Make an AI-generated image editable in Photoshop
Split a ChatGPT, Midjourney or Gemini image into Photoshop layers instead of regenerating it, up to 16 elements per pass.
- per split
- 56–138 s
- per layer (Flash), from
- $0.071
- this page’s sample
- 13 credits
Layers
Resolution
Auto finds every element (up to 16 per pass). Billed per output: at most 17 credits for 17 images. Takes about 56–138 s.
New here?
13 images · 138 s · 13 credits
Background
Full moon
Unfolded old parchment map
Rolled parchment scroll
Bottom ancient book
Brass pocket watch and metal trinkets
Middle ancient book
Pink potion vial
Quill pen with metal cap
Upload an image on the left to split your own.
Short answer
Images from ChatGPT, Midjourney, Gemini or Seedream come out as one flat picture, so fixing one part usually means regenerating the whole thing. LayerPSD splits the finished image into Photoshop layers instead: every element on its own transparent layer (Auto splits up to 16 elements per pass and runs extra passes on busy images), the background filled in behind it, every line of regular-size text as editable Photoshop type (99% of 206 lines in our 30-design test). The illustration on this page was made with GPT Image 2 and split in 138 seconds for 13 credits. Free to try: sign in with Google (a Gmail account) and your first split, background + 4 layers on Flash at 1K with PSD export, or one text edit costs nothing. No card, no subscription.
At a glance
What LayerPSD does, in one table
The same engine runs every page on this site. These are the facts to compare against other tools.
Layers
Every element on its own layer
Auto (up to 16 per pass, extra passes for busy images) or a fixed 3, 4, 6 or 8
Text
Every line of regular-size text is read before the split and lands on its own layer as native Photoshop type you can double-click and retype; stroke, glow, shadow and gradient become layer styles. 99% of 206 lines in our 30-design test, with 99.9% of characters right
Background behind objects
Filled in (amodal inpainting), so moving or hiding a layer leaves no hole
Change words, keep the font
The model redraws the new words in the same lettering, including custom and hand-drawn type
Speed and size
56–138 s per split
input up to 50 MB; output 1K, 1.5K or 2K
Privacy
The PSD is built in your browser.
Images go through our gateway (gen.yingtu.ai, which keeps no images) to the LaoZhang API and the model; LayerPSD stores nothing. Do not upload confidential work
Facts checked 2026-10-10 against the published sample runs and our credit prices. benchmarks.json
How it works
Three steps, one image
- 1
Upload the image
Drop a JPG, PNG or WebP (up to 50 MB), or paste it with Ctrl + V. Nothing to install.
- 2
Split into layers
Seedream 5.0 finds every element, cuts each onto a transparent layer and fills in the background behind it. About 56–138 seconds.
- 3
Edit in Photoshop
Export a layered PSD and move, resize, recolor or swap any element without regenerating the whole image.
What you get
- Your exact image, in layers, instead of a new generation that changes everything
- The background filled in behind each element
- Text layers you can retype in the same lettering, or export as editable type
- A PSD for Photoshop, Photopea, GIMP or Affinity
What it is not (and what to use instead)
Generating the image in the first place
Use a text-to-image model, or change a picture you already have with a prompt on our image to image page. The split starts from a finished image. Image to image AI
Upscaling or restoring the image
Use an upscaler. Splitting keeps the resolution you choose (1K–2K).
Removing AI watermarks or provenance marks
Not a use LayerPSD supports. Leave provenance marks in place.
By hand vs. automatic
Doing it by hand in Photoshop, and doing it here
You can do all of this in Photoshop yourself. This is what each step takes by hand, and what LayerPSD does instead.
Step
In Photoshop, by hand
In LayerPSD
Select each element
In Photoshop, by handObject Selection Tool, then Select and Mask for hair and soft edges, for every object
All elements found in one pass
Fill what was behind it
In Photoshop, by handGenerative Fill or Content-Aware Fill on each hole, then clean up
Background repainted automatically
Make text editable
In Photoshop, by handMatch Font, then retype each line and rebuild stroke and shadow
Export native type layers with effects as layer styles
Time for a 10-element poster
In Photoshop, by handTypically 30–60 minutes
About 1–2 minutes, 11 credits on Flash
Under the hood
How the split works
01
Layer decomposition
The image goes to Seedream 5.0 with layer decomposition on. The model detects each element (instance segmentation), returns it as its own RGBA image with a bounding box and a name, and returns a background with the covered areas repainted (amodal inpainting).
02
Editable text
Before the split, a vision model reads every line of regular-size text: the words, size, color, stroke, glow, shadow and gradient. Each line is cut out onto its own layer and written into the PSD as native Photoshop type with those effects as layer styles. Headlines taller than 8% of the image stay in the picture so the model can repaint behind them; “PSD with editable text” reads them at export. Scores: layerpsd.com/data/text-benchmark.json.
03
Retyping text
To change words, the model redraws only that text in the same lettering and leaves the rest of the image alone, so it also works for custom or hand-drawn type that no installed font matches.
Questions
Short answers on splitting, export and billing. Anything else, email us and a person replies.
How do I split a Midjourney or ChatGPT image into layers?
The models only return one flat image. Upload that image here and press Split into layers to get a layer per element with the background filled in, then export a PSD.
Can I do it in Photoshop by hand?
Yes: select each object, mask it onto its own layer, fill the hole with Generative Fill, repeat. For a 10-element image that is typically 30–60 minutes; here it is one pass of about 1–2 minutes.
Will it change my image?
The elements are cut out as they are. The only new pixels are the filled-in areas behind them, which stay hidden until you move something.
Can I fix the text the AI got wrong?
Yes. Retype a text layer and the model redraws the new words in the same lettering, or use Edit text in image on the flat picture.