Wan 2.1 video inpainting
Generate an edited video segment with SAM3 object selection and Wan 2.1 VACE masked video conditioning.
Workflow details
Workflow details
Runtime profile
- Publisher
- Sweet Tea Automation
- Status
- Listed
- Version
- Checking published version…
- Updated
- Oct 26, 2026
Hover nodes for Pipe controls
Core
Primary Pipe surface
Source video
LoadVideo
Object to edit
CLIPTextEncode
Edit prompt
CLIPTextEncode
Reference image
LoadImage
Output width
PrimitiveInt
Output height
PrimitiveInt
Video segment
ImageFromBatch
Fast sampling
PrimitiveBoolean
Extended
Additional surfaced controls
Negative prompt
CLIPTextEncode
Video guidance
WanVaceToVideo
Mask margin
GrowMask
Object detection
SAM3_Detect
Video framing
ResizeImageMaskNode
Fast sampling LoRA
LoraLoaderModelOnly
Sampling
KSampler
Fast sampling steps
PrimitiveInt
Fast sampling guidance
PrimitiveFloat
Standard sampling steps
PrimitiveInt
Standard sampling guidance
PrimitiveFloat
Sampling shift
ModelSamplingSD3
Video encoding
SweetTeaPreviewVideo
What you need
Hardware and download-size estimates are not available yet.
Output
- Sweet Tea Preview Video · video
Requirements
Models
5sam3.1_multiplex_fp16.safetensors
File unverifiedCheckpoint · Hugging Face · Comfy-Org/sam3.1
umt5_xxl_fp8_e4m3fn_scaled.safetensors
File unverifiedText encoder · Hugging Face · Comfy-Org/Wan_2.1_ComfyUI_repackaged
- File unverified
VAE · Hugging Face · Comfy-Org/Wan_2.1_ComfyUI_repackaged
wan2.1_vace_14B_fp16.safetensors
File unverifiedCheckpoint · Hugging Face · Comfy-Org/Wan_2.1_ComfyUI_repackaged
Wan21_CausVid_14B_T2V_lora_rank32.safetensors
File unverifiedLoRA · Hugging Face · Kijai/WanVideo_comfy
Node packs
1Node pack · GitHub · tea-time-labs/sweet-tea-nodes
Sources
1 source
Sources
1 source