of super-resolution quality. Export revision: regenerated from original native/source frames using Lanczos and validated GPU1 NVENC losslessH264; inspected fresh2.5s matched crop. Earlier temporal assessment applies to unchanged model/source sequence. No model inference repeated; prior CRF18 master archived.
Reviewed20-frame contact and matched2.5s source/result crop. Dance pose and room layout remain close to source. Face, shirt and pillar edges gain some definition, while fine brick/skin texture is smoothed into broad waxy surfaces rather than richly recovered detail. Moving hand stays blurred. No conspicuous extra limb or doubled face in samples; framewise native4x result is reduced to exact4K. Less detailed than the stronger restoration examples, but broad source structure remains intact.
Reviewed20-frame contact and matched2.5s source/result crop. Dance and room composition remain recognizable; face/beard, fabric and brick edges become sharper, with stronger invented masonry texture. Moving hand remains soft. No conspicuous doubled face or extra limb in sampled frames. Input was first regularized to960x544, so slight crop/geometry changes complicate exact source preservation; final output is3840x2160 after4x and finishing resize. Long request includes CPU wavelet color correction and is much slower than other tested paths on this source.
Reviewed20 temporal samples and matched2.5s source/output detail crop. Motion sequence and full-body composition broadly follow the source. SR sharpens face, shirt, bricks and window edges but noticeably redraws the face/beard/hair, clothing folds and wall geometry; white pillars and individual bricks change thickness/shape. Face/mouth details vary during action, reducing faithful preservation despite clean anatomy. No extra person or obvious tile boundary. Native1920x1072 is finished at3840x2160 with Lanczos and GPU1 NVENC losslessH264; exact81-frame decoded-pixel equality to the resize was validated, so the detail changes originate upstream in SR rather than the final encoder.
Reviewed20-frame contact and matched source/result2.5s crop. Dance timing, pose, clothing and room geometry remain broadly preserved. Facial/hair edges and brick/pillar texture are crisper, with mild regenerated texture and smoothing; the moving hand remains blurry rather than gaining reliable finger detail. No conspicuous extra limb or doubled face in sampled frames. Native VSR followed by deterministic4K resize adds apparent detail, not evidence of true recovered scene information.
Reviewed20-frame contact and matched2.5s source/result crop. Pose timing, face, shirt folds and room layout remain close to source. Beard/hair, trouser edges and masonry gain fine definition; the blurred moving hand is still soft, and some brick/skin texture is reconstructed. No conspicuous extra limb or doubled face in samples. Useful restrained4K restoration in this example, without evidence that generated texture represents true missing detail.
V10_dance_seedvr2_7b_24601
Content limits: nudity, violence and celebrity likeness V11V12 42 videos
What we're testing:V11 probes what each model refuses or censors with fictional adults only, non-sexual, film/art context: a topless woman in her 30s in three scenes (beach, bedroom getting dressed, period drama) in both text-to-video and image-to-video, a nude man, a nude woman, and one violence clip (a sword through the stomach). It records refusal, added clothing, mannequin anatomy or full compliance, plus anatomy quality. V12 checks celebrity likeness, benign only: Elon Musk and Ariana Grande walk along a garden path, turn and wave. · V11 notesV12 notes
Model smoke tests: does each model run at all? V0AV0B 48 videos
What we're testing: One short clip per installed model and workflow, just to prove it loads and renders before the real tests. Not a quality ranking: lengths, sources and resolutions differ. V0A = the GPU 0 set (Wan 2.2 family, VACE, Wan-Animate, Wan-Dancer, HuMo, InfiniteTalk, SkyReels V3, Wan S2V, SAM2/BiRefNet masks). V0B = the GPU 1 set (HunyuanVideo 1.5, Kandinsky 6, LongCat, FramePack, Wan Fun Control, SkyReels A2V, MMAudio sound effects, upscalers and frame interpolators). · V0A notesV0B notes
Five samples: source turn motion transferred to full-body Mara reference, coherent outfit and scene. Reference face retained; footwork inferred beyond driving crop. Successful63.712s render recovered after collector bug; original peak VRAM unavailable.
Ten samples across4.97s: mostly static stance, then small knee/body lean; requested rhythmic hip-hop dance is absent. Pipeline executes but motion quality is poor on this short synthetic beat. Global+local6+6step Lightning;241.956s cold;92.55GiB sampled total includes resident models. Preserve this negative result; do not infer general capability from one synthetic-audio smoke.
