Create coherent facial expression differences using AnimateDiff & prompt travel
5.0
1 reviews
Description
What this workflow does
When you input a portrait image, it generates coherent facial expression differences.
Method
•Use HQ-SAM to mask the face part
•Generate a video that changes in the order of “Smile→Sad→Shamed→Angry” by AnimateDiff × Prompt travel
•Combine ControlNet Openpose and IP-Adapter-face
•Select the frame with the best expression and finish with Hires.fix.
Tips about this workflow
•Please try other expressions as well. It might be better to make them a little stronger, like (Smile:1.2).
•The prompt is specialized for female anime drawings, so please modify it if you use male or realistic images.
日本語の解説
AnimateDiff × prompt travelで一貫性のある表情差分を作る
References
Discussion
(No comments yet)
Loading...
Reviews
No reviews yet
Versions (1)
- latest (2 years ago)
Node Details
Primitive Nodes (30)
IPAdapterApply (1)
Reroute (29)
Custom Nodes (62)
- ADE_AnimateDiffLoaderWithContext (1)
- ADE_AnimateDiffUniformContextOptions (1)
ComfyUI
- LatentFromBatch (4)
- CLIPTextEncode (7)
- ConditioningCombine (4)
- KSampler (5)
- SaveImage (4)
- VAEDecode (5)
- CLIPSetLastLayer (1)
- VAEEncode (1)
- VAELoader (2)
- SetLatentNoiseMask (1)
- CLIPVisionLoader (1)
- LoraLoader (1)
- CheckpointLoaderSimple (1)
- MaskToImage (1)
- ImageToMask (1)
- LoadImage (1)
- ControlNetApplyAdvanced (1)
- PreviewImage (1)
- ImpactDilateMask (1)
- OpenposePreprocessor (1)
- IPAdapterModelLoader (1)
- PrepImageForClipVision (1)
- ControlNetLoaderAdvanced (1)
- VHS_DuplicateImages (3)
- VHS_VideoCombine (1)
- NNLatentUpscale (4)
- BatchPromptSchedule (1)
- ImageResize (1)
- GroundingDinoSAMSegment (segment anything) (1)
- GroundingDinoModelLoader (segment anything) (1)
- SAMModelLoader (segment anything) (1)
Model Details
Checkpoints (1)
😎-v1.x\meinamix_meinaV11.safetensors
LoRAs (1)
1.5\🖼️\LowRA.safetensors