# V5 — final results

24 planned videos completed; 13 auxiliary outputs; 6 planned LTX-2.5 slots skipped because access is gated. All actual outputs have verified metadata, full decode checks and explicit visual scores.

## Findings

Animate-2 is useful here for motion transfer, but its native motion-conditioned generation does not preserve a source scene as a localized edit. Both walking and dancing character replacements adopt the reference identity and studio backdrop, removing the source environment. In the two-person conversation, Mara retains the cup-holding gesture but the man and kitchen disappear; Kite is nearly static and loses the cup as well. These are scored quality failures, not rerolled successes.

The motion-only dance tests transfer broad steps and gestures with good identity. Breakdance tests follow the source arms-out hold and late inverted support, but raised feet leave frame. Mara has elongated supporting fingers/wrist geometry; Kite has a cleaner animated outline. The source itself does not contain a complete windmill/headspin sequence, so these are not evidence of full choreography transfer.

The first VACE character attempt incorrectly supplied original editable pixels in the control video. The official Comfy-Org inpainting template instead composites black into the editable area before WanVaceToVideo. Correcting this preprocessing with the same seed and four-step sampler changed the copied source field jacket into Mara’s reference teal jacket and olive top while retaining the street and walking motion. The original unblanked result is retained as an auxiliary diagnostic; canonical VACE results use the explicitly named `lightning-blanked` recipe. This finding concerns input preparation, not a new random seed or a sampler comparison.

Masks were reviewed separately. The cup mask required two corrections: SAM2 selected hands, and the accepted source-specific pale-component tracker follows the cup. The sunglasses mask covers the face/head region, so face changes are a known edit-locality risk. Source clips and mask provenance are provided alongside outputs.

VACE’s corrected walking swaps are the best character replacements in this batch: Mara’s outfit/hair and Kite’s anime identity transfer while the photographic street remains. Both dance swaps change clothing but turn the originally front-facing dancer away from the camera; this is a pose-preservation failure and prevents facial-identity assessment. Both conversation swaps preserve the man, cup and kitchen, but fail target identity, especially Kite, who remains a photographic long-haired woman.

The bridge background is the strongest environment swap, with recognizable consoles and a starfield while both people remain. Mars is recognizable but retains source-like pavement/wall/shadow structure. The diner adds checkerboard tiles and red seating yet still looks like a brick-window studio. Red jacket recoloring succeeds with simplified garment details. Sunglasses persist but alter the face within the broad mask. The cup-to-orb edit fails: a dark cup-like object remains without a blue glow or sphere.

`preservation_index.json` covers 18 planned edits plus the preprocessing diagnostic. Character edits have outside-mask RGB MAE around 1.7–2.9 for VACE versus 61–70 for Animate-2, matching the visual scene-preservation difference. This unaligned, downsampled 0–255 metric is only a change diagnostic: VACE’s failed Kite conversation and cup edits also score low outside the mask. Background edit outside-mask errors are larger (approximately 6–26), indicating changes around the preserved subjects, and are not directly comparable to person-mask errors. Full measurement definitions are embedded in each JSON.

The mask/source preparation includes one auxiliary conversation source, nine initial masks and two cup-mask corrections. Together with the unblanked VACE diagnostic, these are 13 auxiliary outputs, separate from the 24 planned videos. Six LTX-2.5 edit slots are skipped because model access is gated.

Both Wan-Dancer outputs are almost static despite the energetic dance prompt and supplied original 120bpm beat. Identity is strong, choreography absent. The V5 configuration already corrects the short smoke setup to the official 25s global audio context, dynamic global FPS conditioning and high-amplitude local street-dance prompt, with audited LoRA strengths. These two outputs establish failure on this synthetic beat/configuration, not a general ranking of the model or success at beat synchronization.

Best examples to inspect: VACE walking Kite for mixed anime/photo character replacement; VACE walking Mara for realistic outfit/identity transfer; VACE conversation bridge for background replacement; VACE red jacket for a localized edit; Animate-2 dance motion for broad gesture transfer without scene preservation. The main practical lesson is to select the workflow by task: motion-conditioned generation and masked editing have materially different preservation behavior. Masks and input preparation are part of the benchmark, and low pixel error alone does not mean the requested edit happened.

