Load one photo of a person, and MiniMax H3 Image draws a character sheet: the same person from the front, the three-quarter view, the side and the back, side by side in one picture. It draws the sheet at up to 16 megapixels in one run, so there is no upscale step.
Watch the H3 Image video:
https://youtu.be/GTjyJNgIG6Q
Switches show the whole body from a half-body photo, save each view as its own picture, put your photo beside the sheet, and add a row of head close-ups or four expressions under it.
This is the MiniMax H3 Image version of my MiniMax H3 Character Sheet. That one uses the H3 video model. This one uses the new image model: the sheet comes out at 16 megapixels instead of about 5, in about half the time.
How to use it
Load your picture: one person, clearly in view.
Optional: type a background, like "a neon city street at night". Left empty, it is plain white.
Optional: name what your photo doesn't show, like "blue jeans and white sneakers" for a half-body photo.
Run. The first run takes a few minutes, because about 38 GB of models load once.
What you get
One picture, 5504 × 3040 at 16 MP: four views in a row.
The views are framed like your photo. A half-body photo gives half-body views.
Turn Head to toe on to see the whole body, down to the shoes. H3 Image makes up what your photo doesn't show. To choose it yourself, name it in What the photo doesn't show.
The back is H3 Image's guess from what your photo shows.
With Close-ups under the sheet on, the same picture gets a row of four head close-ups under the sheet: the front, the three-quarter view, the side and the back of the head. The picture is then 5504 × 4416.
Size, Body, VRAM and Grain cleanup
Size: 16 MP is the full sheet, about 25 seconds. 4 MP is half as wide and half as tall, for quick tries, about 11. 8 MP sits in between.
Body: Short, Regular or Tall. It sets how tall each view is. H3 Image fills a view with the person, so a taller view draws a taller, slimmer body with longer legs. Regular suits most people. Pick Short for a shorter character and Tall for a taller one.
VRAM: High VRAM is the int8 model file, for 32 GB cards. Low VRAM is the smaller w4a8 file, for 24 GB cards.
Grain cleanup: Off, Light, Medium or Strong. It clears the fine speckle a picture can get on skin and cloth. It opens on Off. Light changes the sheet least, Strong is the smoothest. It takes no extra time. Fast skips it.
The switches
Fast: 2 steps with the pDMD LoRA instead of the model's 6. About 14 seconds a sheet instead of 25.
Head to toe: every view shows the whole body.
Save each view too: the four views are also saved as pictures of their own. Each cut goes to the gap between two views, so no view gets part of its neighbour.
Your photo on the left: your photo goes on the left of the sheet, at the same height, with a thin white line between. The views from Save each view too stay without it.
Close-ups under the sheet: H3 Image runs a second time, from your photo and the finished sheet, and draws the four close-ups. About 30 seconds more, 16 with Fast on. Save each view too still saves the four views of the sheet only.
Expressions instead of head turns: the four close-ups face the camera and show the expressions in the four boxes in 3 · Expressions. Those boxes light up when this is on. They run left to right, like the close-ups. It needs Close-ups under the sheet on too.
What you need
ComfyUI 0.45.0 or newer. That is the first release with MiniMax H3 Image.
AusBoss nodes 2.26.5 or newer. It is waiting for review in the Comfy Registry, so ComfyUI-Manager may not offer it yet. Until it does, install the nightly version of AusBoss nodes: it is the newest code and has everything this workflow needs. How, in two short videos: ComfyUI Nodes Manager (the Extensions button) · classic ComfyUI Manager (the Manager button).
The models:
High VRAM · minimax_h3_image_pruned_int8_convrot.safetensors →
models/diffusion_models· 21.0 GBLow VRAM · minimax_h3_image_pruned_w4a8.safetensors →
models/diffusion_models· 12.5 GBBoth · qwen3vl_32b_minimax_h3_nvfp4_awq.safetensors →
models/text_encoders· 15.7 GB (every H3 workflow uses it, so you may have it already)Both · minimax_h3_image_vae_fp16.safetensors →
models/vae· 1.2 GBOnly for Fast · minimax_h3_pdmd_2nfe_comfyui.safetensors →
models/loras· 1.96 GB
Tips
A view came out wrong? Run again. The seed changes after every run.
It keeps the look of your picture. The prompt names only the layout and the views, so a soft picture gives a soft sheet. The backdrop is plain white with even white light, so a dark photo comes out lit. I only tested photos.
Head to toe makes up the legs and the shoes, and each run can make up something else. Name them in What the photo doesn't show to keep them the same.
Each view's own picture is cut at the gap between two views. When two views touch on the sheet, a little of one can stay in the other's own picture. The sheet keeps both whole.
To change the views or the wording, open the box Write the sheet prompt. The prompt is plain text.
Times on an RTX 5090: about 25 seconds a sheet at 16 MP once the models are loaded, about 11 at 4 MP, about 14 at 16 MP with Fast on.
LoRAs made for H3 Image go in LoRA Loader, under the Fast row.
Credits
MiniMax-H3-Image is by MiniMax and uses the MiniMax-H3-Image Community License. Read it before you use the pictures for work. Powered by MiniMax-H3-Image. The files are Comfy-Org's repackage. The 2-step LoRA is akhaliq's pDMD.
v1.1 (2026-11-20): first release here, the same file as v1.1 on Discord.
Questions or bugs: GitHub issues or the comments here. Help and more workflows on Discord. Videos on YouTube @ausbossdev. More from me on X @Zanzibased and GitHub.
