How to separate an image into layers in Photoshop

Steps

1. 1. Duplicate the background

   Open the JPG or PNG. Duplicate the locked Background so you can hide the original while you work.

2. 2. Select one object

   Use Select Subject for a clear person or product. Use the Pen tool when the edge is hair, glass, or a logo you must follow exactly.

3. 3. Layer via Copy

   Command-J or Control-J puts the selection on a new layer. Name it. Hide it and check the edge at 100 percent.

4. 4. Fill the hole

   On the background copy, select the hole and use Content-Aware Fill, or clone. The fill is the slow part, and it fails on faces and repeating patterns.

5. 5. Repeat for the next object

   Text that is still a live type layer can be edited with the Type tool. Text that was flattened in the JPEG has to be retyped. Photoshop will not recover a font that was never in the file.

6. 6. Or start from an AI split

   Upload the same flat file on image to layers, download the PSD, and open it in Photoshop. You still check edges. You skip the first mask.

The manual method

Photoshop’s own tools are the accurate way to separate a flat image when you have the time. Select Subject, the Object Selection tool, and the Pen tool all end at the same place: a selection, Layer via Copy, then a repair of the background. Do one object at a time. Name the layers. A poster with a headline, a product, and a background is three passes, not one click.

Live text is the exception. If the file is already a PSD with a type layer, do not rasterize it and do not run a splitter. Edit the type layer. The manual method on this page is for a JPG, a PNG, or a flattened PSD where the words are pixels. You retype them with the Type tool and the font you own. Content-Aware Fill does not set type.

Where the manual method gets expensive

Hair, fur, smoke, and glass take longer than the subject in the middle of a white sweep. Content-Aware Fill invents texture. On a face behind a caption it invents a face, and you will paint that by hand. A logo that touches a word becomes two painful paths. This is still the right method when the edge must match a brand guide and you can spend the hour.

There is a shorter article on this site, how to split an image into layers without Photoshop. That one stays outside Photoshop. This one is the in-Photoshop version, plus the point where you stop masking.

The AI shortcut, then Photoshop

Upload the flat image on image to layers or, if you want the file named for what it is, on JPG to PSD with layers. The 2K split is 5 credits and returns a raster PSD plus a PNG ZIP. Open that PSD in Photoshop. The layers are pixels at their original positions, with a filled background. They are not paths, not smart objects, and not the original font.

Use Photoshop for the part the split missed. A fused price and headline still need the Pen tool. A bad fill behind text still needs clone stamp. Edit text on image-layered.app can retype a line before you download, in a font you pick, not in the original file’s font. Remove text hides a line. Recolor is 3 credits. Replace is 4. PhotoCraft can do the browser version of the same PSD if you are not in Photoshop at all. Saved projects keep the stack.

The shortcut is the wrong start when you need vectors or a Live2D rig. You will get raster layers and you will still build the rig somewhere else. It is also the wrong start when you already have the layered PSD. Do not split a file that is already separated.

Questions