An open-weights omni-modal video model with real stereo sound and 2K output — this powerful model is greatly optimized in ComfyUI and can run locally on a 3060.
The MiniMax H3 Reference to Video node isn’t limited to two images—the example workflow only includes two Load Image nodes by default. Add another Load Image node and connect its IMAGE output to the last empty ref_image input. A new empty input will appear automatically, and you can repeat this for up to nine reference images.
Note that the node inputs are zero-indexed, but prompt references are one-indexed: ref_image_0 corresponds to <Picture 1>, ref_image_1 to <Picture 2>, and so on.
Why does the R2V workflow only allow two image references? Isn't the maximum number 9?
The MiniMax H3 Reference to Video node isn’t limited to two images—the example workflow only includes two Load Image nodes by default. Add another Load Image node and connect its IMAGE output to the last empty ref_image input. A new empty input will appear automatically, and you can repeat this for up to nine reference images.
Note that the node inputs are zero-indexed, but prompt references are one-indexed: ref_image_0 corresponds to <Picture 1>, ref_image_1 to <Picture 2>, and so on.