PNG to JPG Converter

Drop in a batch of PNG files, choose the quality and the color that fills every transparent pixel, and read the real size saving per file before you download anything. The images stay on your machine.

PNG to JPG conversion workspace

  • No upload, no queue
  • Up to 24 files at once
  • Alpha is flattened, never dropped

Drop PNG files here or tap to browse

Conversion starts the moment the files land. PNG only, 40 MB per file, 24 per batch.

PNG to JPG is a trade, and the alpha channel is what you hand over

A PNG keeps every pixel exactly as the encoder received it and carries a per pixel alpha channel alongside the color. JPEG keeps neither. The format throws away detail your eye is bad at noticing, and its bitstream has no place to record transparency at all. So the question is never whether a PNG converts. Every PNG converts. The question is what fills the transparent regions and how much detail you are willing to trade for a file a fifth of the size.

This page answers both out loud. The matte color is a control instead of a hidden default, and each file reports the byte count before and after so the saving is a measured number rather than a promise. Batch a folder of screenshots and you will watch some rows shrink by eighty percent while others barely move. Those stubborn rows are telling you something worth reading, and the section on line art below explains what.

Pick the matte before you touch the quality slider

Transparent pixels have to become something. Most converters silently pick white, which works for a product shot heading onto a white page and looks broken the moment the same logo lands on a dark header. The matte you choose is baked in permanently, so it pays to match the background the image will sit on.

The soft edge problem

Anti aliased edges are not fully transparent or fully solid. A logo edge holds a ring of pixels at partial alpha, and each one blends toward your matte. Choose white for an icon destined for a dark page and you get a pale halo tracing every curve, visible on the final page and impossible to remove afterward. Set the matte to the page background color and the halo disappears, because the blend now matches what sits behind the image anyway.

What each quality number does to a photograph

SettingGood forWhat appears first
100An intermediate you will re-editFiles balloon with almost no visible gain over 95
95Print, archive, client deliveryNothing at ordinary zoom, faint noise in flat sky at 400%
90The default here, general web useSlight softening where fine texture meets a flat area
80Blog images, product gridsRinging along hard edges, mild banding in gradients
65Thumbnails and previewsVisible 8 by 8 blocks in shadows and open sky
Below 50Little worth doingColor bleeding, blocking, muddy skin tones

Percentages are not a standard. A quality of 80 in one encoder is not the same as 80 in another, because the number maps onto a quantization table the specification never fixed. Treat the slider as a dial for this page and read the byte counts under each row rather than trusting the number to travel between programs.

Where the saving comes from

PNG compresses losslessly with DEFLATE, so a photograph with a million distinct colors gives the algorithm almost no repetition to work with. JPEG works the other way around. It converts the image to a brightness channel plus two color channels, samples the color channels at half resolution because human vision is far sharper for brightness than for hue, splits the result into 8 by 8 blocks, and rounds the high frequency detail in each block toward zero. The rounding is the compression. Detail you rarely notice gets thrown away first, and the file drops from megabytes to kilobytes.

That design is why the same slider position produces wildly different results on different images. Photographs are mostly gradual change, which fits the model. Screenshots are mostly hard edges between flat areas, which is the worst case for it.

Files you should leave as PNG

Photographs, textures, scans, screenshots without small text and any large image already full of natural color variation are where the conversion pays. If a row in the batch reports a saving under thirty percent, look at the image again. Most of the time it belongs on the list above.

Three steps, all inside your browser tab

  1. Each file is read through FileReader and decoded into an offscreen image. Nothing is posted anywhere. Cut your network connection after the page loads and every feature keeps working.
  2. A canvas the size of the source is filled with your matte color, then the PNG is drawn on top. Partial alpha blends against the matte during that draw, which is where soft edges get their final color.
  3. The canvas is handed to the browser JPEG encoder at your chosen quality. The resulting blob is measured, compared against the original byte count, and held in memory until you download it or clear the batch.

Where this converter stops

Questions people ask before converting a batch

Transparency, quality numbers, and what a browser based encoder will not do for you.

What happens to the transparent parts of my PNG?

They become solid pixels in the color you set under the quality slider, because the JPEG format has no alpha channel to store them in. Pixels at partial alpha, which is most of an anti aliased edge, blend toward that color rather than snapping to it. Match the matte to the background the image will sit on. A white matte on a logo headed for a dark page leaves a pale outline around every curve that no later edit removes.

Which quality setting should I use?

Start at 90 and look at the reported saving. For photographs going onto a web page, anything from 80 to 90 is hard to fault at normal viewing size. Drop to 65 only for thumbnails, where the image is displayed small enough to hide the block artifacts. Go to 95 when the file is being printed or handed to a client. Above 95 the file grows quickly with almost nothing visible in return.

My JPG came out bigger than the PNG. Why?

The source is almost certainly a screenshot, a logo, a chart or line art. Those images hold large flat areas and sharp edges, which PNG compresses efficiently and JPEG does not. The encoder also adds faint noise inside flat regions that costs bytes to store. When a row reports a poor saving or a size increase, the honest answer is to keep the PNG.

Do my images get uploaded to a server?

No. Decoding, matte compositing and JPEG encoding all run in your browser through the canvas API, and the page makes no network request once loaded. Turn off your connection after the page appears and the converter keeps working. Nothing is stored between visits either, so closing the tab clears the batch.

Is EXIF data or the color profile kept?

No. Drawing to a canvas produces raw pixels with no metadata attached, so EXIF fields, ICC profiles, copyright tags and camera information do not appear in the output. Images tagged with a wide gamut profile are converted to the browser display space during decoding, which shifts saturated colors slightly. If either of those matters, encode with a desktop tool instead.

How many files fit in one batch?

Twenty four files per batch, 40 MB per file. The practical limit is memory rather than either number, since every decoded image sits in RAM at four bytes per pixel until you clear the batch. Very large images convert one at a time without trouble, and a long queue of them on a phone will not. Convert in smaller groups when a batch stalls.

Can I convert JPG back to PNG afterwards?

The conversion runs, but no detail returns. JPEG discards information permanently, so a round trip gives you a lossless container holding data that has already been degraded, usually at a larger file size than the JPG. Keep the original PNG if there is any chance you will need it again.

Why does the same quality number look different in Photoshop?

Because the JPEG specification never standardized what a quality percentage means. Each encoder maps the number onto its own quantization tables, so 80 here, 80 in Photoshop and 80 in ImageMagick produce three different files. Compare the byte counts and the visible result instead of trusting the number to carry across programs.

Does the output work everywhere?

Yes. Baseline JPEG is the most widely supported image format in use, readable by every browser, operating system, email client, print workflow and camera made in the last three decades. That reach is the main reason to convert at all when transparency is not needed.