ComfyUI Node
Add Timed Image Reference for MiniMax H3
Adds a Qwen-only image reference at a target time. Chain nodes on the H3 CLIP line, then connect the result to the core MiniMax H3 Reference to Video node.
Add Timed Image Reference for MiniMax H3
- clip
- image
- CLIP
◄prompt_taganchor►
◄time_seconds0.000►
◄image_size512►
Categorymodel/conditioning/minimax
Inputs (5)
| Name | Type | Default | Description |
|---|---|---|---|
| clip | CLIP | — | |
| image | IMAGE | Image presented to Qwen at the selected time. It is not VAE-encoded and does not consume a native H3 reference slot. | |
| prompt_tag | STRING | anchor | Use this name with a leading # in the Ref2VA prompt, for example #anchor. The patch resolves it to the correct <Video N> label. |
| time_seconds | FLOAT | 0.0000–9999 | Target time represented by this image reference, in seconds. Sub-second values are preserved to microsecond display precision. |
| image_size | COMBO | 512 | Approximate pixel-area edge used for Qwen vision encoding. 64 is the technical floor but loses detail; 256-512 is more practical. Source preserves the input resolution. |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| CLIP | CLIP | — |