Should I vibe code
Advanced browser-based raster and vector editor with PSD support
Premium unlocks no features. It hides an ad. Rebuilding a decade of PSD archaeology to dodge a banner is the trade.
?
Their verdict, the Premium price and the build-time estimate come from their entry, MIT-licensed. Checked 2026-08-04.
?
Our verdict, the regret score and everything below it. Editorial and unsponsored — nobody can pay to be moved.
The honest answer
why the verdict is what it is
Start with the thing that makes this entry unusual: there is nothing to save. Photopea's free tier is the entire editor — layers, masks, PSD round-trips, the lot — funded by an advertisement in the right-hand column. The subscription does not unlock features; it hides that column and adds some storage. So the build you are contemplating is not replacing a paid product, it is reimplementing a free one to avoid a banner. And what you would be reimplementing is one of the most quietly absurd pieces of software on the web: a single developer's decade of file-format archaeology, reading PSD, AI, INDD, XCF, Sketch, XD, FIG, CDR, AFPHOTO and raw sensor data from six camera vendors, entirely inside a browser tab with nothing uploaded anywhere. A canvas with layers and a brush is a genuinely fun Saturday and you should have one. The gap between that and opening a client's layered PSD with clipping masks, smart objects and 16-bit channels and handing it back unbroken is measured in years, and it is measured in years for everyone, not just you.
What actually breaks
not "if". the specific failures.
- PSD, about ninety seconds after the first real file: reading the composited pixels is the easy fifth, and adjustment layers, layer effects, smart objects, clipping masks and 16-bit channels are the rest
- Blend modes, which are not one formula each — Photoshop's Overlay, Color Burn and group-isolation rules are where "close enough" becomes visibly, unfixably wrong
- Text layers, which need the exact font, the same metrics, the same OpenType features and the same line-breaking, or every reopened file reflows and the design is subtly not the design
- Colour, on anything from print: CMYK, spot colours and embedded ICC profiles landing on a canvas that thinks in sRGB and has no opinion about ink
- Memory, on a 200-megapixel layered document inside a tab with a hard heap ceiling and no way to ask for more
- Undo, which for a layered editor cannot be a stack of canvas snapshots — that is the implementation an agent will reach for first, and it dies at about step forty
- The long tail of formats people actually open: CDR, AFPHOTO, XD, FIG, SAI, KRI, and camera raw from Nikon, Canon, Sony, Panasonic, Fuji and Olympus
- Saving, which is where a viewer becomes an editor — writing a PSD that Photoshop reopens without complaint is a strictly harder problem than reading one
Is that you?
the verdict is a default, not a law
- You want a canvas, layers, a brush and a couple of filters, and you are building it to learn how they work
- The files are yours, the formats are PNG and JPEG, and nothing has to survive a round trip into somebody else's tool
- Everything stays client-side, which is also why this entry's risk numbers are as low as they are
- You are wrapping existing libraries for the format work rather than writing parsers from the specification
- Anyone downstream expects to reopen your output in Photoshop, Illustrator or Affinity
- The plan is to reimplement PSD reading and writing yourself as a step on the way to something else
- You are doing it to avoid a $5 subscription for a product whose free tier already does everything
- Print is involved and CMYK, spot colour or ICC profiles are part of the job
If you build it anyway
the checklist, then the prompt that enforces it
- Decide first whether you are building a viewer, an editor or a converter. They are three projects and only one of them is a weekend.
- Use an existing format library — ag-psd for PSD, an established raw decoder for camera files — and treat writing those formats as a separate, later, riskier project than reading them.
- Represent a document as layers with parameters, not as a flattened bitmap plus an undo stack. Retrofitting non-destructive editing onto a pixel buffer is a rewrite.
- Work on a downsampled proxy for the interactive view and apply operations to the full-resolution document only on export, or the tab dies on the first real photo.
- Keep everything client-side. Photopea's own strongest property is that files never leave the browser, and copying that choice removes an upload endpoint, a storage bill and a privacy policy in one go.
- Test with files that came out of Photoshop, not files your own code wrote. Round-tripping your own output proves nothing.
- Write down, in the README, which formats you support and to what depth. "Opens PSD" is a promise readers will interpret far more generously than you meant it.
I want to build a browser image editor with layers, in the general shape of
Photopea. Before any code, hold me to the scope, because this project's failure
mode is quietly promising format support I cannot deliver.
