Every time you save or compress an image, you're choosing between lossless and lossy compression — whether you know it or not. Here's exactly what each method does, when to use each, and how it affects your images.
Compress Images FreeLossless compression finds and eliminates redundancy in the image data without changing what the data represents. Common techniques include:
Run-Length Encoding (RLE): Instead of storing "white, white, white, white, white" for a row of 5 white pixels, it stores "5 white." Simple but effective for images with large areas of solid color — like screenshots and diagrams.
LZ77 / DEFLATE (used in PNG): Finds repeated sequences of data and replaces them with references to earlier occurrences. "Here's 12 bytes that match the pattern 40 bytes ago" — the decoder follows the references to reconstruct the original. PNG uses a combination of prediction filters and DEFLATE to typically achieve 10–30% compression on photographic images — good, but much less than lossy.
Palette reduction (GIF): Limits the image to a maximum of 256 colors, then stores each pixel as an index into that palette. Effective for simple graphics but not suitable for photographs.
Lossy compression exploits the limitations of human visual perception to discard image data that is unlikely to be noticed. JPEG — the most widely used lossy format — uses a multi-step process:
1. Color space conversion: The image is converted from RGB to a color space (YCbCr) that separates brightness from color. Human eyes are more sensitive to brightness differences than to color differences, so color information can be compressed more aggressively.
2. Chroma subsampling: Color channel resolution is reduced — often to half the original in each dimension. For most natural photographs, this is nearly invisible to the human eye.
3. Discrete Cosine Transform (DCT): Each 8×8 pixel block is transformed into frequency components. High-frequency components (fine texture, subtle gradients) carry less perceptual weight and are rounded or discarded based on the quality setting.
4. Quantization: The frequency coefficients are divided by a quality matrix and rounded to integers. This is the step where information is permanently lost — the discarded precision cannot be recovered.
5. Entropy coding: The remaining data is losslessly encoded for final compression.
Modern formats like AVIF and WebP use more sophisticated transforms (AV1, VP8) that apply similar principles with less visible artifact production at low bitrates.
| Property | Lossless | Lossy |
|---|---|---|
| Data preservation | 100% — pixel-perfect reconstruction | Permanent data loss — cannot be undone |
| File size (photos) | Large — 2–10× bigger than lossy | Small — 60–90% smaller than lossless |
| File size (graphics, screenshots) | Efficient — flat color compresses very well | Often larger or introduces artifacts on text |
| Re-save quality degradation | None — re-saving losslessly is always safe | Each re-save applies more lossy degradation |
| Best for photos | Unnecessarily large files | Optimal — small files, acceptable quality |
| Best for logos & graphics | Ideal — sharp edges, flat color, no artifacts | Poor — blurry edges, color fringing on text |
| Best for transparency | PNG lossless with alpha | AVIF/WebP lossy can do transparency too |
| Editing workflow safety | Always keep originals in lossless format | Each edit-save cycle compounds quality loss |
| Web delivery (photos) | Too large — poor page speed | Optimized — fast loading, good quality |
| Web delivery (icons/UI) | PNG or SVG — pixel-perfect rendering | Lossy compression blurs thin lines |
| Primary formats | PNG, WebP (lossless), AVIF (lossless), GIF, TIFF | JPEG, WebP (lossy), AVIF (lossy), HEIC |
PNG (Portable Network Graphics) is strictly lossless. It uses DEFLATE compression to reduce file size without discarding any pixel data. PNG is the correct choice for logos, icons, screenshots, UI graphics, and any image with text, sharp edges, or areas of flat color. For photographic images, PNG files are typically 5–10× larger than equivalent JPEGs — a real cost at scale.
Use PNG when: exact pixel fidelity matters, the image has transparency, the image contains text or a logo, or the image will be edited further.
JPEG (Joint Photographic Experts Group) is strictly lossy. Its quality slider (typically 0–100) controls how aggressively data is discarded. Quality 85–92 is the standard web sweet spot — imperceptible quality loss for photos with dramatically reduced file sizes. Below quality 75, blocking artifacts become noticeable. Below quality 50, the image looks visibly degraded.
JPEG cannot be saved losslessly, cannot contain transparency, and should never be re-saved repeatedly — every save cycle applies the lossy algorithm again. Always edit from a lossless original.
WebP is unusual in that it supports both lossless and lossy compression in the same format, and both modes can include an alpha channel. WebP lossy is 25–35% smaller than JPEG at equivalent quality. WebP lossless is typically 20–30% smaller than PNG for the same image.
When converting PNG to WebP, the lossless mode preserves all pixel data. When converting JPEG to WebP, the lossy mode is appropriate. You can convert PNG to WebP or convert JPG to WebP with ImgSwift's free tools.
