How to Remove Objects From Photos With Cleanup.pictures
Cleanup.pictures removes an object from a photo after you paint over it with a brush. No layers, no clone stamp, no selection math. You drag a mask over a bin or a photobomber, the model fills the hole from surrounding texture, and you compare before and after with a slider.
For a product photo or a real-estate listing, this saves a Photoshop session. For anything with a strong repeating pattern or a face you want gone, it needs a second and third pass.
What kind of photo works best?
Textured, evenly lit images with a small object. A bin on a paved courtyard, a cable against a wall, a signpost against sky. The model samples nearby pixels, so the surroundings have to look like something.
Weak cases: an object covering most of the frame, straight architectural lines running through the object, and reflections on glass. Those need manual retouching or an inpainting tool with structure control.
How do you remove an object in three steps?
- Upload the image and choose the brush size, starting a little larger than you think.
- Paint over the object including its shadow. Shadow leftovers are the number one give-away.
- Wait for the fill, then use the slider to inspect edges at 100% zoom.
Repeat with a smaller brush on any leftover fragments. Two passes usually look cleaner than one heavy pass, because a wide mask lets the model invent texture where it should have copied it.
Why do edges look smudged after removal?
Because the fill is generated from the pixels around the mask, and at the mask boundary those pixels blend. On hard edges, like a railing against a wall, the blend reads as blur.
Freeze the result before you judge it: squint. If the removal reads as a soft grey patch when you squint, the edges need another pass.
Practical trick: paint slightly inside the object boundary on the final pass, then paint the remaining few pixels individually. Fewer generated pixels, tighter edge.
What are the file and resolution limits?
Free web use caps output resolution and adds a watermark on some flows; the paid plan removes the cap and supports larger images. On the phone app the same edit works offline-ish via the cloud and is fine for social crops but not for print.
For a print job at 300 DPI, plan on a larger source than the web tool accepts and budget a real retoucher for the final file.
| Use case | Fits Cleanup.pictures? |
|---|---|
| Remove bin from listing photo | Yes |
| Remove tourist from landmark shot | Yes, if small |
| Extend canvas beyond frame | No, use outpainting |
| Fix a scratch across a face | No, use healing brush |
| Batch 500 product shots | Slow, needs a script or another tool |
When should you use something else?
If you need to add content rather than remove it, an outpainting or generative fill tool does that job. For batch work on hundreds of catalog images, Clipchamp is not the answer either, but a batch-capable photo editor is: this is a one-at-a-time tool by design.
Who should skip it?
Professional retouchers who need channel-level control, anyone editing faces for a legal or journalistic context, and anyone needing strict reproduction accuracy for a product where the removal changes what is being sold.
Quick Summary
- Cleanup.pictures removes objects by brushing a mask, then fills from surrounding pixels.
- Works on small objects against textured backgrounds; struggles with strong lines and glass.
- Two light passes beat one heavy mask.
- Free tier caps resolution; paid plan lifts it.
- Not a batch tool, and not for generative extension of the frame.
