Recommended Free Tools
There is no quality setting that works best for every image. Choose a format based on whether the image is photographic, needs transparency, or must preserve exact pixels; resize it to its intended display dimensions; then compare a few exports. Use the smallest file that still looks right at the size people will see it and works in your delivery environment.
Start with the image’s needs
JPEG, PNG, and WebP make different trade-offs. JPEG is lossy; PNG uses lossless compression; WebP supports both lossy and lossless encoding and can preserve transparency. A quality number is an encoder control, not a universal score: the same number in different software—or in different WebP modes—does not guarantee the same appearance or file size.
| # | Preview | Product | Price | |
|---|---|---|---|---|
| 1 |
|
Compressed Image File Formats: JPEG, PNG, GIF, XBM, BMP (SIGGRAPH Series) | $59.37 | Buy on Amazon |
As an Amazon Associate I earn from qualifying purchases.
| Format and mode | Good starting use | What to compare |
|---|---|---|
| JPEG, lossy | Photographs where some irreversible change is acceptable | File size and artifacts in detail, gradients, skin, text, and saturated edges |
| PNG, lossless | Images that need exact decoded pixels, sharp edges, or transparency | File size and encoding effort; compression settings do not change decoded appearance |
| WebP, lossy | Photographs or other images where a visible difference is acceptable | File size, image artifacts, transparency edges if present, and target support |
| WebP, lossless | Images that need exact pixel reconstruction and WebP support | File size, encoding effort, and transparency behavior |
These are starting points, not guarantees. A photograph, a logo with sharp edges, and a graphic with a transparent background can react very differently to the same compression choices.
What settings mean in each format
JPEG: tune the encoder, then inspect the image
JPEG discards image information to reduce file size, so changing quality can change the reconstructed image. The libjpeg-turbo project’s cjpeg usage guide describes its own 0–100 quality scale, where 0 is worst and 100 is best, with a default of 75. It gives 50–95 as a general range for photographic images and recommends finding the lowest setting that still looks visually indistinguishable from the source. Those values describe cjpeg, not every application’s JPEG control.
Higher quality can increase file size sharply for little visible benefit in many cases, according to the same project guidance. Test the image rather than assuming that a value near 100 is worthwhile. Inspect fine detail, gradients, skin, text, and high-contrast or saturated edges for artifacts.
Google’s deprecated PageSpeed image-optimization page historically suggested JPEG quality 85 or lower when the source quality is higher, 4:2:0 chroma sampling, and progressive encoding for larger files. Treat those as historical experiment settings, not current universal rules.
If color is unnecessary, grayscale JPEG may be appropriate. During editing, keep a lossless working copy and export to JPEG once the edit is complete; repeated JPEG re-encoding cannot be undone.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
PNG: compression effort is not visual quality
PNG compression is lossless: changing the compression effort can change file size and processing time without changing decoded pixels. In ImageMagick, the PNG quality setting controls compression level and filter selection, not a visual-quality slider. Check the result against the original; rewriting an already optimized file can sometimes make it larger.
Before keeping PNG, check whether the image actually needs transparency. If every pixel is opaque, removing an unused alpha channel can save bytes. Consider metadata separately: retain color-profile, provenance, or camera information when the use case needs it, and remove only metadata that does not need to travel with the delivered image.
For a limited-palette graphic where some loss is acceptable, a palette-based lossy workflow may be worth testing. That is a separate process from changing PNG’s ordinary lossless compression level; do not expect a compression-level setting alone to make PNG lossy.
WebP: distinguish lossy quality from lossless effort
WebP supports lossy and lossless modes, transparency, and optional metadata. Google’s cwebp documentation gives -q a 0–100 scale and a default of 75. In lossy mode, lower values generally produce smaller files with lower quality. In lossless mode, the same control is an effort/size trade-off: lower values encode faster and usually produce larger output, while higher values spend more effort to reduce file size. The number therefore does not mean the same thing in both modes.
Free tools Windows power users keep installed
One-click scans. No signup required.
The -m method setting ranges from 0 to 6; higher values spend more time exploring encoding options. For transparent images, cwebp’s default alpha quality is 100; lowering it permits lossy alpha compression, so inspect edges against the backgrounds where the image will appear. In lossless mode, RGB values in fully transparent pixels are not preserved unless -exact is used. Metadata copying is configurable and defaults to none.
Use lossy WebP when its visible differences are acceptable, or lossless WebP when exact reconstruction is required. Google lists Chrome, Safari, Firefox, Edge, and Opera among browsers supporting WebP; check the browsers and other software your audience actually uses before choosing it as the only delivered format.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Compare exports with a repeatable workflow
- Keep the original. Make each candidate from the same source rather than compressing one candidate from another. Record the encoder and version, plus the settings, so you can reproduce the result.
- Set the delivered dimensions first. Resize to the intended pixel dimensions before tuning compression. An image much larger than its rendered size may waste more bytes than a small quality adjustment can save.
- Choose a mode based on the requirement. Use lossy encoding if visual change is acceptable; choose a lossless path when exact pixel reconstruction is needed; use a format and mode that support alpha if the image needs transparency.
- Export a small set of candidates. Try a few quality or effort values rather than relying on one “magic” number. Keep other options—such as chroma sampling, metadata handling, and dimensions—consistent when comparing.
- Check the whole delivery result. Compare file size and appearance at the intended rendered size, then zoom in on likely trouble spots: text, fine texture, edges, gradients, and transparency. If you use an objective image metric, name it and apply it consistently; a metric such as SSIM does not replace checking the image for its intended use.
- Choose the smallest acceptable candidate. Repeat the comparison for representative image types instead of assuming that one result applies to every asset.
How much smaller might WebP be?
Google’s WebP overview, last updated in 2025, reports WebP images as 25–34% smaller than comparable JPEG images at equivalent SSIM quality, and WebP lossless images as 26% smaller than PNG. These are Google’s published comparisons, not savings guaranteed for an individual image. The result for a particular asset depends on its content, encoder, settings, and comparison method.
Google’s separate WebP comparison study reports average file-size reductions of 25–34% at equivalent SSIM across its named datasets. The study identifies libwebp 0.1.2, released in Q1 2011, and libjpeg 6b; its encoder versions are historical, so it should not be read as a current benchmark. The overview figures and study results are not a substitute for comparing your own images.
Decide beyond the quality number
File size is only one part of the choice. Consider visible fidelity and artifact character, whether exact pixels or transparency are required, how long encoding takes, the dimensions and display size, metadata and color-profile needs, and support in the target environment. A candidate is useful only if it meets all the delivery requirements—not just if it has the smallest file or the highest quality number.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
