PNG to JPG

What Gets Lost: Transparency, Color Depth, and Metadata

PNG supports an alpha channel — a per-pixel transparency value that lets you place images over any background without a visible rectangle. JPG has no alpha channel at all. When ImageMagick converts a PNG to JPG, it composites the image against a white background by default, so every fully transparent pixel becomes solid white and semi-transparent pixels blend toward white. If your PNG has a transparent background (common with logos, icons, and UI elements), the result will look different once that transparency is flattened.

Color depth also changes. PNG can store 24-bit RGB or 32-bit RGBA (with alpha), and even supports 48-bit deep color. JPG is always 8 bits per channel in RGB — no alpha, no wide color. For standard photographs this makes no practical difference, but for images with subtle gradients or wide color gamut the conversion can clip highlights or compress tonal transitions.

Metadata is another consideration. By default, the converter has Strip Metadata enabled, which removes EXIF data, ICC color profiles, and other embedded chunks during conversion. This reduces file size but means GPS coordinates, camera settings, and color management profiles won't carry over to the JPG. You can disable stripping in Configuration Options if you need to preserve metadata.

Inside the Conversion: From Lossless PNG to Lossy JPG

The conversion pipeline has several stages. First, ImageMagick decodes the PNG file — decompressing the DEFLATE-compressed pixel grid back into raw pixel data. If the PNG has an alpha channel, the decoder composites it against a white background to produce a flat RGB image. If you've specified a target width or height in the configuration, the image is then resampled to the new dimensions using the selected fit mode.

The resized RGB data is then passed to the JPEG encoder, which applies lossy DCT (Discrete Cosine Transform) compression. The encoder splits the image into 8×8 pixel blocks, transforms each block into frequency coefficients, and quantizes those coefficients — rounding less significant values to zero. This is where file size savings come from, and where visual quality is traded away. The quality setting (1-100, default 75) controls how aggressively the quantization tables compress the frequency data. Lower values discard more coefficients, producing smaller files with more visible blocking artifacts.

The three fit modes control what happens when you set width or height. Max (default) fits the image within the specified bounds without upscaling. Crop fills the dimensions exactly and trims overflow. Scale forces the exact dimensions, which may stretch the image if the aspect ratio doesn't match. If you don't set any dimensions, the original pixel size is kept.

This is a one-way lossy step. Once the PNG data has been encoded as JPEG, the discarded frequency information is gone permanently. You cannot recover the original PNG quality from the JPG output — so always keep your original PNG if you might need the lossless version later.

File Size Savings: Measuring What You Gain and Lose

The file size reduction from PNG to JPG varies dramatically depending on image content. Photographic images with complex textures and gradients see massive savings because JPEG's DCT compression excels at encoding continuous tones. Simple graphics with flat colors and sharp edges see smaller gains — and sometimes the JPG is barely smaller because the content doesn't compress well with DCT.

The pattern is clear: the more photographic the content, the bigger the savings. A 2.4 MB screenshot with gradients and UI chrome drops to ~180 KB at quality 75 — a 92% reduction. A DSLR photo exported as PNG (5.8 MB of lossless pixel data) compresses to ~1.1 MB as JPG with minimal visible difference. But a 45 KB icon with flat colors and sharp edges only drops to ~38 KB, and the JPG version introduces ringing artifacts around the edges that weren't there before.

Learn more about the JPG format and how its compression characteristics differ from PNG.

When to Keep PNG Instead of Converting

Not every PNG should become a JPG. Several categories of images are better served by the lossless format, even at the cost of larger file sizes.

Images with transparency. Logos, UI elements, overlays, and any graphic designed to sit on top of other content depends on the alpha channel. Converting to JPG replaces that transparency with solid white, breaking the intended visual presentation. If you need a smaller file with transparency, use WebP instead.

Screenshots with small text. JPEG's 8×8 block quantization blurs the edges of text characters, especially at small font sizes (10-14px). The effect looks like a faint halo or smear around each letter. If your screenshot contains code, terminal output, or fine print that needs to stay readable, PNG preserves every pixel edge exactly.

Graphics with few colors and sharp edges. Logos, icons, diagrams, and illustrations with solid color regions and crisp boundaries are where JPEG artifacts are most visible. The DCT encoder produces ringing — visible halos around high-contrast edges where the mathematical approximation overshoots. These artifacts look especially bad on clean geometric shapes. PNG's lossless DEFLATE compression handles this content perfectly.

Images that will be edited further. Every JPEG encode is a lossy pass. If you convert a PNG to JPG, edit the JPG in Photoshop, and re-save it, the file goes through two rounds of DCT quantization — each one degrading the image slightly. If the image will be opened, modified, and saved multiple times, keep the working copy as PNG and only export to JPG as the final delivery step. Learn more about how images lose quality after conversion.

If the image doesn't fall into any of these categories — especially if it's a photograph or a large screenshot where a few KB of text blur won't matter — JPG is almost always the right choice for its dramatically smaller file size. You can also compress the image after conversion if you need to reduce the JPG size even further.