Nodes/IAMCCS-nodes/IAMCCS Wan-Animate-2 Extends
ComfyUI Node

IAMCCS Wan-Animate-2 Extends

Long-video wrapper for native WanAnimate2ToVideo. It samples 4n+1 chunks, passes the previous generated frame through continue_motion, advances the raw driving-video offset, trims native latent/image anchors, and optionally composites the generated character over a masked background plate.

By IAMCCS·Created 10 months ago·Updated 4 days ago· 112
IAMCCS Wan-Animate-2 Extends
  • model
  • positive
  • negative
  • vae
  • sampler
  • sigmas
  • reference_image
  • pose_video
  • positive_pose
  • clip_vision_output
  • clip_vision_output_pose
  • reference_character_mask
  • composite_mask
  • background_image
  • background_video
  • images
  • native_images
  • frame_count
  • source_frames
  • trimmed_frames
  • chunk_plan
width832
height480
target_frames0
chunk_length81
noise_seed0
seed_modefixed
cfg1.0
reference_image_strength1.00
pose_strength1.00
pose_start_percent0.00
pose_end_percent1.00
add_noisetrue
enable_context_windowstrue
context_length_latents21
context_overlap_latents8
context_schedulestandard_static
context_fuse_methodpyramid
enable_pose_cachetrue
cache_devicecpu
cache_dtypeint8
reference_background_modekeep_reference_background
output_background_modenative_generated
live_chunk_previewmiddle_frame
empty_cache_each_chunkfalse
CategoryIAMCCS/video/WanAnimate-2

Inputs (39)

NameTypeDefaultDescription
modelMODEL
positiveCONDITIONING
negativeCONDITIONING
vaeVAE
samplerSAMPLER
sigmasSIGMAS
reference_imageIMAGE
pose_videoIMAGERaw driving-video frames. Wan-Animate-2 does not require a DWPose render here.
widthINT83216–8192
heightINT48016–8192
target_framesINT00–1000000 uses the complete driving-video frame count.
chunk_lengthINT815–1025Wan-Animate-2 official workflow default is 81 frames.
noise_seedINT00–18446744073709550000
seed_modeCOMBOfixed2 options: fixed, increment
cfgFLOAT1.00–100
reference_image_strengthFLOAT1.000–10
pose_strengthFLOAT1.000–10
pose_start_percentFLOAT0.000–1
pose_end_percentFLOAT1.000–1
add_noiseBOOLEANtrue
enable_context_windowsBOOLEANtrue
context_length_latentsINT211–1024
context_overlap_latentsINT80–1023
context_scheduleCOMBOstandard_static4 options: standard_static, standard_uniform, looped_uniform, batched
context_fuse_methodCOMBOpyramid4 options: pyramid, relative, flat, overlap-linear
enable_pose_cacheBOOLEANtrue
cache_deviceCOMBOcpu2 options: cpu, gpu
cache_dtypeCOMBOint83 options: int8, int4, default
reference_background_modeCOMBOkeep_reference_background2 options: keep_reference_background, isolate_character
output_background_modeCOMBOnative_generated4 options: native_generated, source_video_composite, reference_image_composite, custom_background_composite
live_chunk_previewCOMBOmiddle_frame4 options: off, first_frame, middle_frame, last_frame
empty_cache_each_chunkBOOLEANfalse
positive_poseoptCONDITIONING
clip_vision_outputoptCLIP_VISION_OUTPUT
clip_vision_output_poseoptCLIP_VISION_OUTPUT
reference_character_maskoptMASKWhite character mask used only by isolate_character.
composite_maskoptMASKWhite selects generated character; black selects the background plate.
background_imageoptIMAGE
background_videooptIMAGE

Outputs (6)

NameTypeDescription
imagesIMAGE
native_imagesIMAGE
frame_countINT
source_framesINT
trimmed_framesINT
chunk_planSTRING