Qwen-Image (BudgetPixel)

Alibaba's Qwen-Image built its reputation on one thing: it renders text correctly. Where most diffusion models turn "COFFEE" into a typographic accident, Qwen-Image writes legible, correctly-spelled words - which makes it the pack's go-to for anything with signage, labels, logos or captions baked into the image. At 15 credits per image it's also one of the cheapest nodes in the pack, and it has seed control, so it doubles as a reproducible workhorse for text-heavy asset batches.

Like everything in the BudgetPixel pack, it's a hosted API call. Prompt goes to the API, the Qwen pipeline renders server-side, and the image returns as a core IMAGE tensor in your graph. No local GPU, no model files. Qwen-Image is closed-weight here - the pack is calling Alibaba's hosted endpoint, not a local checkpoint - which is exactly the "call it because you can't download it" pattern the whole API-node category lives on.

The inputs

• prompt (required) - your description. With Qwen-Image, put the text you want rendered in quotes inside the prompt; that's the convention the model responds to best.
• seed - default -1 for random, or set for reproducible results. Lock it and vary the prompt to iterate.
• aspect_ratio - nine options including the tall 9:21 and wide 21:9, plus the usual 16:9/9:16/4:3/3:2 family.
• num_images - 1–4.

Output is images into Preview Image or Save Image.

Where it sits in the pack

The Qwen family in this pack forms a clean ladder. This base Qwen-Image is the cheap text-rendering specialist at 15 credits - pure text-to-image, seed control, no editing. Qwen Image 2.0 (45 credits) adds editing and multi-image fusion with a negative prompt. Qwen Image 2.0 Pro (85 credits) is the premium tier with stronger semantic adherence. Qwen Image 3.0 (40 credits) balances speed and quality. And if you need text but care even more about cost, the P-Image node at 10 credits is Qwen-derived - but it auto-enhances prompts, which dilutes your typography control. When the text itself is the deliverable, base Qwen-Image is the best credits-to-typography deal in the pack.

Install and issues

Install via ComfyUI Manager (search "BudgetPixel") or:

cd ComfyUI/custom_nodes
git clone https://github.com/budgetpixel-ai/comfyui-budgetpixel.git
pip install -r comfyui-budgetpixel/requirements.txt

Restart, find it under Add Node → BudgetPixel → Image, and set your API key via BUDGETPIXEL_API_KEY or a budgetpixel.json file - there's no key widget, deliberately, so shared workflows don't leak secrets.

Standard pack caveats: a recent ComfyUI for core types, and interrupting a run abandons the local wait while the server-side job keeps going (still billing its 15 credits - low stakes, but it's the pack's rule). And remember: hosted Qwen means the model's filters are what they are. There's no local weights file to tweak, so anything Alibaba's endpoint refuses stays refused.