ComfyUI Node
MiniMax H3 Multi Keyframe [darkilNodes]
A ComfyUI node in darkilNodes/minimax with 14 inputs and 3 outputs.
MiniMax H3 Multi Keyframe [darkilNodes]
- model
- clip
- vae
- first_frame
- last_frame
- middle_frames
- positive
- latent
- model
◄prompt—►
◄width1344►
◄height768►
◄length124►
◄middle_positions►
◄position_unitsratio►
◄middle_fitcover►
◄keyframe_noise_aug0.999►
CategorydarkilNodes/minimax
Inputs (14)
| Name | Type | Default | Description |
|---|---|---|---|
| model | MODEL | H3 diffusion model. The returned model carries the layout patch and has to reach the sampler | |
| clip | CLIP | — | |
| vae | VAE | — | |
| prompt | STRING | — | |
| width | INT | 134432–16384 | — |
| height | INT | 76832–16384 | — |
| length | INT | 1245–3600 | Frame count at 24 fps, snapped up to the model's 17k+5 grid (124 = ~5s, trained range is ~124-362) |
| middle_positions | STRING | One position per image in middle_frames, comma separated. Empty spreads them evenly between the first and the last frame | |
| position_units | COMBO | ratio | ratio: 0.0-1.0 of the clip. frames: absolute frame index on the snapped timeline |
| middle_fit | COMBO | cover | How a middle frame is fitted to the canvas. cover keeps the aspect and crops, like the stock last_frame |
| keyframe_noise_aug | FLOAT | 0.9990–1 | Condition strength for every keyframe row. 0.999 is the stock value, lower values blend noise into the anchors and let the model deviate from them |
| first_frameopt | IMAGE | — | |
| last_frameopt | IMAGE | — | |
| middle_framesopt | IMAGE | Batch of images anchored between the first and the last frame, one anchor per image |
Outputs (3)
| Name | Type | Description |
|---|---|---|
| positive | CONDITIONING | — |
| latent | LATENT | — |
| model | MODEL | — |