Timing includes ComfyUI execution and may include cold loading. Sampled peak GiB is total GPU0 residency, not isolated model allocation. Visual scores use the five PLAN axes; additional reference or likeness assessments are separate. No phoneme-level lip-sync accuracy is inferred from silent clips.

| Test | Role | Settings | Seconds | Peak GiB | Scores | Review / flags |
|---|---|---|---:|---:|---|---|
| V5_breakdance_motion_Kite_wan-animate2-bf16_45110 | primary | 10steps,cfg1,lcm,1280x720,81frames,16fps; motion transfer,seed45110 | 340.23 | 73.73 | {"adherence": 3, "motion": 3, "anatomy_face": 4, "temporal": 4, "style": 5, "reference_identity": 5} | 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. Flags: {"artifacts": "raised-leg crop; approximate inverted support and transitions", "text_garbling": false, "watermark": false} Final encoding: Native H.264/yuv420p preserved; FFmpeg faststart remux, metadata stripped |
| V5_breakdance_motion_Mara_wan-animate2-bf16_45109 | primary | 10steps,cfg1,lcm,1280x720,81frames,16fps; motion transfer,seed45109 | 341.29 | 73.73 | {"adherence": 3, "motion": 3, "anatomy_face": 3, "temporal": 4, "style": 5, "reference_identity": 4} | 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. Flags: {"artifacts": "raised-leg crop; elongated supporting fingers/wrists; approximate inverted support", "text_garbling": false, "watermark": false} Final encoding: Native H.264/yuv420p preserved; FFmpeg faststart remux, metadata stripped |
| V5_dance_background_wan22-vace14-lightning-blanked-fp8_45112 | primary | 4steps,cfg1,official masked-source blanking,1280x720,81frames,16fps; background mask,seed45112 | 114.84 | 58.04 | {"adherence": 3, "motion": 4, "anatomy_face": 4, "temporal": 4, "style": 3, "source_preservation": 4} | 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. Flags: {"artifacts": true, "text_garbling": false, "watermark": false} Final encoding: Native H.264/yuv420p preserved; FFmpeg faststart remux, metadata stripped |
| V5_dance_character_Kite_wan-animate2-bf16_45104 | primary | 10steps,cfg1,lcm,1280x720,81frames,16fps; native motion-conditioned regeneration, no spatial edit-mask input,seed45104 | 339.98 | 73.73 | {"adherence": 2, "motion": 4, "anatomy_face": 4, "temporal": 4, "style": 5, "reference_identity": 5, "source_preservation": 1} | 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. Flags: {"artifacts": "reference backdrop replaces source; repeated head/foot crop; small contact/detail ambiguity", "text_garbling": false, "watermark": false} Final encoding: Native H.264/yuv420p preserved; FFmpeg faststart remux, metadata stripped |
| V5_dance_character_Kite_wan22-vace14-lightning-blanked-fp8_45104 | primary | 4steps,cfg1,official masked-source blanking,1280x720,81frames,16fps; tracked target-person mask,seed45104 | 119.43 | 60.67 | {"adherence": 2, "motion": 4, "anatomy_face": 4, "temporal": 4, "style": 3, "reference_identity": 3, "source_preservation": 3} | 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. Flags: {"artifacts": true, "text_garbling": false, "watermark": false} Final encoding: Native H.264/yuv420p preserved; FFmpeg faststart remux, metadata stripped |
| V5_dance_character_Mara_wan-animate2-bf16_45103 | primary | 10steps,cfg1,lcm,1280x720,81frames,16fps; native motion-conditioned regeneration, no spatial edit-mask input,seed45103 | 347.66 | 73.48 | {"adherence": 2, "motion": 4, "anatomy_face": 4, "temporal": 4, "style": 5, "reference_identity": 5, "source_preservation": 1} | 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. Flags: {"artifacts": "reference background replaces source; repeated head/foot crop; small hand/support ambiguity", "text_garbling": false, "watermark": false} Final encoding: Native H.264/yuv420p preserved; FFmpeg faststart remux, metadata stripped |
