Lossy vs Lossless GIF Compression Explained

Lossless compression: no quality loss

Lossless GIF compression finds patterns in the pixel data and encodes them more efficiently — without changing a single pixel. Gifsicle's --optimize flag does this by:

• Rewriting frames to only encode the pixels that changed from the previous frame
• Sorting the color palette to improve LZW run-length compression
• Removing redundant metadata and extension blocks

This typically achieves 10–30% reduction on unoptimized GIFs. On already-optimized GIFs, the gain is smaller.

Lossy compression: trading quality for size

Lossy GIF compression (Gifsicle's --lossy flag) deliberately alters pixel colors to create longer LZW run-length sequences. This makes the compression algorithm much more effective at the cost of visible artifacts:

• At --lossy=20: very subtle, barely noticeable
• At --lossy=80: moderate dithering and color banding in gradients
• At --lossy=200: significant artifacts, blocky appearance

Color reduction: the biggest lever

Reducing the number of colors in the palette is technically lossless within the reduced palette — the encoder picks the best 64 or 128 colors and maps all other pixels to the nearest match. This is highly effective and often yields 20–50% size reduction with minimal perceptual quality loss on simple GIFs.

Which to use?

GifMash's Quality slider maps across these tiers automatically. The Target Size mode combines all three techniques in a binary search to hit your exact KB target.