ComfyUI Node

H3 Cast Board

H3 Cast Board — the reference cast, uploaded into the node itself. Roles decide tags, not file order. The H3 guide is explicit that a character, product or style image is NOT automatically a <Picture N>: if it only supplies reusable identity it is a <Subject N>, and <Picture N> is reserved for a concrete first/last/key/edited frame or composition anchor. The four kinds are numbered independently, so <Subject 1> and <Picture 1> can be different assets. Numbering follows card order within each kind, so reordering the board renumbers the tags — and the wired IMAGE outputs move with them, which is what keeps the tags in a prompt pointing at the slot they were written for. The three reference inputs grow on their own: connect ref_image_0 and ref_image_1 appears. That is ComfyUI's own Autogrow, the same mechanism the MiniMax H3 sampler uses for its ref_image_ / ref_video_ / ref_audio_ families, so the two nodes behave identically. Outputs cannot do the same: an output link is bound to a slot INDEX, so a node that adds or hides output slots silently repoints every connection downstream of it. They are a fixed set instead.

By AIJigyasa·Created 2 days ago·Updated 2 days ago· 1
H3 Cast Board
  • ref_images
  • ref_audios
  • ref_videos
  • cast
  • image_1
  • image_2
  • image_3
  • image_4
  • image_5
  • image_6
  • image_7
  • image_8
  • audio_1
  • audio_2
  • audio_3
  • audio_4
  • video_1
  • video_2
  • video_3
  • video_4
  • report
cast_json{"entries": []}
CategoryH3 Planner

Inputs (4)

NameTypeDefaultDescription
cast_jsonSTRING{"entries": []}managed by the Cast Board UI; press JSON on the node to edit it by hand
ref_imagesoptCOMFY_AUTOGROW_V3
ref_audiosoptCOMFY_AUTOGROW_V3
ref_videosoptCOMFY_AUTOGROW_V3

Outputs (18)

NameTypeDescription
castH3_CAST
image_1IMAGE
image_2IMAGE
image_3IMAGE
image_4IMAGE
image_5IMAGE
image_6IMAGE
image_7IMAGE
image_8IMAGE
audio_1AUDIO
audio_2AUDIO
audio_3AUDIO
audio_4AUDIO
video_1IMAGE
video_2IMAGE
video_3IMAGE
video_4IMAGE
reportSTRING