Nodes/ComfyUI-SA-Nodes-QQ/Add Masked Guide for MiniMax H3
ComfyUI Node

Add Masked Guide for MiniMax H3

Anchor an image, a short clip, audio, or a clip with its soundtrack at any frame of a MiniMax H3 video, with an optional mask limiting where the guide applies. The latent output hard-protects masked regions; connect it to the sampler.

By siraxe·Created 11 months ago·Updated about 19 hours ago· 73
Add Masked Guide for MiniMax H3
  • positive
  • latent
  • vae
  • audio_vae
  • image
  • audio
  • mask
  • positive
  • LATENT
frame_idx0
mask_threshold0.50
CategoryQQ/conditioning/minimax

Inputs (9)

NameTypeDefaultDescription
positiveCONDITIONING
latentLATENT
frame_idxINT0-9999–9999Frame index to anchor the image or the clip first frame at. Negative values are counted from the end of the video.
mask_thresholdFLOAT0.50-1–1Binarize the mask above this value for a hard anchor edge (recommended: kills feathering at the mask border). Set below 0 to keep the mask soft, with gray values giving partial guide strength.
vaeoptVAEVideo VAE, needed when an image is connected.
audio_vaeoptVAEAudio VAE, needed when an audio is connected.
imageoptIMAGEImage or video frames to anchor. Multi-frame batches are anchored as a clip and cropped down to the model valid clip lengths: 5, 22, 39... (17k + 5) frames. Batches shorter than 5 frames use only the first image.
audiooptAUDIOSoundtrack to anchor starting at the same frame index, cropped to the video remaining duration.
maskoptMASKMask over the image guide: 1 keeps the guide anchored, 0 lets the model generate those regions freely. A single-frame mask applies to every frame of a guide clip; a multi-frame mask maps onto the clip by relative time.

Outputs (2)

NameTypeDescription
positiveCONDITIONING
LATENTLATENT