Reference-guided creation
AI image-to-image generation that keeps your direction
Upload a reference—or choose any finished RasterBloom image—describe what should change, and create a new version from that exact source.
Transform a reference, not your whole workflow
Use image-to-image AI for restyling, composition changes, lighting explorations, product concepts, character variations, and visual iteration. Your source stays attached to the project, and every result remains available for comparison.
Every branch starts from the selected image
When you work inside a grouped project, RasterBloom locks the next generation to the image you selected. That prevents an accidental text-to-image render from being mixed into an image-to-image sequence.
Private, practical iteration
References and results live in your private library rather than a public gallery. Select a version, adjust the prompt, preserve its aspect ratio, and continue building without recreating the project from scratch.
Image-to-image FAQ
What is image-to-image AI?
Image-to-image AI uses an uploaded or generated image as a visual reference, then applies your written direction to create a new result.
Can I continue from a generated result?
Yes. Select any finished image in a project and the next generation uses that exact image as its source.
How much does image-to-image cost?
Each reference-guided image costs 2 credits. Current packs provide up to 90, 200, or 450 image-to-image generations.