MiniMax H3 is a video model, but if you ask it for a single frame it turns out to be a really good image editor. Give it a photo plus the thing you want in it (a logo, a label, a poster, a mural) and it puts it in properly: wrapped around a can, painted into brick, glowing on a wall with the reflection in the wet street. It also does plain text-to-image, and the 16 MP workflows take either one up to about 16 MP while staying sharp when you zoom in.The trick comes from Patient_Ratio4177 on r/StableDiffusion. I packaged it into ComfyUI workflows for Mac and for NVIDIA.<h3 id="what-you-get">What you get</h3>Six workflows:<ul><li>Edit (Mac) and Edit (CUDA): two images in, one edited image out</li><li>Inpaint (Mac) and Inpaint (CUDA): paint over the part you want changed. Everything outside it stays untouched, apart from a soft blend at the edge.</li><li>Edit 16 MP (Mac) and Edit 16 MP (CUDA): the edit workflow at 4 MP, then a 2x latent upscale and a 16-step refine, so about 16 MP comes out. Asking H3 for 16 MP in one go looks fine from a distance but melts when you zoom in; this route stays sharp.</li></ul>Each one opens with an example already loaded (the demo images are in the zip). The inputs you actually touch are grouped at the top left, and there's a how-to note right there.<h3 id="getting-it-running">Getting it running</h3><ol><li>Copy the images from the zip's <code>input/</code> folder into ComfyUI's <code>input/</code> folder</li><li>Install the small <code>h3_single_frame</code> node from the GitHub repo and restart ComfyUI. It's what lets H3 render one frame instead of a video.</li><li>Mac: you also need ComfyUI-GGUF, ComfyUI-ClipProj and comfyui-obvpm. For inpainting (Mac or CUDA): ComfyUI-MAINodes. For the 16 MP workflows: Comfyui_Minimax_h3_latent_Upscaler (LBH-123-AI) and its 0.7 GB model.</li><li>Grab the models. The repo README lists every file, with links and where it goes.</li></ol>Speed: about 11 minutes for a 4 MP edit on an M5 Mac, about 5 minutes on an RTX PRO 4500. The 16 MP workflow takes about 10 minutes on the RTX PRO 4500. CUDA needs a Blackwell card (RTX 50xx / RTX PRO) on driver 580 or newer.<h3 id="writing-prompts-that-work">Writing prompts that work</h3><ul><li>Start with <code>Task: Reference-guided generation.</code></li><li>Refer to your images as <code>&lt;Picture 1&gt;</code>, <code>&lt;Picture 2&gt;</code> (first loaded image = Picture 1)</li><li>Say what each picture is for, and what it is NOT for: "&lt;Picture 2&gt; is the label artwork only. It does not supply a background or lighting."</li><li>Say "exactly one" when you want one of something</li><li>List what has to stay the same in Picture 1</li><li>Go to 4 MP if there's small lettering. At 2 MP small text turns to mush.</li></ul><h3 id="what-it's-not-good-at">What it's not good at</h3><ul><li>A normal edit redraws the whole picture, so fine details can shift a little. If the rest has to stay pixel-perfect, use the inpaint workflow.</li><li>Tattoos look like stickers (last image). Detailed illustrations get distorted, but text and logos hold up.</li><li>Inpainting can leave a very faint grid on smooth areas like sky or plain walls</li><li>Faces from a reference photo don't hold a real likeness when they're small in the frame</li><li>It's not an upscaler for existing photos. Qwen-Image-Edit 2.1 does that job well.</li></ul><h3 id="want-to-script-it">Want to script it?</h3>The repo also has a command-line tool that runs the same workflows: batch edits, several seeds at once with a contact sheet, extending an image past its edges, automatic face fixes, and building huge images one region at a time. <code>pip install git+https://github.com/Bambushu/h3image</code>Repo, full docs and model list: <a target="_blank" rel="ugc" href="https://github.com/Bambushu/h3image">github.com/Bambushu/h3image</a>All brands in the demos are made up and every demo image is AI-generated.

Files with SHA256: 7474acc5be2681ee8df996c492c2de31b49a5c90eba6a4f9baa7f42fd3175499