Nodes/WanVideoLooper/WanVideo Looper
ComfyUI Node

WanVideo Looper

A ComfyUI node in WanVideoLooper with 27 inputs and 2 outputs.

By SquirrelRat·Created 10 months ago·Updated 7 months ago· 9
WanVideo Looper
  • model_high
  • model_low
  • clip
  • vae
  • start_image
  • positive_prompts
  • clip_vision
  • model_clip_sequence
  • color_match_ref
  • images
  • last_frame
negative_prompt
seed0
steps6
cfg_motion_noise1.0
cfg_noise1.0
sampler_name
scheduler
model_switch_point0.900
sigma_shift8.00
denoise1.00
width832
height480
frame_merge1
duration_sec5
color_match_methodDisabled
color_match_reference_frameFirst Frame
color_match_strength1.00
enable_dry_runfalse
CategoryWanVideoLooper

Inputs (27)

NameTypeDefaultDescription
model_highMODELThe BASE high-noise model (can be pre-patched with global LoRAs).
model_lowMODELThe BASE low-noise model (can be pre-patched with global LoRAs).
clipCLIPThe BASE CLIP model (can be pre-patched with global LoRAs).
vaeVAEThe VAE model used for decoding the final image.
start_imageIMAGEThe seed image that the first loop will start from.
positive_prompts*Connect the 'prompt_list' output from the WanVideo Multi-Prompt node.
negative_promptSTRINGA single, global negative prompt applied to all loops.
seedINT00–18446744073709550000The random seed. This will be used for all loops.
stepsINT61–10000The total number of sampling steps for each loop.
cfg_motion_noiseFLOAT1.00–100CFG for the first step ONLY. A value > 1.0 will engage the negative prompt for initial motion guidance. Set to 1.0 to disable motion guidance.
cfg_noiseFLOAT1.00–100Classifier-Free Guidance scale for both high and low noise models. 1.0 disables guidance.
sampler_nameCOMBOThe sampling algorithm (e.g., dpmpp_2m_sde_gpu).
schedulerCOMBOThe noise scheduler (e.g., karras).
model_switch_pointFLOAT0.9000–1Timestep to switch from high-noise to low-noise model. 0.9 (I2V) or 0.875 (T2V) recommended.
sigma_shiftFLOAT8.000–100Applies sigma shift to both models. 8.0 is a good default.
denoiseFLOAT1.000–1Denoise amount. 1.0 means generate a full new image.
widthINT83264–8192The width of the output video.
heightINT48064–8192The height of the output video.
frame_mergeINT10–8Number of frames to overlap/merge between loops for smooth transitions.
duration_secINT51–10Duration of each segment in seconds. 5s is recommended, max 10.
color_match_methodCOMBODisabled'Disabled': No color matching is applied. 'Reinhard (GPU)': Applies Reinhard color transfer. 'MKL (GPU)': Applies MKL color transfer.
color_match_reference_frameCOMBOFirst Frame'First Frame': Matches all frames to the very first frame of the entire sequence. 'Last Frame': Matches all frames to the very last frame of the entire sequence. 'Sequential': Matches each segment to the last frame of the previous segment. 'Reference': Matches all frames to the optional 'color_match_ref' image.
color_match_strengthFLOAT1.000–1Strength of the internal color matching (1.0 = full match).
enable_dry_runBOOLEANfalseIf enabled, skips sampling and decoding to quickly check setup and logs.
clip_visionoptCLIP_VISION(Optional) A CLIP Vision model for guiding the start image.
model_clip_sequenceoptANY(Optional) Connect a WanVideo Lora Sequencer to use different pre-patched models/clips per segment.
color_match_refoptIMAGE(Optional) Static reference image for internal color matching. If omitted, uses last frame of previous segment.

Outputs (2)

NameTypeDescription
imagesIMAGE
last_frameIMAGE