Is PNG Lossy or Lossless? The Direct Answer

PNG is lossless. A normal PNG file preserves every pixel exactly. You can decode it, edit it, and save it again without JPEG-style generation loss. The confusion comes from tools that create smaller PNG files by reducing colors first. The PNG format is lossless; some PNG optimization workflows are not.

TL;DR: PNG is a lossless image format. Its compression is reversible and does not throw away image data. Use PNG for screenshots, UI images, diagrams, transparent graphics, and raster logos when exact pixels matter. The exception is not the PNG format itself. It is tools like TinyPNG, pngquant, or PNG-8 export settings that reduce colors before saving. Those workflows can be lossy even though the output file is still .png.

Is PNG Lossy or Lossless?

PNG is lossless by design.

The W3C PNG Third Edition specification describes PNG as a lossless, portable, well-compressed format for static and animated raster images. That means the format is built to store pixel data without discarding it.

If you save a clean screenshot as PNG, the text and UI edges remain exact. If you open that PNG and save it again as PNG, it does not degrade the way JPEG does.

The important distinction:

So the clean answer is: PNG is lossless, but a PNG file can contain image data that was already damaged or reduced before it became PNG.

How Does PNG Compression Work?

PNG compression has two main ideas: filtering and DEFLATE.

1. PNG Filters the Pixel Rows

Before compression, PNG can transform each row of pixels with prediction filters. Instead of storing every value directly, the file can store differences from nearby pixels.

For example, a row like this:

200, 201, 202, 203

can become something closer to:

200, 1, 1, 1

That is easier to compress because repeated small values are common.

PNG has several filter types, including None, Sub, Up, Average, and Paeth. The encoder chooses filters that make the data easier to compress.

2. PNG Uses DEFLATE Compression

After filtering, PNG uses DEFLATE compression. DEFLATE combines repeated-pattern matching with entropy coding. It is the same broad family of compression used by ZIP and gzip.

The important part: this compression is reversible.

Nothing is thrown away. The decoder can reconstruct the original pixel values from the compressed stream.

That is why PNG is good for screenshots, interface images, flat graphics, diagrams, and text-heavy visuals. These images often have repeated shapes, flat regions, and sharp edges that PNG can represent cleanly.

When Can a PNG Workflow Be Lossy?

The PNG format is lossless. The workflow around it can still be lossy.

1. Converting JPEG to PNG

Saving a JPEG as PNG does not restore lost quality.

JPEG compression already discarded image data. If the JPEG has block artifacts, color shifts, or blurred text, the PNG will preserve those damaged pixels exactly.

That usually gives you the worst result: a larger PNG file with the same JPEG artifacts.

2. TinyPNG and pngquant

Tools like TinyPNG and pngquant can make PNG files much smaller by reducing colors. This is useful, but it can be lossy.

The common method is color quantization: reduce a full-color image to a smaller palette, often 256 colors. For simple icons or flat logos, the change may be invisible. For gradients, shadows, photos, or complex UI screenshots, it can create banding or color shifts.

The output is still a valid PNG file. It is just not pixel-identical to the original.

3. PNG-8 Export

PNG-8 usually means indexed-color PNG with up to 256 colors.

If your image genuinely has only a small set of colors, PNG-8 can be fine. If your image has gradients, transparency, shadows, or anti-aliased edges, reducing it to PNG-8 can throw away color detail.

This is why "PNG is lossless" and "this PNG export was lossy" can both be true.

4. Resizing, Cropping, or Color Conversion

PNG does not protect you from edits.

If you resize an image, change color profiles, flatten transparency, or remove colors before saving, those changes are baked into the PNG. PNG will preserve the result losslessly, but it cannot preserve data you already changed.

PNG-8, PNG-24, and PNG-32 Explained

These names are common in design tools, but they are not the cleanest technical names from the spec. Still, they are useful shorthand.

PNG-8 can be smaller, but palette reduction can be lossy when the source has more colors than the palette can represent.

PNG-24 and PNG-32 can preserve much more visual detail, but they are often larger.

For transparent UI assets, PNG-32 or lossless WebP is usually safer than forcing PNG-8.

PNG Optimization: Lossless vs Lossy

There are two kinds of PNG optimization.

Lossless PNG Optimization

Lossless optimizers keep the pixels the same and try to store them more efficiently.

Examples:

• oxipng
• OptiPNG
• pngcrush
• zopflipng

These tools can remove unnecessary metadata, try better filters, and search for smaller DEFLATE encodings. When used in lossless mode, decoded pixels stay the same.

This is safe for source files, screenshots, and graphics where exact pixels matter.

Lossy PNG Optimization

Lossy PNG optimizers change the pixels to make the file smaller.

Examples:

• TinyPNG
• pngquant
• PNG-8 palette export

These can be useful for web delivery, but they are not pixel-perfect. The quality loss may be invisible on simple graphics and obvious on gradients or screenshots.

Quick Note: If your goal is smaller web delivery with exact pixels, try lossless WebP. Google's WebP lossless and alpha study found WebP lossless could beat PNG encodings in its test corpus while preserving pixels.

When Should You Use PNG?

Use PNG when exact pixels, transparency, or compatibility matters.

PNG is still excellent for documentation screenshots, UI examples, simple diagrams, and transparent raster graphics.

PNG is a poor default for photography. A photo saved as PNG can be many times heavier than the same visual image delivered as JPEG, WebP, or AVIF.

For a wider format comparison, see image formats explained and WebP vs AVIF vs JPEG.

PNG vs WebP Lossless

PNG is the compatibility baseline. WebP lossless is the modern web upgrade.

For websites, the practical setup is simple:

<picture>
  <source srcset="/images/screenshot.webp" type="image/webp" />
  <img
    src="/images/screenshot.png"
    alt="Dashboard screenshot showing image optimization settings"
    width="1200"
    height="800"
    loading="lazy"
  />
</picture>

Modern browsers get WebP. Everyone else gets PNG.

An image CDN format conversion setup can handle this automatically through the browser's Accept header.

PNG on the Web

PNG is still everywhere.

HTTP Archive's 2024 Media chapter reported PNG at 28.4% of mobile image resources, behind JPEG at 32.4% and ahead of GIF, WebP, SVG, and AVIF in raw format share.

That does not mean PNG is always the best delivery format. It means PNG is deeply embedded in web workflows because it is reliable, simple, transparent, and universally supported.

The better 2026 pattern is:

• keep PNG when exact pixels and compatibility matter
• serve lossless WebP where browser support allows
• use JPEG/WebP/AVIF for photos
• use SVG for true vector logos and icons

For more PNG-specific delivery advice, read image CDNs for PNG images.

Summing Up!

PNG is lossless. The format preserves pixels exactly and does not degrade from repeated saves the way JPEG does.

The catch is workflow. If you start from a damaged JPEG, reduce colors with pngquant or TinyPNG, export as PNG-8, resize, or flatten transparency, the resulting PNG will preserve those changed pixels. PNG cannot restore information that was already removed.

Use PNG for screenshots, UI images, transparent graphics, and compatibility. Use lossless WebP when you control modern web delivery. Use JPEG, WebP, or AVIF for photos. For production sites, an image CDN can keep your PNG originals intact while serving smaller modern formats to browsers that support them.