| V5_dance_character_Mara_wan22-vace14-lightning-blanked-fp8_45103 | primary | 4steps,cfg1,official masked-source blanking,1280x720,81frames,16fps; tracked target-person mask,seed45103 | 124.85 | 60.67 | {"adherence": 2, "motion": 4, "anatomy_face": 4, "temporal": 4, "style": 5, "reference_identity": 3, "source_preservation": 3} | 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. Flags: {"artifacts": true, "text_garbling": false, "watermark": false} Final encoding: Native H.264/yuv420p preserved; FFmpeg faststart remux, metadata stripped |
| V5_dance_detail_sunglasses_wan22-vace14-lightning-blanked-fp8_45116 | primary | 4steps,cfg1,official masked-source blanking,1280x720,81frames,16fps; localized tracked detail mask,seed45116 | 116.11 | 58.04 | {"adherence": 4, "motion": 4, "anatomy_face": 4, "temporal": 4, "style": 5, "source_preservation": 3} | 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. Flags: {"artifacts": true, "text_garbling": false, "watermark": false} Final encoding: Native H.264/yuv420p preserved; FFmpeg faststart remux, metadata stripped |
| V5_dance_motion_Kite_wan-animate2-bf16_45108 | primary | 10steps,cfg1,lcm,1280x720,81frames,16fps; motion transfer,seed45108 | 341.82 | 73.73 | {"adherence": 3, "motion": 4, "anatomy_face": 4, "temporal": 4, "style": 5, "reference_identity": 5} | 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. Flags: {"artifacts": "repeated head/foot crop; approximate hand placements/support", "text_garbling": false, "watermark": false} Final encoding: Native H.264/yuv420p preserved; FFmpeg faststart remux, metadata stripped |
| V5_dance_motion_Mara_wan-animate2-bf16_45107 | primary | 10steps,cfg1,lcm,1280x720,81frames,16fps; motion transfer,seed45107 | 342.58 | 74.21 | {"adherence": 3, "motion": 4, "anatomy_face": 4, "temporal": 4, "style": 5, "reference_identity": 5} | 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. Flags: {"artifacts": "repeated head/foot crop; small hand and support ambiguity", "text_garbling": false, "watermark": false} Final encoding: Native H.264/yuv420p preserved; FFmpeg faststart remux, metadata stripped |
| V5_mask_background_dance_birefnet-support_0 | auxiliary | BiRefNet general foreground inverted; four-pixel dilation and four-pixel blur, validated edge treatment;1280x720,81frames,16fps; auxiliary mask | 14.25 | 43.26 | {"mask_coverage": 4, "temporal": 5} | 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. Flags: {"artifacts": true, "text_garbling": false, "watermark": false} Final encoding: FFmpeg libx264 CRF18 yuv420p faststart (legacy collector) |
| V5_mask_background_talk_birefnet-support_0 | auxiliary | BiRefNet general foreground inverted; four-pixel dilation and four-pixel blur, validated edge treatment;1280x720,81frames,16fps; auxiliary mask | 14.48 | 43.26 | {"mask_coverage": 4, "temporal": 5} | 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. Flags: {"artifacts": true, "text_garbling": false, "watermark": false} Final encoding: FFmpeg libx264 CRF18 yuv420p faststart (legacy collector) |
| V5_mask_background_walk_birefnet-support_0 | auxiliary | BiRefNet general foreground inverted; four-pixel dilation and four-pixel blur, validated edge treatment;1280x720,81frames,16fps; auxiliary mask | 25.67 | 43.26 | {"mask_coverage": 4, "temporal": 5} | 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. Flags: {"artifacts": true, "text_garbling": false, "watermark": false} Final encoding: FFmpeg libx264 CRF18 yuv420p faststart (legacy collector) |
| V5_mask_detail_cup_retry1_sam2-support_0 | auxiliary | auxiliary mask correction1; revised cup-positive/hand-negative points; one-pixel dilation;1280x720,81frames,16fps | 6.79 | 47.26 | {"mask_coverage": 1, "temporal": 5} | 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. Flags: {"artifacts": true, "text_garbling": false, "watermark": false} Final encoding: FFmpeg libx264 CRF18 yuv420p faststart (legacy collector) |
| V5_mask_detail_cup_retry2_deterministic-mask-cpu_0 | auxiliary | CPU color-component cup tracking;ROI420,490,550,615;minRGB205,range28,dilate2;1280x720,81frames,16fps,seed0;auxiliary correction2 | 6.09 | N/A | {"mask_coverage": 4, "temporal": 5} | 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. Flags: {"artifacts": "minor shadowed-edge omissions and dilated edge spill", "text_garbling": false, "watermark": false} Final encoding: FFmpeg libx264 CRF18 yuv420p faststart (legacy collector) |
