The input does two things at once: it supplies content and it supplies structure. How much of that structure survives is the whole game, and it is governed by a strength setting — low values nudge the picture, high values keep only the rough layout and repaint the rest.
This is the mode behind most "make it look like" work: turn a phone photo into a studio shot, restyle a frame into illustration, produce six variants of one hero image that still read as the same composition. It is also the mode behind fixes, because a small strength value is a targeted correction rather than a fresh roll of the dice.
Masked variants of the same idea get their own names — inpainting when you protect everything except a region, outpainting when you push past the original edges.
In practice
- Strength is the dial that matters; prompt wording barely registers if strength is near zero.
- Structure survives better than texture — layout and pose persist long after surface detail is gone.
- Running the output back in as the next input compounds drift quickly.
Image-to-image models
Catalog entries that take a picture in and hand a changed picture back. 30 of the 296 models in the Versely catalog qualify.
| Model | Provider | Type |
|---|---|---|
| Mai Image 2.5 Edit | Microsoft | Image |
| Nano Banana 2 | Image | |
| HunyuanImage 3.0 Instruct Edit | Hunyuan | Image |
| Kling Image O1 | Kling | Image |
| Qwen Image Edit 2511 | Qwen | Image |
| Flux 2 Klein 4B Base Edit | Flux | Image |
| Runway Gen4 Image | Runway | Image |
| Qwen Image 2 Edit | Qwen | Image |
Browse all 10 spec pages for full settings, resolutions and credit costs.
The mistake to avoid
Expecting an identity to survive a high strength value. Past roughly the halfway mark the face is being redrawn, not edited, and it will no longer be the same person.
Where you will run into it
- Text to Image Generator — One prompt. Every image model. One studio.
Related terms
Denoising strength
Denoising strength decides how much of your input picture gets thrown away before regeneration — low keeps it nearly intact, high keeps only the general shape.
Inpainting
Inpainting regenerates a region you have masked while leaving the rest of the picture untouched, so a change stays local.
Outpainting
Outpainting extends an image beyond its original borders, generating new content that continues the scene outward.
Text-to-image
Text-to-image renders a still picture from a written description, with no picture going in.
Reference image
A reference image is a picture supplied alongside the prompt so the model can copy an identity, product or style from it, without that picture becoming a frame of the output.
The all-in-one AI studio for creators. 60+ models for video, image, voice, music and lipsync in a single app.