Modelverse Flux Kontext Pro text2image

Some nodes are for flagship moments; this one is for Tuesday. Flux Kontext Pro text2image is the text-to-image face of Kontext Pro - same prompt-and-aspect-ratio simplicity as the Max T2I node, but on the cheaper, more conservative model tier. In a pack full of video heavyweights and flagship editors, this is the node you'll actually have in the default workflow: the everyday image generator that won't make you check the console after every run.

What it is

Pure text-to-image. Prompt in, aspect ratio, batch knobs, one IMAGE out. No image inputs, no editing mode, no surprises. It's the "generation" companion to the Pro editing node - and because the interface matches the Max T2I node field-for-field, you can swap between Pro and Max to A/B quality against cost without rebuilding anything.

The inputs that matter

• client - required, from the Modelverse Client / Secret Client node.
• prompt - the whole job. Kontext Pro follows descriptive prompts well; give it scene, subject, and style.
• aspect_ratio - 21:9 to 9:21, default 1:1. Your only composition control, so set it deliberately.
• num_images (1–4) and num_requests (1–10) - batch knobs and bill multipliers.
• guidance_scale - 2.5 default (0–10). Gentle is the house style here.
• seed - -1 for random; set it to nail down a keeper.

Output is a single image (IMAGE tensor) for preview or save.

Installing and keying up

One plugin for the whole pack. ComfyUI Manager → search "ComfyUl-UCloud" → Install → restart, or:

cd ComfyUI/custom_nodes
git clone https://github.com/ucloud/comfyui-plugin

restart ComfyUI. Dependencies are light (requests, pillow, numpy, openai, GitPython); no weights to download since the model runs server-side. Setup is a ModelVerse API key from the UCloud console - via the Modelverse Secret Client node (name in workflow, key in secrets.json) or a config.ini MODELVERSE_API_KEY=... entry.

Common issues

• Output is more conservative than Max - that's the Pro tier's personality: reliable and faithful over wild. If you need bold stylization, that's the Max T2I node.
• "API key is not set" - empty client node. First check, always.
• Aspect ratio surprises - no width/height fields, so the dropdown is the only shape control. Check it before you queue, not after.
• Costs still exist - Pro is cheaper than Max but not free. num_requests × num_images multiplies; iterate at 1×1.
• "Meh" results on thin prompts - one-liner prompts give one-liner images. Add style and mood words.

Treat it as the pack's default generator: cheap enough to iterate on, good enough that you'll forgive it the occasional run. When a take needs more muscle, hand the prompt to the Max T2I node and watch the difference.