Transparent PNG layers
Each element comes back as a PNG with a clean alpha channel, ready to drop into Photoshop, Figma, Canva, a slide deck, or a game engine.
A new dimension for your images
Magic Layers is an AI tool that turns a flat image into editable layers. Get named, transparent PNGs ready for your next design.
From flat to flexible
One flat poster, six independent layers. Select a layer to see exactly where Magic Layers found it in the original image.
Layers out (6)
An illustration of real Magic Layers output. Every layer comes back named, described, and positioned.
Made for your workflow
Not a folder of anonymous crops. Every separated element arrives ready to use in the tools you already work in.
Each element comes back as a PNG with a clean alpha channel, ready to drop into Photoshop, Figma, Canva, a slide deck, or a game engine.
The AI labels every layer it finds — "Gold circle", "Headline", "Product bottle" — so you know what a file is before you open it.
Each layer keeps its detected position and stacking order. Click a layer to highlight exactly where it sits in the original image.
Describe what you want back — every person, just the clothes, the headline — and pick the target count from 2 to 15 images. You pay only for what comes back.
Three steps. More possibilities.
Upload, submit, and return for your layers.
Drop in a PNG or JPEG. Large images are resized automatically before upload. Your file stays private.
Choose a target of 2–15 images and optionally describe the split. Submit the task; after confirmation, you can leave the page. Tasks run one at a time per account.
Follow progress in My tasks. When a task completes, preview the layers, download the ones you need, or delete the task and its files.
Image layer separation is the process of taking one flat, already-rendered image and splitting it back into its individual visual elements — background, subject, text, and decorations — as separate images with transparency.
Magic Layers does this automatically. You upload a single PNG or JPEG, optionally describe what to separate — for example all the people in a photo or the clothes on a model — and get back a base image plus one transparent PNG per element, each with a name, a short description, and the position it occupied in the original. Nothing about the source file is needed: the separation is inferred from the pixels, so it works on images you no longer have the project file for.
A background remover gives you one subject on a transparent background. Magic Layers separates every major element, so you can move, replace, or restyle any part of the image.
Built for the work you do
Anywhere a finished visual needs to become editable again.
Split a finished campaign visual into background, headline, product, and badges — then test new copy and layouts without hunting for the lost source file.
Recompose one thumbnail for vertical stories, square posts, and wide video covers instead of masking the subject out by hand every time.
Pull charts, icons, labels, and shapes out of a flat slide screenshot and rebuild it as a clean, editable composition.
Turn flattened concept art and interface mockups into separate characters, panels, and props for parallax scenes and clickable prototypes.
Separate the product, price label, and background surface so you can restage a shot for a new placement or marketplace.
See how a visual is built by stepping through its layers — useful for learning composition and planning your own original version.
Room for every creative ambition
Every layer costs 2 Credits, and you are only charged for the layers you get. Choose a monthly plan or a one-time pack.
$29/month
400 Credits · up to 200 layers
For individual creators with regular layer-separation work.
$79/month
1,200 Credits · up to 600 layers
For professionals producing frequent creative variations.
$149/month
2,400 Credits · up to 1,200 layers
For studios and teams with a high monthly volume.
Layer counts, formats, credits, and what Magic Layers can and cannot recover.
Magic Layers analyses one flat PNG or JPEG and separates the major visual elements inside it — text, subjects, decorations, and background — into individual images. You get a base image plus a set of transparent PNG layers, each with a name, a description, and its position in the original.
A background remover produces a single cutout of one subject and throws the rest away. Magic Layers separates every major element it finds, so the headline, the product, the badge, and the background all come back as independent layers you can edit separately.
You choose the target count — from 2 to 15 images per job, counted as one base image plus layers. The model aims for your target and may return fewer; you are billed for every image the job actually returns.
Yes, after the upload finishes and the task submission is confirmed. Processing continues in the background. Each account runs one task at a time, in submission order, with up to 10 unfinished tasks. My tasks shows progress and results. If the next task needs more Credits, the queue pauses until you add Credits or cancel that waiting task. Running tasks cannot be cancelled; completed tasks can be downloaded or deleted.
Each output image costs 2 credits, including the base image, and starting a job needs at least 2 credits per image you select. There is no upfront hold: after the job you are billed for every image actually returned. If the model returns more than you selected, that can take your balance below zero — a negative balance must be topped up before the next job. Failed jobs are never charged.
Some images are refused by the processing provider — for example copyright-protected or adult content. A refused image fails before billing: you are not charged and your balance is unchanged. No job that fails, for any reason, is ever charged.
One PNG or JPEG per job, up to 10 MB; larger images are downscaled automatically. The image needs to be at least 512×512 and within the model's supported aspect-ratio range, so ordinary photos, posters, screenshots, and artwork all work.
They are transparent raster PNGs, not a recovered Photoshop or Figma document. You can move, resize, mask, and composite them in any editor, but fonts, vectors, and effects that only existed in the original authoring file cannot be recovered from a flat image.
New successful source images and generated layers are stored privately in R2 until you delete the task or account. You can revisit, preview, download, and delete results from My tasks. Unfinished uploads expire after seven days; older temporary result URLs may already be unavailable.
Magic Layers gives you the technical separation; it does not grant you rights to the artwork. Upload only images you own or have permission to use, and check the result before publishing it commercially.
Upload a PNG or JPEG and get editable layers back in a few minutes.
Upload an imageThe claims on this page about file formats and transparency come from the published specifications, not from our own benchmarks. The references below are where to check them.
“Indexed-color, greyscale, and truecolor images are supported, plus an optional alpha channel.”
This optional alpha channel is what lets every separated element sit on a transparent background instead of a solid rectangle.
Page last updated