AVIF, like WebP, supports both lossless and lossy compression. AVIF lossy produces the smallest file sizes of any widely-supported web format — typically 40–55% smaller than JPEG and 20–35% smaller than WebP lossy. AVIF lossless is effective but encodes slowly.
AVIF is the best choice for photographic web images in 2026 where encoding time is not a constraint. Use it with a JPEG fallback via the HTML <picture> element.
Lossy compression is not binary — it's a spectrum. Most lossy formats let you choose a quality level that determines how aggressively data is discarded.
Every pixel is preserved. File size is larger but quality is always maximum.
This is the web sweet spot. Files are 60–80% smaller than lossless with no visible quality difference for photographs.
Aggressive compression with visible artifacts, blocking, and color banding. Only appropriate for thumbnails or situations where bandwidth is extremely limited.
One of the most important practical consequences of lossy compression is generation loss. Every time a JPEG is opened, edited, and re-saved as JPEG, the lossy algorithm runs again on already-compressed data. Each cycle discards more information. After 3–5 re-saves at moderate quality settings, the image shows clear degradation — increased blocking, color smearing, and loss of fine detail.
This is why professional photographers and designers always work from RAW or lossless source files and only export to JPEG as the final delivery step. Never use a JPEG as your working file. The original should be a RAW, TIFF, or lossless format. Export to JPEG once, for delivery, and never re-save it.
Lossless formats don't have this problem. You can open, edit, and re-save a PNG file thousands of times with zero quality degradation.
Camera RAW, TIFF, or uncompressed formats for photography. PNG or vector (SVG) for graphics and UI assets. This is your master file — never discard it.
All resizing, cropping, color correction, and retouching happens on the lossless source. You can save work-in-progress as lossless (PSD, TIFF, PNG) as many times as needed without quality loss.
For photographs: export as AVIF (with JPEG fallback) or WebP lossy (with JPEG fallback). For logos and icons: use SVG where possible, PNG lossless otherwise. For screenshots and UI: PNG lossless or WebP lossless. Use ImgSwift's compressor to optimize without needing to install software.
Your web-ready JPEG or WebP lossy is a delivery artifact. If you need to make changes, go back to the lossless original and re-export. Never open your web JPEG, make a small change, and re-save it as JPEG — you'll compound the lossy degradation.
Lossless and lossy compression aren't competitors — they're tools for different jobs. Lossless compression is the correct choice for any image where exact pixel fidelity matters: logos, icons, screenshots, transparency, archives, and working files. Lossy compression is the correct choice for photographs delivered over the web, where smaller file sizes meaningfully improve performance without perceptible quality loss.
The most common mistake is using the wrong type for the job — saving a photograph as a giant PNG when JPEG or WebP would be 80% smaller with no visible difference, or saving a logo as a JPEG and watching the crisp edges get blurry and the text get blocky. Understanding the difference is one of the most practical improvements any web developer or designer can make to their workflow.
Lossless compression reduces file size without discarding any image data — the decompressed image is pixel-for-pixel identical to the original. Lossy compression achieves much smaller files by permanently discarding image data the human eye is unlikely to notice. Lossless is used where exact fidelity matters (logos, screenshots, medical imaging). Lossy is used where file size matters most and minor quality loss is acceptable (web photos, social media).
PNG is the most common lossless web image format. GIF also uses lossless compression but is limited to 256 colors. WebP and AVIF both support lossless modes in addition to lossy. TIFF and BMP can be stored losslessly. RAW camera formats are lossless or near-lossless captures of sensor data.
JPEG is the most widely used lossy image format. WebP lossy mode, AVIF lossy mode, and HEIC are modern lossy formats. Most social media platforms re-encode uploaded images with lossy compression at their own quality settings, regardless of the original format you uploaded.
Yes. Lossy compression permanently discards image data. Once an image has been saved as a lossy format like JPEG, the discarded information cannot be recovered. Re-saving a JPEG as JPEG applies lossy compression again, compounding the quality loss with each save — a problem known as generation loss. Always keep an original lossless or uncompressed copy if you plan to edit the image again.
Not for photographs. PNG's lossless compression produces much larger files than JPEG for photographic content — often 5–10× larger with no visible quality benefit for most viewers. PNG is better than JPEG for graphics with text, flat colors, sharp edges, logos, and any image requiring transparency. JPEG is better for photographs where file size matters. Choosing the wrong format for the image type is one of the most common optimization mistakes on the web.
Yes, using lossless compression. ImgSwift's image compressor uses lossless optimization for PNG files — removing metadata, optimizing color tables, and applying better compression without changing a single pixel. You can also strip EXIF metadata from images to reduce file size without any quality loss at all. For JPEG files, removing metadata and re-encoding with optimal settings can reduce size somewhat, but further reduction beyond that requires accepting some quality trade-off.