| V5_mask_detail_cup_sam2-support_0 | auxiliary | SAM2.1 large bf16; explicit tracked positive/negative points; four-pixel dilation;1280x720,81frames,16fps; auxiliary mask | 7.27 | 41.44 | {"mask_coverage": 1, "temporal": 5} | 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. Flags: {"artifacts": true, "text_garbling": false, "watermark": false} Final encoding: FFmpeg libx264 CRF18 yuv420p faststart (legacy collector) |
| V5_mask_detail_jacket_sam2-support_0 | auxiliary | SAM2.1 large bf16; explicit tracked positive/negative points; four-pixel dilation;1280x720,81frames,16fps; auxiliary mask | 9.66 | 41.72 | {"mask_coverage": 5, "temporal": 5} | 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. Flags: {"artifacts": false, "text_garbling": false, "watermark": false} Final encoding: FFmpeg libx264 CRF18 yuv420p faststart (legacy collector) |
| V5_mask_detail_sunglasses_sam2-support_0 | auxiliary | SAM2.1 large bf16; explicit tracked positive/negative points; four-pixel dilation;1280x720,81frames,16fps; auxiliary mask | 7.25 | 41.54 | {"mask_coverage": 4, "temporal": 4} | 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. Flags: {"artifacts": true, "text_garbling": false, "watermark": false} Final encoding: FFmpeg libx264 CRF18 yuv420p faststart (legacy collector) |
| V5_mask_person_dance_sam2-support_0 | auxiliary | SAM2.1 large bf16; explicit tracked positive/negative points; four-pixel dilation;1280x720,81frames,16fps; auxiliary mask | 13.35 | 55.10 | {"mask_coverage": 5, "temporal": 5} | 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. Flags: {"artifacts": false, "text_garbling": false, "watermark": false} Final encoding: FFmpeg libx264 CRF18 yuv420p faststart (legacy collector) |
| V5_mask_person_talk_sam2-support_0 | auxiliary | SAM2.1 large bf16; explicit tracked positive/negative points; four-pixel dilation;1280x720,81frames,16fps; auxiliary mask | 9.85 | 41.60 | {"mask_coverage": 5, "temporal": 5} | 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. Flags: {"artifacts": false, "text_garbling": false, "watermark": false} Final encoding: FFmpeg libx264 CRF18 yuv420p faststart (legacy collector) |
| V5_mask_person_walk_sam2-support_0 | auxiliary | SAM2.1 large bf16; explicit tracked positive/negative points; four-pixel dilation;1280x720,81frames,16fps; auxiliary mask | 14.75 | 54.38 | {"mask_coverage": 4, "temporal": 4} | 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. Flags: {"artifacts": true, "text_garbling": false, "watermark": false} Final encoding: FFmpeg libx264 CRF18 yuv420p faststart (legacy collector) |
| V5_music_dance_Kite_wan-dancer-global-local-fp8_45118 | primary | 6+6steps,cfg1,1280x720,149frames,30fps; original120bpm beat25s global context, first5s local output; dynamic globalFPS conditioning and high-amplitude street-dance local prompt,seed45118 | 788.83 | 76.95 | {"adherence": 1, "motion": 1, "anatomy_face": 4, "temporal": 5, "style": 5, "reference_identity": 5} | 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. Flags: {"artifacts": false, "text_garbling": false, "watermark": false} Final encoding: Native H.264/yuv420p preserved; FFmpeg faststart remux, metadata stripped |
| V5_music_dance_Mara_wan-dancer-global-local-fp8_45117 | primary | 6+6steps,cfg1,1280x720,149frames,30fps; original120bpm beat25s global context, first5s local output; dynamic globalFPS conditioning and high-amplitude street-dance local prompt,seed45117 | 832.52 | 65.31 | {"adherence": 1, "motion": 1, "anatomy_face": 4, "temporal": 5, "style": 5, "reference_identity": 5} | 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. Flags: {"artifacts": false, "text_garbling": false, "watermark": false} Final encoding: Native H.264/yuv420p preserved; FFmpeg faststart remux, metadata stripped |