1. Make me answer one question first and put the answer in the README: is this a
viewer, an editor, or a converter? Do not start until it is written down.
2. Model the document as an ordered tree of layers with parameters — position,
opacity, blend mode, mask, effects. Never as a single flattened bitmap.
Rendering composites that tree; it does not mutate it.
3. Build undo as an operation log against that tree, not as canvas snapshots. If
you were about to snapshot the canvas, say so and explain why you are not.
4. Interactive editing runs on a downsampled proxy; full-resolution pixels are
touched only on export. Tell me the memory ceiling you assumed.
5. Everything stays in the browser. No uploads, no server-side decoding, no
cloud storage. If I ask for a share link, tell me what I have just signed up
to host and store.
6. For PSD, use ag-psd or equivalent. On open, report which features were not
honoured — adjustment layers, smart objects, effects, 16-bit channels —
rather than silently flattening them.
7. Writing PSD is a separate milestone with its own test: Photoshop reopens the
file with no warning dialog. Until that passes, the button says "export".
8. Text layers need the font to render identically. If the font is missing, say
so on the layer rather than substituting silently and reflowing my design.
9. Preserve ICC profiles, or convert to sRGB explicitly and log it. Refuse CMYK
and spot colour outright rather than approximating them.
10. Test against files produced by Photoshop and Illustrator, not by this app.
Round-tripping your own output proves nothing.
11. Out of scope: generative fill, camera raw, INDD, CDR, Sketch, XD,
collaboration, and an asset library.
12. Finally: remind me Photopea's free tier is the whole editor with an ad in
the sidebar. If I am building this to save $5 a month, say so and stop.That one keeps you out of trouble. For the prompt that actually builds it, canivibecodeit.com has one.
their build prompt ↗Or don’t build it
the boring option, and the way back out
Almost always, and the honest version of the sentence is: you do not have to buy anything. Photopea's free tier is the whole editor and it costs nothing, so "buy instead" here means "just use it, with an ad". The $5 removes that ad, adds cloud storage and lengthens the history stack, and it goes to a developer who wrote a decade of file-format compatibility that nobody else attempted. That is a tip with a feature attached, and it is a considerably better trade than your next six months.
$5/mo is cheaper than your weekend.
The nicest exit in the category, because there is no lock-in to escape: inputs are files, outputs are files, and deleting the project leaves your images exactly where they were. Keep originals in their native format and export flattened PNG or TIFF copies of anything you care about, and a homemade editor's disappearance costs you nothing but the tool. The only way to create an exit problem here is to invent your own document format and store work in it — so if you do, write the exporter in the same week.
Browser-based layered image editor in plain JavaScript — the closest open equivalent, and a good read for how much of this fits in one repo.
Maintained JavaScript library for reading and writing PSD files, which is the single hardest dependency you would otherwise write yourself.
Open-source browser design tool covering the vector half of the job rather than the raster half.
Questions
If the free tier does everything, why is this entry here at all?
Because canivibecodeit lists it, and because the question it raises is genuinely interesting: what does "should I build this" mean when the incumbent is free? The answer turns out to be the cleanest version of the argument this site makes. There is no subscription to escape, no vendor to be locked into, no data being held hostage. All that is left is the engineering — and once the money is off the table, six months of PSD compatibility is very obviously not worth it.
Is Photopea open source? There is a repo with its name on it.
No. github.com/photopea/photopea exists but its README says plainly that Photopea is not fully open-source and the repository is for bug reports, feature requests and discussion. It is a useful read for the supported-format list, which is the best available summary of what you would be signing up to reimplement, but there is no source to start from.
Why is buildEase 3.2 when the upstream estimate says one sitting?
Because the two answers are about different things, which is why we are marked harsher here. A canvas with layers, a brush and a few filters really is one sitting, and that build is worth doing. The product in this entry is the format compatibility, and there is no sitting length that gets you a PSD with smart objects and clipping masks reopening cleanly in Photoshop. We scored the product, not the demo.
Every week, someone ships something they shouldn’t have.
New verdicts, the worst thing that landed in the trap, and the occasional incident report. No other email, ever.
Every operation is a library call. The product is that it runs in a browser tab on a 40MP file without dying.
Your version exports PNGs. The product is a document you can reopen next year and reshape into thirty formats.
The editor is hard, the template library is harder, and the font licensing is a legal department.
last reviewed 2026-08-05 · verdict is editorial and unsponsored · shared entry data from canivibecodeit under MIT · not legal advice