Nodes/Difforum/Difforum · H3 Shot
ComfyUI Node

Difforum · H3 Shot

Everything a MiniMax H3 first-last-frame shot needs, from Difforum. Give it the frames of a Storyboard / Feedback render / Guide Frames and it returns the first and last frame of the chosen segment, the length on the H3 17k+5 grid, the size on the 32 px grid, and the camera move of that segment in words for the prompt. Clips longer than one H3 generation are split into segments: render segment 0, then feed its last frame as the next segment's first frame.

By chillithebillis·Created 3 months ago·Updated about 7 hours ago· 8
Difforum · H3 Shot
  • frames
  • direction
  • params
  • camera
  • masks
  • first_frame
  • last_frame
  • length
  • width
  • height
  • prompt
  • segments
  • last_mask
  • info
◄segment_lengthauto (from frames)►
◄segment0►
◄shot_description►
◄include_looktrue►
◄prompt_styleH3 structured►
◄soundscape►
◄music►
CategoryDifforum/5 · Video model bridges

Inputs (12)

NameTypeDefaultDescription
framesIMAGEThe clip whose first and last frame H3 should connect (Guide Frames, Storyboard, Feedback render).
segment_lengthCOMBOauto (from frames)Frames per H3 generation. Clips longer than this are split into segments.
segmentINT00–999Which segment to render: 0, then 1 with segment 0's last frame as its first frame...
directionoptDIFFORUM_DIRECTIONThe Director's wire: camera, energy curve, scene prompts, keys and look in one cable.
paramsoptDIFFORUM_PARAMSFrom Setup: frame count, fps, width and height. Plug the same Setup everywhere.
cameraoptDIFFORUM_CAMERAA camera track from Camera (keys), Camera (expressions) or Camera Import. Overrides the Director's camera.
shot_descriptionoptSTRINGWhat is on screen at the start: subject, place, light. Goes into the prompt.
masksoptMASKGuide Frames masks: returns the last frame's revealed area for Fill Reveal.
include_lookoptBOOLEANtrueAdd the Director's look as the style of the shot.
prompt_styleoptCOMBOH3 structuredH3 structured: this segment's scenes, camera, keys and look in MiniMax H3's native FL2VA format (alignment line + fields). sentence: description + camera sentence + look, as plain text.
soundscapeoptSTRINGoverall_soundscape: ambience and physical sounds only.
musicoptSTRINGnon_diegetic_music: instruments, tempo, dynamics. Empty = N/A.

Outputs (9)

NameTypeDescription
first_frameIMAGE—
last_frameIMAGE—
lengthINT—
widthINT—
heightINT—
promptSTRING—
segmentsINT—
last_maskMASK—
infoSTRING—