For Wan VACE, compare the control video and mask separately with the length set on WanVaceToVideo. Longer inputs are truncated; shorter control videos are padded with neutral gray, while shorter masks are padded with 1.0. The node also scales both inputs to the requested width and height, so check the target canvas as well as frame coverage. Its strength controls VACE conditioning, not LoRA weights.
Check frame coverage in order
- Set the intended output
length,width, andheightonWanVaceToVideofirst. The node documentation lists 81 frames as the default length, but use the value in your workflow rather than assuming the default applies. - Check the actual frame count of the control-video input. If it exceeds
length, the extra frames are truncated; if it falls short, the missing frames are filled with neutral gray, value 0.5. With no control video, the node creates a neutral-gray sequence of the target length. - Check the mask input independently. Longer masks are truncated to
length; shorter masks are extended with value 1.0. If no mask is connected, the node uses a full white mask. Treat those as documented node behaviors, not as a universal rule about what black, white, or gray means in other editing tools. - Compare each input’s dimensions and aspect ratio with the requested canvas. The node scales control video and masks to
widthandheight; if their proportions differ, do not assume the original framing remains unchanged. Check the resulting composition in the workflow’s available preview or output. - Confirm the
strengthfield on the VACE node separately from any LoRA nodes. The documentation identifies it as VACE control strength and says LoRA weights are applied through separate nodes.
For the node’s exact input and padding behavior, consult the WanVaceToVideo documentation. Its input definitions also identify width, height, length, and strength as separate controls. When inspecting a workflow, use the node’s actual field labels and input types; connector color alone is only a clue.
Make a frame-coverage worksheet
The following is a hypothetical audit, not a test result or a guarantee about generated content. Assume the target is 81 frames at 832 × 480. Record counts from the media and workflow, then use the table to make a coverage decision before generation.
| Input | Hypothetical input | Node handling at length 81 | Coverage decision |
|---|---|---|---|
| Control video | 90 frames, 832 × 480 | 9 frames beyond the target are truncated | Confirm the desired motion fits within the retained 81 frames |
| Mask | 76 frames, 832 × 480 | 5 frames are padded with 1.0 | Decide whether the final five frames should use this documented padding behavior |
| Control video, alternate case | 76 frames, 832 × 480 | 5 missing frames are filled with neutral gray (0.5) | Mark the ending as lacking supplied video frames; do not count padding as authored motion |
| Canvas | Both inputs are 832 × 480 | Inputs are scaled to the specified dimensions | No size change is indicated by these assumed dimensions; inspect framing if source proportions differ |
Use separate rows for video and mask even if they came from the same edit: equal frame counts do not establish that their content is aligned. This worksheet records the node’s documented length handling and your own acceptance decision; it does not predict the generated result.
Limits and next steps
Padding makes an input sequence reach the requested length, but it does not add newly authored control footage. If the padded interval overlaps important action, prepare a control sequence that contains the intended frames, then recount it. If truncation removes an event you need, increase the target length only if that is the intended output, or revise the control edit to place required content within the retained range.
If the canvas changes the framing, review the scaled input in context and adjust the source composition or target dimensions before generating. The documentation specifies scaling to the target dimensions but does not spell out the resize algorithm or guarantee preservation of aspect ratio. Do not infer crop, stretch, or padding behavior from the node description alone. The documentation also does not define mask-value semantics in terms of black or white meaning “editable”; it says the mask separates inactive and reactive control-video parts. Confirm the actual input type and consult the specific workflow’s documented mask convention rather than importing assumptions from another tool.
Common questions
Does a short mask receive the same gray padding as the video? No. The documented mask padding value is 1.0; neutral gray at 0.5 applies to a short control video.
Does setting VACE `strength` change my LoRA strength? No. They are separate controls: the node’s field is for VACE conditioning, and LoRA weights are handled through separate LoRA nodes.
Can I assume an 81-frame output always means 81 latent frames? No. The documentation gives a separate latent-length formula, `((length - 1) // 4) + 1`. For `length` 81, that formula yields 21 latent frames; this is a latent count, not a replacement for checking video and mask frame counts.
Sources and next step
This guide interprets public documentation and uses explicitly hypothetical acceptance examples, not application tests. Controls and availability can change; verify the cited manual against your installed version.
For work created in Imagild, carry the input requirements and acceptance checklist into the film workspace. The external software controls discussed above belong to their respective products.