The short version

A .psd holds two things: the editable document, which is layers and masks and live type, and a flattened picture of the result, saved so that other programs have something to show. Nothing outside Photoshop redraws the first. Everything reads the second.

This reads the second, shows it to you, and puts it on a page at the size and resolution you choose. CMYK, transparency, indexed colour and 16 bit files all work, and the ones that cannot be converted honestly are refused with a reason.

Photoshop's own format is not really an image format. It is a document format that happens to contain an image, in the same way a word processor file is not a page of text. That distinction explains everything that goes wrong with PSD converters.

What a Photoshop file actually holds

Five blocks, one pictureWhich part of the file is the image
Inside the .psdHeadersize, channels, colour modeColour mode datathe palette, if indexedImage resourcesguides, paths, thumbnailLayers and masksthe editable documentImage datathe flattened pictureOnly the last block is readReproducing layers meansblend modes, masks, smartobjects, effects and livetype: a painting program.The flattened picture iswhat every other programalready shows you.
The layers block is the document you edit and it is by far the largest part of most files. The image data block at the end is the flattened result, and it is what a file manager thumbnail, a placed image in a layout program and this converter all read.
In the image data blockWhat is read, and how
  • One channel at a time, not pixel by pixelplanar
  • Scanlines packed with PackBitscompressed
  • Greyscale, indexed, RGB or CMYKcolour mode
  • 8 or 16 bits a channel, or 1 for bitmapdepth

None of that is exotic, and all of it is documented. What makes a PSD reader fiddly rather than hard is that each of those choices is independent: a 16 bit CMYK file with an alpha channel and PackBits compression is a perfectly ordinary thing to be handed, and it exercises four separate code paths at once.

The setting that decides whether any of this works

Photoshop writes the flattened picture only when Maximise Compatibility is on. It can be turned off, it makes files noticeably smaller, and a good many people have turned it off years ago and forgotten.

What that looks like from the outside
A file with no composite in it shows a blank thumbnail in the file manager, refuses to place in InDesign, and makes every converter say something unhelpful about a corrupt file. Nothing is wrong with it. It simply has no picture in it for anything but Photoshop to read.
1

Open the file in Photoshop

Any version. Nothing needs changing in the document itself.

2

Preferences, then File Handling

On a Mac it is under the Photoshop menu; on Windows it is under Edit.

3

Maximise PSD and PSB File Compatibility, set to Always

Rather than Ask, which is the default and which people click through.

4

Save the file again

It gets bigger, by roughly the size of the flattened image, and every other program can read it.

If Photoshop is not available, anything that can open the file and export can produce a PNG or a TIFF instead, and Images to PDF takes either.

Colour, and the file that arrives as a negative

A CMYK Photoshop file stores its channels inverted: 255 means no ink rather than full ink, because that is how printing thinks. Read it the way you would read RGB and every file comes out as a photographic negative.

The same four bytesWhy CMYK has to be read upside down
Read inverted, as storedC 255 M 255 Y 255 K 255reads aswhite paperRead the obvious wayC 255 M 255 Y 255 K 255reads assolid black
A pixel with all four channels at 255 is the whitest thing in the document, because no ink is being laid down. The conversion from there to RGB is a simple multiplication, and it is done without a colour profile, which is why the result is close rather than exact.

The honest position on colour is worth stating plainly. A proper CMYK to RGB conversion needs the source profile, the destination profile and a rendering intent, and none of those are available to a browser. What happens here is the standard arithmetic conversion, which is what most software does when it has nothing better, and it is perfectly good for a proof, a share, or a document going on screen. For anything going to press, hand the printer the original.

Modes
Six

RGB, CMYK, greyscale, indexed, duotone, bitmap.

Depth
1, 8 or 16

Sixteen is read at eight, which is what a page carries.

Refused
With a reason

Lab and 32 bit, rather than a guess about colour.

Transparency, and the choice it forces

If the file has an alpha channel there are only two honest answers and you should get to pick.

