DEDUPED Hot AI repos, turned into use cases Get the Monday 5

How to generate Design concept images with Z-Image and LM Studio with Mini Max H3 Director Cut Studio

Create concept reference images with the Z-Image workflow and write prompts through a local LM Studio model.

  • Content
  • Intermediate
  • 2 hours
  • From the README
  • Any agent

Who it is forA creator or art director preparing reference frames for a video project who wants local image generation and a local text model.

The problemProducing consistent concept images and matching text prompts by hand takes time, and the two steps are hard to keep in sync.

The outcomeConcept reference images from the Design page, using the Z-Image text-to-image workflow, with prompts generated through the LM Studio OpenAI-compatible API.

What you need first

  • A Windows PC with ComfyUI installed, plus a high-VRAM machine if you want to run the local server roles described in the README
  • Completed Step 1 deployment (Studio Runtime, Qwen3-TTS and VoxCPM2) from the README's step1.md
  • The model files listed in the README, saved under the exact file names it gives
  • Model and source-footage licenses checked by you; the README says weights are not included in the source code

Steps

  1. Place z_image_turbo_bf16.safetensors, qwen_3_4b.safetensors and ae.safetensors in the ComfyUI folders named in the README.
  2. Load a text model in LM Studio and start its local server at http://127.0.0.1:1234/v1.
  3. Start the studio with .\run_h3_prompt_studio.bat and open the Design page.
  4. Enter the LM Studio Base URL and pick a Model ID from the live list returned by /v1/models, then click TEST CONNECTION.
  5. Generate the concept images and review the captions the local BLIP model creates for them.

Prompt for your agent

Using the repo karuvanan/MiniMax-H3-Director-Cut-Studio, configure the Design page to use the LM Studio Base URL http://127.0.0.1:1234/v1 and a Model ID that I select from the live list, then generate one 1024 by 576 concept image for <scene description>. Expected output: one Z-Image concept image and its caption. Verify that TEST CONNECTION succeeds before generating.

Caveat. The README says the current Z-Image workflow is fixed to the qwen_3_4b and ae files, so a different diffusion checkpoint must be compatible with that text encoder and VAE or it will fail during sampling or decoding.

Share: 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