Image compression results depend on the source image, its format, and the settings used. A photograph, a logo, and a detailed scene do not necessarily respond to compression in the same way.
To show what that means in practice, we compressed six image files with Optimize My Image: two PNG files, two WebP files, and two JPEG files. We then measured the original and compressed files rather than relying only on the rounded size labels shown in the interface.
The results below are from this specific test set. They are examples, not a promise that every image will achieve the same reduction.
Test results at a glance
| Format | Test image | Original size | Compressed size | Reduction | Dimensions |
|---|---|---|---|---|---|
| PNG | Frappe drink | 4.05 MiB | 637.6 KiB | 84.62% | 4147 × 4135 |
| PNG | Phone Store logo | 389.7 KiB | 164.7 KiB | 57.75% | 5000 × 5000 |
| WebP | Dune scene | 6.34 MiB | 170.7 KiB | 97.37% | 3840 × 2160 |
| WebP | Dragonfly | 13.14 MiB | 382.2 KiB | 97.16% | 9999 × 6666 |
| JPEG | Flower | 1.09 MiB | 432.5 KiB | 61.31% | 6023 × 4015 |
| JPEG | Bird | 3.54 MiB | 1.35 MiB | 61.91% | 6000 × 4000 |
Sizes are shown using binary units (1 MiB = 1,048,576 bytes; 1 KiB = 1,024 bytes). Percentages are calculated from the exact file byte counts and rounded to two decimal places.
How we ran the tests
We used six original files and their corresponding compressed outputs from the image compressor. Each output was checked for its file format and pixel dimensions.
The captured settings for these runs were:
- PNG: Balanced profile.
- WebP: 80% image quality.
- JPEG: 80% image quality.
The PNG Balanced profile uses a 192-color palette. The JPEG and WebP tests used the 80% quality setting shown in the compressor interface.
We calculated file-size reduction with this formula:
Reduction (%) = ((Original bytes − Compressed bytes) ÷ Original bytes) × 100
For example, the Frappe PNG started at 4,245,632 bytes and the compressed output was 652,867 bytes:
((4,245,632 − 652,867) ÷ 4,245,632) × 100 = 84.62%
The dimensions stayed the same for all six pairs. These tests therefore measure file-size changes from compression, not savings from resizing.
What the results tell us
PNG: graphics and images can respond differently
The Frappe image was reduced by 84.62%, while the Phone Store logo was reduced by 57.75%. Both used the Balanced PNG profile, but their source content was different.
PNG palette optimization reduces the number of colors used to represent an image. This can work well for some images, but it may affect gradients, shadows, color transitions, and partially transparent edges.
The compressed PNG files in this test use indexed palette color. Transparency is retained, but the compressed files do not preserve every original alpha-channel level exactly. For logos, icons, and graphics with soft edges, inspect the output against its intended background before publishing.
WebP: large reductions in these two samples
The Dune scene was reduced by 97.37%, and the Dragonfly image by 97.16%, both at the captured 80% quality setting.
These are substantial reductions for this particular pair of source files. They should not be treated as typical or guaranteed WebP results. The source files were relatively large, and different WebP images can produce very different outcomes depending on their existing encoding and visual content.
JPEG: both samples were reduced by about 61%
The Flower image was reduced by 61.31%, and the Bird image by 61.91%.
The two results are close, but two samples are not enough to establish an expected reduction for all JPEG photographs. Fine detail, noise, texture, existing compression, and image dimensions can all influence the result.
Did compression change the image dimensions?
No. The original and compressed dimensions matched in all six tests:
- PNG Frappe: 4147 × 4135 before and after.
- PNG Phone Store logo: 5000 × 5000 before and after.
- WebP Dune: 3840 × 2160 before and after.
- WebP Dragonfly: 9999 × 6666 before and after.
- JPEG Flower: 6023 × 4015 before and after.
- JPEG Bird: 6000 × 4000 before and after.
This distinction matters: compression changes how image data is stored, while resizing changes the number of pixels. If an image is much larger than its display area requires, resizing may reduce the file further, but that is a separate optimization step.
How to evaluate an image after compression
A smaller file is useful only if the resulting image still meets your needs. Before replacing an original file:
- Compare the original and compressed versions side by side.
- View the image at the size it will actually appear on your website.
- Zoom in on text, edges, fine textures, gradients, and faces.
- For PNG files, check transparent edges against the background where the image will be used.
- Check the final dimensions and file format.
- Keep the original file so you can try a different setting if necessary.
The preview comparison is a practical visual check, not a formal perceptual-quality score. These six tests did not use a standardized image-quality metric or a controlled user study.
Why your results may be different
Your results can vary based on:
- The image's content, colors, texture, and detail.
- The original file's compression and encoding.
- The selected quality or PNG palette profile.
- The image dimensions and color information.
- The browser's image decoding and encoding implementation.
An image that is already well optimized may show only a small reduction. In some cases, a new encoded file may even be larger than the source. Compare the result rather than assuming that a particular format or setting will always produce a smaller file.
Try the compressor with your own images
You can run your own test with the free online image compressor. Select a JPG, PNG, or WebP file, choose the available setting, and compare the original and compressed sizes before downloading.
Your images are processed in your browser. Keep a copy of the original and review the compressed result before using it on a live website.
Frequently asked questions
Does WebP always reduce file size by more than 90%?
No. The two WebP samples in this test were reduced by more than 97%, but those results depend on the particular source files and settings. Other WebP files may produce much smaller reductions.
Does PNG compression preserve transparency?
The PNG output can retain transparency, but palette-based optimization can change the available alpha levels. Check soft edges and partially transparent areas, especially on logos and interface graphics.
Does image compression reduce dimensions?
Not in these tests. All six compressed files kept the same pixel dimensions as their originals. Resizing is a separate operation.
Is 80% quality the best setting for every JPEG or WebP image?
No. It is one setting to test, not a universal recommendation. Inspect the result and adjust the setting based on the image's content and intended use.
How do I calculate the percentage saved?
Subtract the compressed file size from the original file size, divide the difference by the original file size, and multiply by 100. Use the same unit for both sizes, or calculate from exact bytes.
Final takeaway
These six tests show why image compression should be measured on real files. The reductions ranged from 57.75% to 97.37%, while all six outputs kept their original dimensions. The best result for your own image will depend on its content, format, and settings.
Measure the file-size reduction, inspect the image carefully, and choose the smallest output that still meets your visual requirements.