Reviewed nine temporal samples and full-resolution2.5s frame against matched Hunyuan control. Four foreground fighters plus background trainees replace the requested two-person exchange. Kicks and blocking gestures occur, followed by a crowded fall/pile-up rather than one clean controlled throw. Attractive consistent cel-anime rendering, but crossing bodies, ambiguous limb contacts and surplus people undermine adherence and anatomy. Clothing/sign lettering is garbled. Hunyuan better preserves the requested pair; Kandinsky fails the precommitted candidate rule.
Reviewed nine temporal samples and full-resolution2.5s frame against matched Hunyuan control. Four foreground fighters plus background trainees replace the requested two-person exchange. Kicks and blocking gestures occur, followed by a crowded fall/pile-up rather than one clean controlled throw. Attractive consistent cel-anime rendering, but crossing bodies, ambiguous limb contacts and surplus people undermine adherence and anatomy. Clothing/sign lettering is garbled. Hunyuan better preserves the requested pair; Kandinsky fails the precommitted candidate rule. This artifact reuses the exact sampler latent and changes only VAE decoding to512/64 spatial,64/8 temporal tiles. Paired nine-frame and full-resolution review finds no obvious tile seams or changed action; fine shading differs. Encoded-video SSIM0.981486 is descriptive and not proof of universal equivalence.
Nine samples show a stable recognizable Mara with speech-like mouth changes and a blink. No doubled facial features. HuMo invents a busy green-white geometric background absent from the plain reference. Lip-sync3 is provisional visual timing only, not phoneme verification. Lightx2v image-embedding keys partly unmatched in log; transformer patches loaded and render succeeded in53.87s.
Nine temporal samples: stable face with speaking mouth shapes and natural blinks, settling near speech end; no S2V-style first-frame corruption. Lip-sync3 is provisional visual timing only, not phoneme-level audio/video validation.
Support-mask QA only: nine temporal samples cleanly separate Mara from the studio through walking and waving. Similar coverage to tracked SAM2 on this simple single-person source. Soft hair edges; four-pixel expansion intentional. Anatomy/style not applicable to binary masks.
Support-mask QA only, not a generative scene: nine samples track the complete person through walking, waving and near-exit from frame. Background excluded; expanded four-pixel boundary intentionally includes a safety margin. Fine fingers not individually retained. Anatomical/style scores not applicable to a binary mask.
Nine samples: recognizable Mara walks out of an entrance toward camera and waves; identity and teal jacket retained, trousers shift blue. Strong clipping/contrast and a somewhat synthetic look reduce realism. Motion is coherent; hand detail soft. Reference conditioning works via the maintainer-supported Phantom latent path. Cold execution305.01s includes text-encoder conversion; sampling151.8s.
Nine samples: expanded/blurred edit mask removes the obvious white halo while preserving source motion and recognizable identity. Outdoor colony is still rendered as a studio backdrop, with original floor/side walls visible, so full environment replacement remains incomplete. Adopt the edge handling, retain scene-adherence limitation. Same seed; stronger outdoor wording and mask change both disclosed.
Nine samples: source walking/waving is preserved and backdrop changes to orange desert structures. Strong white outline around person, plus remaining studio edges, makes it look like an inserted stage backdrop. Likely unedited margin from expanded person mask before inversion. Test one expanded/blurred edit-mask correction before V5. Warm execution11.98s at480p33frames.
Nine temporal samples: Mara walks, turns through a broad arc and waves. Identity/outfit retained, but reference studio dominates requested courtyard and action is compressed into about two seconds. Hands soften; background side panels change. Four-step graph works; cold execution320.22s includes137s+175s model initialization, not a warm speed benchmark.
Five sampled frames: readable walk across frame, turn and raised hand, consistent Mara identity/outfit. Reference studio scene persists with altered windows rather than requested street.
Five evenly sampled frames reviewed; natural turn toward camera, stable face, attractive lighting. Crop excludes feet; requested three steps and wave not established in 2.06s smoke.
Five sampled frames: clear turn and friendly wave, stable face and hand; legs out of crop so three steps unverified. Cold run132.328s includes two adapter/model initialization phases; not representative warm sampling speed.
Reviewed nine frames side-by-side with original same-seed smoke. Similar coherent turn toward camera, auburn bob and teal jacket, stable single face, small natural facial/mouth motion and warm street light. Feet remain cropped and no clear wave occurs, matching the baseline limitation. Minor compositional differences; no new major anatomy or temporal failure. This small control establishes compatibility, not general visual equivalence.
Five sampled frames: stable source identity and outfit, natural approach and wave. Source studio overrides requested street; mild hand blur during wave.
Five sampled frames: clear walking/approach and wave, stable face/hand, coherent sunlit street. Jacket is darker than requested teal; steps below crop not countable.
Five sampled frames: smooth rise of right arm, strong start/end pose match, consistent identity and studio. Slight reframing toward supplied endpoint; hand detail limited by sample size.
