Ideogram 3.0 Replace Background

Ideogram built its name on the one thing image models kept botching - text in images - and 3.0 was the closed generation that made it the benchmark. This node takes Ideogram 3.0's replace background capability and puts it in ComfyUI via Runware: you feed a photo and a prompt, and the subject stays put while the background is reimagined. If you've ever needed to drop a product onto a different backdrop without re-shooting it, this is the node.

That's the whole job and it's a genuinely useful one. It does the segmentation-and-inpaint dance for you - the model figures out what's the subject and what's the backdrop, then regenerates only the backdrop from your prompt. No mask painting, no cutout, no background-removal step. It's the closest thing in the pack to a one-click e-commerce hero shot.

The inputs that matter

• seedImage (required, IMAGE): the photo. Subject-forward, cleanly lit, whole subject in frame.
• positivePrompt (required, STRING, multiline): the new background. "Studio backdrop, soft grey gradient, subtle shadow beneath the product" - concrete prompts read best.
• providerSettings.ideogram.styleType: AUTO, GENERAL, REALISTIC, DESIGN, RENDER_3D, ANIME, FICTION. REALISTIC is the default instinct for product work; RENDER_3D when you want the CGI look.
• providerSettings.ideogram.stylePreset (63 presets): from 80S_ILLUSTRATION to BLUEPRINT and beyond. This is Ideogram's secret sauce - if you want a specific art direction, this dropdown does more than any prompt.
• providerSettings.ideogram.magicPrompt (AUTO/ON/OFF): Ideogram's prompt expansion. AUTO is fine; OFF if you want exactly what you wrote.
• providerSettings.ideogram.renderingSpeed: TURBO / DEFAULT / QUALITY. QUALITY costs more; TURBO for drafts.
• providerSettings.ideogram.styleReferenceImages (IMAGE): style guidance images, for "make it look like this."
• providerSettings.ideogram.styleCode: an 8-character hex style code, if you've saved one from Ideogram's style system.

Output is a native image (IMAGE), with the standard numberResults (up to 20), safety, ttl, outputFormat and outputQuality block.

How it works

Pack-standard: your image and prompt go up as a REST request, Ideogram 3.0's replace-background runs on Runware's GPUs, and the composited result comes back as an IMAGE. The cloud handles segmentation, so you never touch a mask. Cost per run shows on the title bar.

Install and gotchas

Install the pack once - ComfyUI Manager, search Runware - or git clone https://github.com/Runware/ComfyUI-Runware into custom_nodes + pip install -r requirements.txt, restart, and set a Runware API key (Settings → Runware API key, RUNWARE_API_KEY, or runware auth login).

The gotchas: this is a closed Ideogram API, so moderation applies and commercial-use terms are Ideogram's, not yours - check before you build a product on it. Quality tracks the source photo; a subject whose edges melt into the background will produce a subject that still melts. And it's per-generation metered, so iterate on one result at TURBO before committing to QUALITY batches. For catalog-style background swaps with zero mask work, it's hard to beat.