Keep it

The image goes onto the page as PNG, which is lossless and carries alpha.

Transparent areas show whatever is behind the image in the PDF, which is usually nothing.

Much larger: a photograph as PNG can be ten times the size.

Right when the image will be placed over something else.

Flatten it

Transparent areas are composited onto white, which is what paper does.

The image goes on as JPEG, which is a fraction of the size.

What you want for almost anything that will be printed or emailed.

The default, because most people converting to PDF are heading for paper.

The checkbox is only enabled when the file actually has an alpha channel, so it never offers a choice that would do nothing.

What this is not

  • Not a layer renderer. It does not composite layers, apply blend modes, run layer effects or set live type. Those are Photoshop, and a browser tool claiming otherwise is claiming too much.
  • Not a print production tool. No colour management, no spot colours, no overprint, no bleed or trim marks. A press-ready PDF comes out of Photoshop or InDesign with a profile attached.
  • Not an editor. The picture comes out as it was saved. If a layer was hidden, it stays hidden.
  • Not a way around a missing composite. If the flattened image is not in the file, no tool can invent it, and the ones that appear to are rendering the thumbnail.

When this is the right tool

Sending a design to somebody who does not have Photoshop, and should not need it. Getting a comp in front of a client in a format they can annotate. Putting a mock-up into a document. Printing a layout at a known size. Archiving a finished piece in something that will still open in twenty years.

For several designs in one document, convert each and then merge them. To bring the file size down afterwards, the compressor works well on this kind of page, because a design is mostly photograph.

Why this runs in your browser

Design files are client work, unreleased product, brand assets and things under an agreement not to share. Uploading one to a free converter is a poor idea, and the usual justification for it, that this kind of decoding needs a server, has not been true for years.

The file is read in memory, the picture is drawn, the PDF is written and checked by reopening it before it is offered to you. Nothing is transmitted and nothing is stored.

The tool is right above

Drop a .psd, look at the picture, and put it on a page.

Jump to the tool

Common questions

The setting that stops it working, colour, and what it deliberately does not do.

01It says there is no flattened image in my file. What now?

Photoshop only writes the flattened picture when Maximise Compatibility is on, and it can be turned off to save space. Open the file, go to Preferences then File Handling, set Maximise PSD and PSB File Compatibility to Always, and save it again. The file gets bigger and every other program on earth can suddenly open it.

02Does it use my layers?

No, and nothing outside Photoshop really does. It reads the flattened composite, which is the same picture a file manager thumbnail and a page layout program show. If a layer was hidden when the file was last saved, it is not in the composite and will not appear.

03What happens to a CMYK file?

It is converted to RGB without a colour profile, which is the honest limit of what a browser can do. The result is close rather than exact, and saturated colours are where you would notice. For anything going to a printer, give them the original and let their workflow handle the conversion.

04Is transparency kept?

If the file has an alpha channel you can choose. Keeping it means the image goes onto the page as PNG, which is lossless and much larger. Not keeping it composites the transparent parts onto white, which is what would happen on paper anyway, and the image goes on as JPEG.

05Can it open a .psb?

Yes. A .psb is the large document format, which is the same layout with a different version number and 64-bit lengths, and it is read the same way. What will stop you is memory rather than the format: a 2 gigabyte file is not something a browser tab can hold.

06What resolution does the PDF come out at?

Whatever you pick. Fitting the page to the image lets you choose 72, 150 or 300 dots per inch, which decides how large the page is in millimetres for a given number of pixels. Choosing a paper instead fits the image onto that sheet and the resolution follows from the size.

07Why is my file refused for being in Lab or 32 bit?

Lab needs a colour engine to convert and 32 bit is a floating point format for high dynamic range work. Both can be converted here only by guessing, and a guess about colour is worse than a refusal. Save as 8 or 16 bit RGB or CMYK in Photoshop and it will go through.

08Is my file uploaded?

No. The .psd is read and the PDF built in your browser. Design files are among the last things anybody should be uploading to a converter, and there is no reason to.