| V5_source_talk_wan22-t2v14-lightning-fp8_45000 | auxiliary | auxiliary source;4steps,cfg1,1280x720,81frames,16fps,seed=45000 | 223.82 | 64.64 | {"adherence": 4, "motion": 4, "anatomy_face": 4, "temporal": 5, "style": 5} | 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. Flags: {"artifacts": "minor hand/cup contact ambiguity", "text_garbling": false, "watermark": false} Final encoding: FFmpeg libx264 CRF18 yuv420p faststart (legacy collector) |
| V5_talk_background_wan22-vace14-lightning-blanked-fp8_45113 | primary | 4steps,cfg1,official masked-source blanking,1280x720,81frames,16fps; background mask,seed45113 | 115.28 | 58.04 | {"adherence": 5, "motion": 4, "anatomy_face": 4, "temporal": 5, "style": 4, "source_preservation": 4} | 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. Flags: {"artifacts": false, "text_garbling": false, "watermark": false} Final encoding: Native H.264/yuv420p preserved; FFmpeg faststart remux, metadata stripped |
| V5_talk_character_Kite_wan-animate2-bf16_45106 | primary | 10steps,cfg1,lcm,1280x720,81frames,16fps; native motion-conditioned regeneration, no spatial edit-mask input,seed45106 | 342.65 | 73.73 | {"adherence": 1, "motion": 2, "anatomy_face": 4, "temporal": 5, "style": 5, "reference_identity": 5, "source_preservation": 1} | 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. Flags: {"artifacts": "untouched man and kitchen removed; cup/interaction missing; nearly static output", "text_garbling": false, "watermark": false} Final encoding: Native H.264/yuv420p preserved; FFmpeg faststart remux, metadata stripped |
| V5_talk_character_Kite_wan22-vace14-lightning-blanked-fp8_45106 | primary | 4steps,cfg1,official masked-source blanking,1280x720,81frames,16fps; tracked target-person mask,seed45106 | 118.96 | 60.67 | {"adherence": 1, "motion": 4, "anatomy_face": 4, "temporal": 5, "style": 1, "reference_identity": 1, "source_preservation": 4} | 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. Flags: {"artifacts": false, "text_garbling": false, "watermark": false} Final encoding: Native H.264/yuv420p preserved; FFmpeg faststart remux, metadata stripped |
| V5_talk_character_Mara_wan-animate2-bf16_45105 | primary | 10steps,cfg1,lcm,1280x720,81frames,16fps; native motion-conditioned regeneration, no spatial edit-mask input,seed45105 | 341.82 | 73.69 | {"adherence": 2, "motion": 4, "anatomy_face": 4, "temporal": 5, "style": 5, "reference_identity": 4, "source_preservation": 1} | 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. Flags: {"artifacts": "untouched man removed; kitchen replaced; framing changed to single full-body figure", "text_garbling": false, "watermark": false} Final encoding: Native H.264/yuv420p preserved; FFmpeg faststart remux, metadata stripped |
| V5_talk_character_Mara_wan22-vace14-lightning-blanked-fp8_45105 | primary | 4steps,cfg1,official masked-source blanking,1280x720,81frames,16fps; tracked target-person mask,seed45105 | 123.32 | 60.67 | {"adherence": 2, "motion": 4, "anatomy_face": 4, "temporal": 5, "style": 5, "reference_identity": 2, "source_preservation": 4} | 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. Flags: {"artifacts": false, "text_garbling": false, "watermark": false} Final encoding: Native H.264/yuv420p preserved; FFmpeg faststart remux, metadata stripped |
| V5_talk_detail_cup_wan22-vace14-lightning-blanked-fp8_45115 | primary | 4steps,cfg1,official masked-source blanking,1280x720,81frames,16fps; localized tracked detail mask,seed45115 | 114.73 | 58.04 | {"adherence": 1, "motion": 4, "anatomy_face": 4, "temporal": 5, "style": 5, "source_preservation": 4} | 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. Flags: {"artifacts": true, "text_garbling": false, "watermark": false} Final encoding: Native H.264/yuv420p preserved; FFmpeg faststart remux, metadata stripped |
| V5_walk_background_wan22-vace14-lightning-blanked-fp8_45111 | primary | 4steps,cfg1,official masked-source blanking,1280x720,81frames,16fps; background mask,seed45111 | 114.40 | 58.59 | {"adherence": 4, "motion": 4, "anatomy_face": 4, "temporal": 4, "style": 4, "source_preservation": 4} | 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. Flags: {"artifacts": true, "text_garbling": false, "watermark": false} Final encoding: Native H.264/yuv420p preserved; FFmpeg faststart remux, metadata stripped |
