What Reduction Should You Realistically Expect from Your PNG?

Open a PNG and ask yourself how much can a png compress. The honest answer lands anywhere from zero to 85 percent. The spread is not random and it is not a quality slider. It is entirely controlled by what the image contains. A flat colour screenshot and a photograph saved as PNG share the same file format but compress to dramatically different sizes. The table below gives the real numbers by image type for the aggressive approach, palette reduction. Lossless-only numbers follow after.

Compression Percentages by Image Type

Image TypeLossy Palette ReductionLossless Only
UI screenshot70 85 percent5 15 percent
Logo, icon or line art60 80 percent5 15 percent
Chart or diagram70 80 percent5 15 percent
Illustration with gradients40 65 percent5 15 percent
Photograph stored as PNG40 70 percent0 5 percent
Already-optimised PNG0 percent0 percent

How PNG Prediction and Noise Affect File Size

PNG is always compressed and always lossless. There is no uncompressed PNG. The algorithm works in two stages. First it filters each row of pixels. The encoder picks one of five filter types (None, Sub, Up, Average, Paeth) that predicts each pixel from its neighbours and stores the difference. Then DEFLATE, the same compression used in ZIP, compresses those filtered bytes.

Flat colour, sharp edges and repeated patterns make the prediction correct. The differences are mostly zero, and zeros compress to almost nothing. That is why a screenshot of a dialog box with a solid background shrinks to a fraction of its original size.

Photographic sensor noise makes the prediction wrong nearly every time. You get small random differences that are close to incompressible. This is the whole reason photos are huge as PNG. JPEG is lossy: it discards high-frequency detail the eye is poor at noticing before compressing. PNG promises exact pixels, so it can never do that.

How Much Can a PNG Compress Percentage

The question how much can png compress percentage depends entirely on whether you stay lossless or accept a visible change.

Lossless optimisation searches harder for the best filter per row, re-runs DEFLATE at maximum effort, and strips metadata chunks like tEXt, iTXt, eXIf and iCCP colour profiles. The decoded pixels are identical. Typical saving is 5 to 15 percent. Sometimes zero on an already-optimised file. Re-encoding can produce a larger file than the original; a correct tool returns the original in that case.

Palette reduction changes the image. It rebuilds the image from an adaptive palette of typically 256, 128 or 64 colours chosen from the colours actually present. Two compounding effects happen. Each pixel needs 7 or 8 bits instead of 24 or 32. Thousands of near-identical shades collapse into one entry, creating exactly the repetition DEFLATE needs.

The visible failure mode is banding in smooth gradients, because a soft fade needs many closely spaced colours to look continuous. The second failure mode is a stair-step fringe on semi-transparent edges. The soft alpha band around a rounded logo needs enough palette entries or it snaps into discrete steps. A gentler setting fixes it.

Tools That Do Palette Reduction

In Photoshop use File Export Export As and tick the Smaller File (8-bit) checkbox. That switches to a 256-colour indexed palette. Save for Web (Legacy) gives finer control: PNG-8 with a Colors dropdown from 2 to 256, a Dither slider, and a live file size readout.

In GIMP use Image Mode Indexed, choose Generate optimum palette and set maximum number of colours, then export. GIMPs Export As dialog also has a Compression level slider from 0 to 9. That slider is entirely lossless. Level 9 is slower and slightly smaller. It does not reduce colours.

pngquant does palette reduction on the command line. TinyPNG, the online service, does the same thing server-side.

Does PNG Compress at All?

Yes, does png compress is a question with a clear answer: PNG always compresses, and the compression is always lossless. What varies is the compression ratio. A PNG file that was saved from a modern image editor is already compressed once. Running it through a lossless optimiser may still find a few more percent by trying harder filter choices and stripping metadata.

On a photograph, lossless compression is practically a rounding error. The 0 to 5 percent you might get barely registers. The real saving for photographs comes from changing format. A photograph as PNG is roughly ten times the size of the same photo as JPEG at a quality difference invisible at normal viewing distance.

On flat artwork and screenshots, lossless compression is a solid starting point but palette reduction is where the big numbers live. A UI screenshot at 70 to 85 percent off is a genuinely different file size than the original. If you need to shrink further, reduce dimensions first. An image displayed at 800 pixels wide and stored at 3000 carries roughly fourteen times the pixel data anyone will see. For high-density screens serve about twice the display width. Roughly 2560px on the long edge covers full-screen use. A blog image needs about 1600px.

Why Format Choice Beats Any Percentage

A number like 85 percent sounds final. It is not. The format you choose determines the entire range of possible sizes. WebP lossless is about 26 percent smaller than PNG. JPEG has no alpha channel at all. Converting a transparent PNG to JPEG flattens transparency onto a solid colour, usually white. JPEG handles sharp edges badly, producing visible ringing around text and lines, and on flat artwork often produces a larger file than PNG.

WebP supports a full alpha channel like PNG and is supported by all current browsers. Some older desktop applications still cannot open it. That is the trade-off.

The cheapest order to reduce a file is: strip metadata losslessly, then reduce the palette in steps, then reduce dimensions, then change format. Settings move in discrete steps, so a result often lands well under the requested ceiling rather than exactly on it. Palette reduction cannot be undone. Keep the original and compress a copy. Compressing an already palette-reduced file again gives a worse result than going to the low colour count once, because the second pass chooses from an already impoverished set. Lossless re-compression is safe to repeat.

When To Ignore Every Number On This Page

Print resolution is far higher than screen and exposes banding invisible on a monitor. Send print the untouched original. For web use every percentage on this page applies directly to the file size your users will download. Resize before compressing. Compressing pixels you are about to discard is wasted work and the compressor performs better on the smaller image.