Guide
How to Compress a PNG on Windows (Without Losing Quality)
A PNG can be made considerably smaller without discarding a single pixel. PNG is a lossless format, so compressing one is a matter of storing the same image more efficiently — not of throwing anything away. The result is byte-for-byte identical when decoded, and often 40–70% smaller on screenshots and graphics.
The short answer
- Open Resize & Compress in PhotoTray.
- Drop in a PNG, a selection, or a whole folder.
- Choose lossless optimisation.
- Pick where the results go. The originals are left alone unless you say otherwise.
Nothing is uploaded, there is no file-count limit, and there is no watermark on the output.
Why PNG files get so big
PNG stores every pixel exactly as it is. That is the whole point of the format — it was designed so that an image could be opened, edited and re-saved any number of times without degrading. The cost is size.
A photograph is the worst case. Photographic detail is close to random from a compressor's point of view, so there is very little redundancy to exploit. A 12-megapixel photo saved as PNG will often exceed 20 MB, where a JPG of the same photo at a quality nobody could distinguish would be under 2 MB.
Screenshots, diagrams, logos and UI exports are the opposite case. They are full of large flat areas of identical colour, long runs of the same pixel, and a limited palette. That is exactly what PNG's compression is good at — and exactly what a poorly-written PNG encoder fails to take advantage of.
What "lossless compression" actually does
Three things, none of which touch the image:
- Better filtering. PNG applies a per-row filter before compressing, and choosing a better filter for each row makes the data more compressible. The pixels are unchanged; only the intermediate representation differs.
- A smaller palette. If an image only uses 200 distinct colours, storing it as a 256-colour indexed image instead of full 24-bit RGB is exact and much smaller. Most encoders never check.
- Dropping metadata and chunks. Colour profiles, text chunks, timestamps and editor cruft all take up space in the file and none of it is the picture.
Every one of those is reversible in the sense that matters: decode the optimised file and you get precisely the original image back.
Why the obvious way is worse
The default answer to "compress png" is a website. Upload the file, wait, download it back. It works, and it has three costs that are easy to overlook.
You upload the image. For a screenshot of a spreadsheet, a client mockup, or a photo of something private, that is a real decision that a convenient interface makes feel like no decision at all.
There is a cap, and it is low. Free tiers typically stop at 20 files and 5 MB each. If you have a folder of 300 exported assets, you are doing it in fifteen batches, or paying.
It is a one-off. You will have exactly the same folder of oversized PNGs next month, and you will do exactly the same thing again.
Doing it on your own machine removes all three. There is no upload step, so there is no size limit and nothing to think about; and because it runs locally, it can be made to happen automatically.
Make it happen by itself
The version of this that actually saves time is a rule rather than a task. Point PhotoTray at the folder your screenshots land in and tell it: when a PNG shows up here, optimise it.
From then on every screenshot you take is compressed before you go looking for it. No dialog, no upload, no remembering. That is the thing a website structurally cannot do for you, and it is the reason to have a tool on the machine at all.
When to convert instead
Compression is the right answer when the image needs to stay exact. Sometimes it does not.
- A photograph that is going on a web page should probably be a JPG or a WebP. The saving is an order of magnitude larger than lossless optimisation can achieve, and on a photograph the quality difference is invisible.
- A logo or screenshot going on a web page should probably be WebP — it keeps transparency, is lossless if you want it to be, and is still smaller than an optimised PNG.
- Anything being archived, edited further, or handed to a designer should stay a PNG. Optimise it and leave it lossless.
The honest limitation
PhotoTray compresses PNGs using its image engine's own encoder. That is a real reduction and it is genuinely lossless. A dedicated PNG optimiser that tries many more filter and palette combinations will sometimes squeeze out a few percent more, at the cost of taking much longer per file.
If those last few percent matter to you — you are shipping assets where every kilobyte is budgeted — it is worth knowing that a specialist tool exists. For the folder of screenshots and exports that most people actually have, the difference is not something you would notice.
The tool for this — Resize & Compress
Cap the long edge, hit a target file size, crop, or squeeze a PNG losslessly — whole folders at a time. No upload, so no file-size ceiling and no queue, and the settings you pick can become a rule that runs on every screenshot you take from then on. Learn more about the Resize & Compress.