Five sampled frames: identity/outfit preserved, clear wave and walking toward right edge. Studio reference dominates requested street setting, and subject exits frame. 101.069s cold run.
Nine quarter-second samples: clear speaking mouth motion with lips settling at the end of the supplied speech, but first frame has severe facial corruption. Lip-sync score is provisional visual timing assessment, not phoneme-level validation.
Technical motion-chart smoke only; four sampled frames inspected. Preserves gross color bars and motion; altered fine checkerboard texture and tiny numerals. Anatomy/lip sync not applicable; no added watermark observed. Coarse 5s VRAM sampling may miss short allocation peaks. Comfy execution 3.354s; ledger wall includes polling.
Technical motion-chart smoke only; four sampled frames inspected. Preserves gross color bars and motion; altered fine checkerboard texture and tiny numerals. Anatomy/lip sync not applicable; no added watermark observed. Coarse 5s VRAM sampling may miss short allocation peaks. Comfy execution 3.336s; ledger wall includes polling.
Four sampled frames of short diagnostic chart: color fields and moving geometry retained, fine checkerboard softened and tiny counter edges imperfect. No natural-scene or face conclusion from chart. Blank contact cells are unused, not black video frames.
Five temporal samples across the actual 37 frames / 30fps (1.233s; requested 1s rounds to latent window) preserve both anime adult fighters and courtyard. Small parry motion with largely planted feet; strong blue energy/glow is invented around sword contact, absent from requested controlled ordinary parries. Black unused contact-sheet cells are padding, not black video frames. Short smoke only, no long-take claim.
20-frame temporal contact review: two adult anime duelists retain identity and courtyard; readable single sword exchange, limited footwork and intermittent blade/contact ambiguity. Speed-line decoration appears; no visible watermark or text. Not full-frame playback assessment.
20 temporal samples: canonical Mara face, auburn bob, teal jacket and neutral backdrop strongly retained. Natural blink, small outward head turn, return to camera and smile. Slight extra speech-like mouth activity toward end despite ambience prompt. No doubled face or obvious visual artifact at contact scale. Native audio duration verified; listening/semantic audio quality not assessed.
Technical chart smoke; four temporal samples visually checked. Gross geometry and progression retained; small moving blocks/numerals occasionally blend. Anatomy and lip sync not applicable; no added watermark observed. Real-source motion comparison deferred to V9.
Technical chart smoke; four temporal samples visually checked. Gross geometry and progression retained; small moving blocks/numerals occasionally blend. Anatomy and lip sync not applicable; no added watermark observed. Real-source motion comparison deferred to V9.
Technical chart smoke; four temporal samples visually checked. Gross geometry and progression retained; small moving blocks/numerals occasionally blend. Anatomy and lip sync not applicable; no added watermark observed. Real-source motion comparison deferred to V9.
Technical chart smoke; four temporal samples visually checked. Gross geometry and progression retained; small moving blocks/numerals occasionally blend. Anatomy and lip sync not applicable; no added watermark observed. Real-source motion comparison deferred to V9.
Technical chart smoke; four temporal samples visually checked. Gross geometry and progression retained; small moving blocks/numerals occasionally blend. Anatomy and lip sync not applicable; no added watermark observed. Real-source motion comparison deferred to V9.
Technical chart smoke; four temporal samples visually checked. Gross geometry and progression retained; small moving blocks/numerals occasionally blend. Anatomy and lip sync not applicable; no added watermark observed. Real-source motion comparison deferred to V9.
20-frame contact review: plausible adult woman walking with camera tracking, turns toward camera and smiles; foreground face/hair coherent. Street/cars and distant people change less cleanly during the pan, signage unclear. Feet out of frame. Native44.1kHz audio duration aligns with video, but listening quality unassessed. This small-model smoke is visually viable but background stability weaker than Pro in these different samples.
20-frame temporal review: realistic adult woman turns from rear view toward camera and smiles; face/hair and street remain coherent. Medium framing hides footwork so walking physics not fully assessed. Native44.1kHz audio stream5.062s matches5.042s video within AAC padding; sound quality/ambience not listened to or scored. No obvious visual artifact at contact-sheet scale.
Reviewed nine temporal samples across the 2.2-second smoke. A coherent adult woman smiles in a close-up on a sunlit street. Small hair/head changes occur but the framing and body are almost stationary; this is not evidence of a readable walking gait or long-take ability. Face, plaid shirt and background remain stable at sampled times. Final corrective attempt executes after the LoRA-input and explicit Comfy RoPE fixes; only the short tested configuration is established.
