Soku AIUpload up to four images
One image to edit, or several when the edit needs a reference — the product you want inserted, the palette you want matched, the style you want it to resemble.
All tools
AI Image Editor Say What Should Change
Most ad-image work is not creating something new, it is changing one thing about something that already exists: a different colourway, a cleaner background, the product a third larger. Upload the image, describe the change in a sentence, and get the edited version back — no masks, no selection tools, no layers.Most ad-image work is not creating something new, it is changing one thing about something that already exists: a different colourway, a cleaner background, the product a third larger.
Image EditingPrompt-Based
Three deliberate steps
Describe the change in one sentence
Name what should be different and leave the rest alone. Short, specific instructions land far better than a full re-description of the scene.
Take the result into Creatives
The edited image hands into the Creatives canvas, so the same asset can be sized and shipped to Meta, Google and TikTok without a round trip through a desktop editor.
Use cases
Who reaches for this, and when
Produce a colourway per audience
One product photo becomes the whole range. Ask for each colour by name and test which one a given audience responds to, without a reshoot per variant.
Clean up usable-but-not-quite footage
A shot that is right except for the cluttered background, the wrong season, or a distracting object is normally binned. Describing the fix is faster than masking it.
Bring stock imagery on-brand
Licensed stock rarely matches a brand palette. Re-render it with your colours and lighting so it stops looking like stock.
Questions before you run it
- How is this different from a normal image editor?
- A conventional editor asks you to select the region you want to change and then operate on it. Here you describe the change in words and the model works out what to alter. That is faster for content changes and worse for precise, pixel-exact ones.
- Does it edit only the part I mentioned?
- Not literally. The model re-renders the whole image with your instruction applied, so untouched areas can shift slightly in grain, lighting or fine detail. For most ad creative that is invisible; if you need the rest of the frame byte-identical, this is the wrong tool.
- Can I use it to remove an object?
- You can, and describing what to remove often works. For a clean removal that leaves the rest of the frame untouched, the dedicated AI Object Remover uses an inpainting model built for exactly that and is the better choice.
- Why would I upload more than one image?
- To give the edit something to copy. Uploading your product alongside a scene lets you ask for the product to be placed into it; uploading a brand image alongside a stock photo lets you ask for the palette to be matched.
- Is it free?
- Yes — create your account on this page and your free credits cover your first edits. Each edit is a flat credit charge regardless of how large the instruction is.
