Uploaded September 2026 | Updated November 2026, 3 days ago
I tested Ming Image 0.1 in ComfyUI to see what this new design model is really good at. It made a clean phone ad and an Instagram screen with readable text. But when I asked for a realistic photo, the skin looked like plastic.
Ming Image 0.1 comes from inclusionAI. It is made for design work, not for photos: UI screens, dashboards, posters, infographics, and text inside the image. It can also make cutouts with a transparent background (RGBA output).
What you will see:
- The model files and their sizes. The main model is about 12.3 GB (BF16) or 6.18 GB (INT8). The text encoder is about 34.6 GB (BF16) or 17.7 GB (INT8). These are file sizes, not the VRAM you need.
- The error I got and the fix that worked on my system: update ComfyUI, then start it with --disable-dynamic-vram, or use the PowerShell method. Your setup may not need this.
- UI tests: a music app, an Instagram screen, and a phone ad
- My 5-step prompt method that fixed my broken icons
- How I use Qwen 3.5 9B as a prompt enhancer, with or without a reference image
- The one phrase that gives you a transparent background
- Where it fails: realistic photos, storyboards, and character-sheet consistency
My 5-step prompt method:
1. Say what to make, the viewpoint, and the style.
2. List the sections and how they connect.
3. Write the exact text you want to see.
4. Describe the important icons and their shapes.
5. Say what you do not want, like extra words or extra icons.
My verdict: I would use Ming Image 0.1 for UI screens and ad designs, especially with a good prompt or a prompt enhancer. For realistic photos, storyboards, and character sheets, I would use another model for now. In my opinion, Qwen Image 2.1 is much better for realistic images.
Resources
Workflow: patreon.com/ComfyuiWorkflowBlog/posts/ming-image-0-1-170445299
Ming-Image-0.1-Design (inclusionAI): huggingface.co/inclusionAI/Ming-Image-0.1-Design
Ming-Image-0.1-Design-Layer: huggingface.co/inclusionAI/Ming-Image-0.1-Design-Layer
Model files for ComfyUI: huggingface.co/Comfy-Org/Ming-Image
Official GitHub: github.com/inclusionAI/Ming-Image
#ComfyUI #MingImage #AIDesign
I tested Ming Image 0.1 in ComfyUI to see what this new design model is really good at. It made a clean phone ad and an Instagram screen with readable text. But when I asked for a realistic photo, the skin looked like plastic.
Ming Image 0.1 comes from inclusionAI. It is made for design work, not for photos: UI screens, dashboards, posters, infographics, and text inside the image. It can also make cutouts with a transparent background (RGBA output).
What you will see:
- The model files and their sizes. The main model is about 12.3 GB (BF16) or 6.18 GB (INT8). The text encoder is about 34.6 GB (BF16) or 17.7 GB (INT8). These are file sizes, not the VRAM you need.
- The error I got and the fix that worked on my system: update ComfyUI, then start it with --disable-dynamic-vram, or use the PowerShell method. Your setup may not need this.
- UI tests: a music app, an Instagram screen, and a phone ad
- My 5-step prompt method that fixed my broken icons
- How I use Qwen 3.5 9B as a prompt enhancer, with or without a reference image
- The one phrase that gives you a transparent background
- Where it fails: realistic photos, storyboards, and character-sheet consistency
My 5-step prompt method:
1. Say what to make, the viewpoint, and the style.
2. List the sections and how they connect.
3. Write the exact text you want to see.
4. Describe the important icons and their shapes.
5. Say what you do not want, like extra words or extra icons.
My verdict: I would use Ming Image 0.1 for UI screens and ad designs, especially with a good prompt or a prompt enhancer. For realistic photos, storyboards, and character sheets, I would use another model for now. In my opinion, Qwen Image 2.1 is much better for realistic images.
Resources
Workflow: patreon.com/ComfyuiWorkflowBlog/posts/ming-image-0-1-170445299
Ming-Image-0.1-Design (inclusionAI): huggingface.co/inclusionAI/Ming-Image-0.1-Design
Ming-Image-0.1-Design-Layer: huggingface.co/inclusionAI/Ming-Image-0.1-Design-Layer
Model files for ComfyUI: huggingface.co/Comfy-Org/Ming-Image
Official GitHub: github.com/inclusionAI/Ming-Image
#ComfyUI #MingImage #AIDesign










