I've repeated Oleg Kikin's PNG compression tools test with the following competitors:
- Adobe Photoshop CS4
- AdvanceCOMP 1.15
- OptiPNG 0.6.3
- PNGCrush 1.7.6
- PNGOut r20091108
I've made two tests. First, with a sprite image (26 x 2532 pixels) with 70 icons in it (used in real life in one of my previous project) and second, with a set of those 70 icons in separated files.
Here are the results:
| Sprite | Improvement | Separated files | Improvement | |
|---|---|---|---|---|
| PS CS4 (GIF) | 10253 | -27% | 14534 | 39% |
| PS CS4 (PNG) | 7522 | — | 20171 | — |
| AdvanceCOMP from PNG | 7224 | 4% | 17360 | 16% |
| OptiPNG from PNG | 7510 | 0% | 17413 | 16% |
| OptiPNG from GIF | 7522 | 0% | 19869 | 2% |
| PNGCrush from PNG | 7513 | 0% | 19901 | 1% |
| PNGOut from PNG | 6596 | 14% | 16318 | 24% |
| PNGOut from GIF | 6591 | 14% | 16308 | 24% |
So, my result is, if you want to serve small files, you should use GIFs. Why? Maybe because of the header size? Maybe the LZW compression is better with small amount of data than DEFLATE…
My next test will be using data:uri with these files using base64 serving using no encodings and gzip, and will see the results.
No comments:
Post a Comment