Who it is forA designer or content maker who has several reference pictures and needs one composite image that uses a specific element from each.
The problemMerging elements from several photos by hand in an editor is slow, and a single-source prompt cannot name which picture supplies which element.
The outcomeOne composite image built from the roles assigned to each reference picture.
What you need first
- ComfyUI 0.37.0 or newer
- A CUDA/PyTorch installation that supports the chosen quantization
- Three model files: minimax_h3_fl2va_pruned_int8_convrot.safetensors, qwen3vl_32b_minimax_h3_nvfp4_awq.safetensors, and an H3 video VAE
- Enough GPU memory for the chosen megapixel size; out-of-memory errors are possible at larger sizes
Steps
- Set up Image Studio and the three model files as described in the README.
- Open the Edit workflow and add picture sockets to Image Edit, which grow as needed up to nine pictures total.
- Load each reference image into its own socket, noting which picture is which.
- Write a prompt that assigns each picture an explicit role, for example keeping the person and clothes from one picture and using the pose from another.
- Queue the workflow and check that each role was applied, then rerun with a different seed if needed.
Prompt for your agent
Using the astropuzzo/ComfyUI-MiniMax-H3-Image-Studio repo, run the Edit workflow in ComfyUI with the reference images <list of image paths> and the role prompt <role prompt>. Expected output: one composite image from Save Image. Verify that each reference contributed the role the prompt assigned to it.
Caveat. Multi-reference results vary and the README does not guarantee that the model follows every role assignment.
Every Monday: the 5 hottest repos and exactly how to use them. Free, one email a week, unsubscribe anytime.
Get them freeShare: X LinkedIn Hacker News Email Copy link
Disclaimer
Deduped lists open-source repositories that belong to other people. We do not own, maintain, audit, or vouch for any of them, and a repo appearing here is not an endorsement or a safety check. The use cases, steps, and prompts are written by AI and may be incomplete, outdated, or just wrong. Where a use case is marked "Seen in the wild", we link to public posts where someone described something similar, but we have not verified those posts and the use case itself is still our AI's own write-up. Use all of this at your own risk. Only run code, commands, or agents on systems and accounts you are authorized to use, and read the code and every command before you run it, especially anything that touches credentials, files, or the network. Nothing here is professional, security, legal, or financial advice. All trademarks and project names belong to their respective owners, and their use here does not imply affiliation. Content is provided as is, without warranties of any kind. Found a mistake, a safety problem, or something you want removed? Email [email protected] and we will look at it promptly. Full disclaimer page