Make a PNG smaller by saving it as a JPG, with the quality under your control and the size change shown before you commit. Everything happens in this tab, on your own machine.
Drop images here, or click to choose
Drop PNG files here. Transparent areas become white.
Free · no sign-up · nothing uploaded. Your files are decoded and re-encoded on your own device, so a photo with a face or a location in it never leaves the machine.
One, or a folder of them. There is no upload, so a batch of large screenshots starts immediately.
Set the quality. 85 is the default and the right answer for most photographs.
Each card shows the old size, the new size and the difference. Save one file, or the batch as a ZIP.
PNG to JPG is worth doing when the file is a photograph and the size is a problem. It is worth skipping in more cases than people expect, and the honest version of this page says so before the tool does the work.
| Your PNG is | Convert? | Why |
|---|---|---|
| A photograph | Yes | Often a tenth of the size at quality 85, with nothing you can see missing. |
| A screenshot with text | No | JPG puts a halo round every letter, and the file often gets bigger anyway. |
| A logo with a clear background | No | The transparency becomes a white box. That box shows up on any coloured page. |
| A chart or diagram | No | Flat colour and hard lines are exactly what PNG compresses best and JPG handles worst. |
| A photo, but something demands PNG | No | An upload form asking for PNG usually means it needs the transparency or the exactness. |
If the goal is only a smaller file and the format does not matter, the image converter offers WebP as well, which beats JPG by roughly a quarter at the same quality and keeps transparency too.
A JPG cannot be partly see-through. There is no alpha channel in the format, so when a PNG with transparency is converted, every transparent pixel has to become some solid colour.
Left alone, a browser canvas starts out with transparent black, and the JPG encoder writes those pixels as black. That is why converting a logo somewhere else so often hands back a logo on a black rectangle. This tool fills the canvas with white first, because white matches the page most logos end up on and is the outcome people expect.
Neither is correct in every case. If your picture belongs on a dark background, or on a brand colour, the only conversion that preserves your options is one to a format that keeps the alpha channel. Convert to PNG or WebP and composite it later.
85 does not mean 85% of the picture survives. It is a value handed to the JPG encoder that decides how aggressively it may discard detail, and the same number means something different in JPG, WebP and AVIF, because all three are doing different maths underneath.
What it looks like in practice: at 90 to 95 a photograph is visually identical to the source and the file is large. At 80 to 85 it is still hard to tell apart and the file is roughly half the size, which is why the slider starts there. At 70 the flat areas of sky and skin begin to band. Below 60 the eight-by-eight blocks JPG works in start to become visible, especially around hard edges.
Because every file is re-encoded from the original bitmap rather than from the previous output, moving the slider is free. You are not stacking a fresh lossy pass with every nudge, which is what would happen if the tool converted its own results.
They come out white. JPG has no alpha channel, so there is nowhere for transparency to go and something has to be put behind the picture. White is chosen here deliberately: a browser left to fill that space on its own uses black, which is a far nastier surprise on a logo. The card warns you when a file had transparency in it, so you find out before you save rather than after you have sent it to someone. If you need it kept, convert to PNG, WebP or AVIF instead, all three of which hold an alpha channel.
Yes. JPG is lossy, so the encoder discards detail it judges you will not notice, and that detail does not come back. How much depends on the quality number and on what the picture is: a photograph survives quality 85 with nothing visible, while text and hard edges pick up a faint halo at the same setting. Every file here is re-encoded from the original you dropped in and never from the last result, so you can move the slider as many times as you like and the tenth version at 85 is exactly as good as the first.
Because the source was the kind of picture PNG is good at. PNG compresses flat colour and hard edges without losing anything, so a screenshot, a logo, a chart or a diagram is often smaller as a PNG than any JPG worth using. JPG wins on photographs, where PNG has to record the noise in every leaf and every patch of sky. If your file grew, the honest answer is that PNG was the right format for it and converting was the wrong move. The card shows the change before you save, so you can see which way yours went.
85 for a photograph, and leave it alone. That is where most people stop being able to see the difference from the original, and it is where the slider starts. Below 70 the flat areas start to band and edges pick up a halo. Above 92 the file grows quickly and the picture barely improves. For a screenshot or anything with text in it, either push to 95 or accept that PNG is the better format and do not convert at all. The number is a setting handed to the encoder, not a percentage of the picture that survives.