Sometimes you want a new still from a sentence. Sometimes you want to change one thing in a photo you already have — a door color, a sky, a small prop — without rebuilding the whole scene. The free Flux Kontext tool on CanvaPro covers both paths: Text to Image and Image to Image. You do not need an account, and the usual download does not pick up a watermark from the page.
Image to Image accepts a JPG, PNG, GIF, or WEBP up to 10 MB. GIF support is worth noting if your reference is already in that format. Aspect ratios are Auto, 1:1, 3:2, 2:3, 9:16, 16:9, 3:4, and 4:3. After you write a prompt, complete the security check and click Generate Image. Image to Image requires a file. Text to Image does not. A preview typically appears in about a minute.
Edits that stay anchored
For a change to an existing photo, say what should stay and what should change. “Keep the tiles, cobblestones, and plant; change only the door to deep teal” gives the model a clearer job than “make it nicer.” A sharp, straight-on still usually edits more cleanly than a tiny, busy thumbnail. Only upload a reference you have the right to edit.
When you want a looser variation of a still, not a tightly locked edit, transform the photo with image to image is often the better starting point.
How to generate or edit an image, step by step
- Choose Text to Image or Image to Image. In Image to Image, upload a JPG, PNG, GIF, or WEBP under 10 MB.
- Write a prompt that names the subject, setting, and — for an edit — the parts to keep.
- Pick an aspect ratio, complete the security check, and click Generate Image once.
- Compare the preview with your intent (and with the reference, if you used one). Download the still if the change looks right. Keep earlier files for comparison.
When Flux Kontext is a good fit
Creators use it for concept stills and for small, controlled edits to photos they already own. Designers try a color or prop change before a reshoot. Treat every result as synthetic media. A generated door color is a visualization, not proof that the building was painted that way, and it should not be used to mislead a client or a viewer.
If the edit drifts
Tighten the prompt, confirm the 10 MB limit, and try a cleaner reference. A busy queue can add a short wait. If you would rather start from a different model with a broader set of image-to-image choices, open the image to image tool, which lets you pick among several models after you upload a reference and write a transformation prompt.
Uploads and prompts travel over HTTPS. Requests are handled to complete generation and are not sold.
A sentence extracted from the still with the image to prompt tool can help you tighten the brief before you try again.
After the Flux Kontext preview appears
Look closely at the parts you asked to keep — a face, a product shape, a tile pattern. If they drifted, say so more plainly in the next prompt rather than running the same vague instruction again. Save the download with a note of the mode, prompt, and ratio.
Check the parts you wanted to keep, and if they slipped, tighten the prompt and run the Flux Kontext editor again from the original photo.
Small edits versus a whole new scene
Flux Kontext is often at its best when the change is local and clearly named: a door color, a sky, a piece of furniture, a time of day. Those edits are easier to judge against the original photo. If you ask for a complete restyle, you can still get a strong still, but you should expect the reference to loosen. Read the preview against your brief, not only as a pretty picture.
GIF references are welcome on this uploader, which helps if the still you want to edit already lives as a GIF frame or a short animation export. Convert or crop first if you only need one frame, and stay under 10 MB either way.
For a series of related edits — several colorways of the same storefront, for example — keep the same reference and change only the instructed detail. That usually looks more consistent than generating each version from a slightly different crop.
A photo you already like can also be turned into a written brief with the guide to extract a prompt from an image.
A short checklist
- Pick Text to Image, or Image to Image with a JPG, PNG, GIF, or WEBP under 10 MB.
- Write one clear prompt, and for edits name what should stay.
- Choose an aspect ratio, then click Generate Image once.
- Download the still only after you have compared it with your brief.
If you are editing a photo for a client, keep the original reference and the prompt in the same folder as the download, and say that the change was generated. A local, clearly described edit is easier to approve than a vague restyle with no record of what you asked for.
When you are ready, choose a mode, describe a single change or scene, and save the preview once it looks anchored. One careful edit is more useful than a stack of loosely related variations.