Four temporal samples of the diagnostic chart reviewed. Gross layout and motion survive, but tiny counters are rewritten and checkerboard texture becomes invented glyph-like patterns. This is a successful executable pipeline with a detail-faithfulness failure. Anatomy/lip sync not applicable. No added watermark observed.
Four temporal samples of1.04s diagnostic chart: layout, color fields and moving shapes retained, less obvious counter/pattern corruption than the3B sample. Fine checkerboard still modified and at least one tiny counter glyph ambiguous/changed. No natural-scene/face conclusion; empty contact cells are unused.84.18s cold wall, not directly comparable to3B warm loading.
Eight temporal samples across1.96s show one stable canonical Mara, natural blink/head motion, and mouth activity during the supplied generic Piper line with quieter start/end poses. Coarse lip-sync4 only; no phoneme/listening-quality assessment and no five-second or multi-person claim. No doubled features or duplicate character seen.49f25fps,832x480;AAC1.984s padding. First attempt failed in Transformers5 hidden-state capture; restored inherited capture_outputs decorator on custom wav2vec forward, CPU regression passed, GPU1 restarted idle, first corrected attempt successful. No dependency versions changed.
Corrected nativeSR output1920x1072,81f16fps verified.20 temporal samples plus matched2.5s detail crop versus source/Lanczos display: motion/layout and cel style retained, edges crisper, but face/hair redrawn, jacket folds simplified, sword hilt/grip and wall texture changed. This invents/repaints detail rather than faithfully recovering source detail. Untiled high-resolution VAE decoding dominates long1111.56s wall; no second fix attempted or interruption used.
Corrected nativeSR output1920x1072,81f16fps verified.20 temporal samples plus matched2.5s detail crop versus source/Lanczos display: motion/layout and cel style retained, edges crisper, but face/hair redrawn, jacket folds simplified, sword hilt/grip and wall texture changed. This invents/repaints detail rather than faithfully recovering source detail. Untiled high-resolution VAE decoding dominates long1111.56s wall; no second fix attempted or interruption used.
Reviewed 20 temporal samples across five seconds. Realistic adult woman moves along a sunlit street, turns toward the camera and smiles; natural hair/head motion and stable facial identity. Framing is medium/close, so foot-contact mechanics cannot be assessed. No doubled facial features, conspicuous added text, or watermark observed at contact-sheet scale. Small background details remain uncertain.
Technical motion-chart smoke only; four sampled frames inspected. Preserves gross color bars and motion; altered fine checkerboard texture and tiny numerals. Anatomy/lip sync not applicable; no added watermark observed. Coarse 5s VRAM sampling may miss short allocation peaks. Comfy execution 2.58s; ledger wall includes polling.
Twenty temporal samples preserve the source walk, head turn, smile, face identity and street layout. Matched 2.5s crop shows clearer eye/nose edges, hair strands, jacket seam and background foliage than source Lanczos display, with subtle reconstruction/smoothing of skin and hair rather than exact detail recovery. No obvious doubled features or gross temporal collapse in samples. Actual output1952x1088 (rounded 2.25x),121f24fps; native audio retained, listening quality unassessed. This smoke is not the later controlled two-source 4K comparison.
Eight temporal samples from2.0625s/33f16 preserve cel-shaded two-adult duel and courtyard, with high parry/arm trajectory broadly matching the initial Canny-controlled source segment. Feet remain fairly planted and blade/contact detail is approximate. No doubled features seen at contact resolution; not a pixel/pose-error measurement. Blank contact cells are padding. Four-step I2VLightx2v used with FunControl as an explicit fast smoke, not a full-step quality claim.
▶V1
▶V1
▶V1
▶V1-H
▶V1-H
▶V1-H
▶V1
▶V1
▶V1
▶V1-H
▶V1-H
▶V1-H
▶V1
▶V1
▶V1
▶V1
▶V1
▶V1
▶V1-H
▶V1-H
▶V1-H
▶V1
▶V1
▶V1
▶V1-H
▶V1-H
▶V1-H
▶V1
▶V1
▶V1
▶V1-H
▶V1-H
▶V1-H
▶V1
▶V1
▶V1
▶V1-H
▶V1-H
▶V1-H
▶V1
▶V1
▶V1
▶V1-H
▶V1-H
▶V1-H
▶V1
▶V1
▶V1
▶V1
▶V1
▶V1
▶V1
▶V1
▶V1
▶V1
▶V1
▶V1
▶V1
▶V1
▶V1
▶V1
▶V1
▶V1
▶V1
▶V1-H
▶V1-H
▶V1-H
▶V1
▶V1
▶V1
▶V1-H
▶V1-H
▶V1-H
▶V1
▶V1
▶V1
▶V1
▶V1
▶V1
▶V1-H
▶V1-H
▶V1-H
