Load one photo of a person, and Qwen Image 2.1 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. Switches add a row of head close-ups under the sheet, or four expressions, make it faster, make it twice as big with SeedVR2, redraw each view for more detail, save each view as its own picture, or put your photo beside the sheet. If your photo leaves something out, like the legs in a half-body photo, a box lets you name it, and the views show it.

It's the Qwen Image 2.1 version of my MiniMax H3 Character Sheet: much faster and smaller, about 7 seconds a sheet.

Two versions

• v1.1: the regular Qwen Image 2.1 model, 25 steps, or 6 with Fast on.

• v1.1 Turbo: the official Qwen Image 2.1 Turbo model in its own 8 steps. Everything else is the same, with no Fast switch.

How to use it

1. Load your picture: one person, clearly in view.

2. Optional: type a background, like "a neon city street at night". Left empty, it is plain white.

3. Optional: name what your photo doesn't show, like "blue jeans and white sneakers" for a half-body photo.

4. Run. The first run takes longer, while about 17 GB of models load.

What you get

• One picture, 1344 × 768: four views in a row, each 336 × 768.

• The views are framed like your photo. A half-body photo gives half-body views, a full-body photo full-body views.

• To see more than your photo shows, name it in What the photo doesn't show. "Blue jeans and white sneakers" turned a chest-up photo into head-to-toe views in that outfit.

• The back is Qwen'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 1344 × 1104.

The switches

• Fast (regular version only): the Viggle Turbo LoRA with its own 6 steps instead of 25, about twice as fast. The sheets held up in my tests.

• Upscale 2x: SeedVR2 makes the sheet twice as big on each side, 2688 × 1536, and sharpens it.

• Redraw each view: Qwen redraws each of the four views once more at twice the size, so faces, hair and clothes come out sharper. About 40 seconds more, about 10 on Turbo. The sheet comes out 2688 × 1536, as with Upscale 2x, which gives it the cleanest start.

• Save each view too: the four views are also saved as pictures of their own, 336 × 768 each (672 × 1536 with Upscale 2x or Redraw each view).

• 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: Qwen runs a second time, from your photo and the finished sheet, and draws the four close-ups. About 10 seconds more. Fast works on them too. With Upscale 2x on they are made twice as big as well, so the picture is 2688 × 2208.

• Expressions instead of head turns: the four close-ups face the camera and show the expressions in the four boxes under the switches (calm, a big smile, surprised, annoyed to start with). Type your own. It needs Close-ups under the sheet on too.

What you need

• ComfyUI 0.38 or newer.

• AusBoss nodes 2.24.0 or newer. In ComfyUI-Manager, search "AusBoss". Already have them? Update first.

• The models:

  • qwen_image_2.1_int8_convrot.safetensors → models/diffusion_models · 7.26 GB (the regular version)

  • qwen_image_2.1_turbo_int8_convrot.safetensors → models/diffusion_models · 7.26 GB (the Turbo version)

  • qwen3vl_8b_int8_convrot.safetensors → models/text_encoders · 9.35 GB

  • qwen_image_2.1_vae_bf16.safetensors → models/vae · 676 MB

• Only for Fast: Qwen-Image-2.1-viggle-turbo-v0.3-6step-lora-r128.safetensors → models/loras · 680 MB

• Only for Upscale 2x, SeedVR2:

  • seedvr2_3b_int8_convrot.safetensors → models/diffusion_models · 3.46 GB

  • seedvr2_ema_vae_fp16.safetensors → models/vae · 501 MB

• Optional, an uncensored text encoder (a community-made "Heretic" file; the regular one stays picked, see the second note in the workflow): qwen3-vl-8b-heretic-1.3.0_fp8_e4m3fn.safetensors → models/text_encoders · 10.02 GB

Tips

• A view came out wrong? Run again. The seed changes after every run.

• One person, clearly in view, works best. If the person is small in your photo, crop to them with the blue handles first.

• Each view's own picture is an exact quarter of the sheet, so a wide sleeve at the edge of a view can get cut off there. The sheet keeps it whole.

• Going from half-body to full body can lose small things, like a shirt under a jacket. Name those in What the photo doesn't show too.

• To change the views or the wording, open the box Write the sheet prompt. The prompts are plain text.

• Times on an RTX 5090: about 7 seconds a sheet once the models are loaded, about 3 with Fast on. The close-ups add about 10 seconds, Redraw each view about 40.

• Your own LoRAs for Qwen Image 2.1 go in the LoRA Loader: under the row Fast turns on and off, or in the empty loader of the Turbo version. Turbo already runs in its own 8 steps, so leave Viggle out there.

Credits

Qwen Image 2.1 is by Qwen, under the Qwen Research License: not for commercial use without a licence from Qwen. The files are Comfy-Org's repackage. Viggle Turbo is by Viggle, under the Qwen Research License too. SeedVR2 is by ByteDance Seed. The Heretic text encoder is DreamFast's, under Apache 2.0.

v1.1 (2026-11-12): Redraw each view, and a Turbo version on its own tab.

v1.0 (2026-11-04): first release.

Questions or bugs: GitHub issues, the comments here, or Discord. Videos on YouTube @ausbossdev. More from me on X @Zanzibased and GitHub.