| V5_walk_character_Kite_wan-animate2-bf16_45102 | primary | 10steps,cfg1,lcm,1280x720,81frames,16fps; native motion-conditioned regeneration, no spatial edit-mask input,seed45102 | 344.77 | 73.73 | {"adherence": 2, "motion": 4, "anatomy_face": 4, "temporal": 4, "style": 5, "reference_identity": 5, "source_preservation": 1} | 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. Flags: {"artifacts": "source background replaced; late head/foot crop; minor contact/detail ambiguity", "text_garbling": false, "watermark": false} Final encoding: Native H.264/yuv420p preserved; FFmpeg faststart remux, metadata stripped |
| V5_walk_character_Kite_wan22-vace14-lightning-blanked-fp8_45102 | primary | 4steps,cfg1,official masked-source blanking,1280x720,81frames,16fps; tracked target-person mask,seed45102 | 118.80 | 60.67 | {"adherence": 4, "motion": 4, "anatomy_face": 4, "temporal": 4, "style": 5, "reference_identity": 5, "source_preservation": 4} | 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. Flags: {"artifacts": false, "text_garbling": false, "watermark": false} Final encoding: Native H.264/yuv420p preserved; FFmpeg faststart remux, metadata stripped |
| V5_walk_character_Mara_wan-animate2-bf16_45101 | primary | 10steps,cfg1,lcm,1280x720,81frames,16fps; native motion-conditioned regeneration, no spatial edit-mask input,seed45101 | 584.15 | 75.01 | {"adherence": 2, "motion": 4, "anatomy_face": 4, "temporal": 4, "style": 5, "reference_identity": 5, "source_preservation": 1} | 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. Flags: {"artifacts": "source background replaced by reference studio; late head/lower-body crop", "text_garbling": false, "watermark": false} Final encoding: Native H.264/yuv420p preserved; FFmpeg faststart remux, metadata stripped |
| V5_walk_character_Mara_wan22-vace14-lightning-blanked-fp8_45101 | primary | 4steps,cfg1,official masked-source blanking,1280x720,81frames,16fps; tracked target-person mask,seed45101 | 124.16 | 60.67 | {"adherence": 4, "motion": 4, "anatomy_face": 4, "temporal": 4, "style": 5, "reference_identity": 4, "source_preservation": 4} | 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. Flags: {"artifacts": false, "text_garbling": false, "watermark": false} Final encoding: Native H.264/yuv420p preserved; FFmpeg faststart remux, metadata stripped |
| V5_walk_character_Mara_wan22-vace14-lightning-fp8_45101 | auxiliary | 4steps,cfg1,1280x720,81frames,16fps; tracked target-person mask,seed45101 | 223.42 | 58.84 | {"adherence": 1, "motion": 4, "anatomy_face": 4, "temporal": 4, "style": 5, "reference_identity": 2, "source_preservation": 5} | 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. Flags: {"artifacts": false, "text_garbling": false, "watermark": false} Final encoding: Native H.264/yuv420p preserved; FFmpeg faststart remux, metadata stripped |
| V5_walk_detail_jacket_wan22-vace14-lightning-blanked-fp8_45114 | primary | 4steps,cfg1,official masked-source blanking,1280x720,81frames,16fps; localized tracked detail mask,seed45114 | 116.25 | 58.04 | {"adherence": 4, "motion": 4, "anatomy_face": 4, "temporal": 5, "style": 5, "source_preservation": 4} | 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. Flags: {"artifacts": false, "text_garbling": false, "watermark": false} Final encoding: Native H.264/yuv420p preserved; FFmpeg faststart remux, metadata stripped |

## Comparison viewing aids

[Maras source/Animate-2/VACE grid](comparisons/mara_character_swaps.mp4) and [Kites source/Animate-2/VACE grid](comparisons/kite_character_swaps.mp4) show rows for walk, dance and conversation. Both grids passed full decode/metadata checks and were visually checked at 2.5s for panel ordering and labels. These are downscaled viewing aids; use the full-resolution originals for detail.