22
5
I've just been trying to compress an image with the PNG image format and have had varying results from different applications. For reference, the image I am trying to compress is a screenshot of this site. Here's a download of the original image.
Here are some results I have so far in order of best to worse (original image is 239KB):
PNGGauntlet (Xtreme! / None filter) ...................... 141KB
FastStone Capture v5.3 (16M / MAX/ No filter) ............ 166KB
Paint.NET 3.36 (Autodetect) .............................. 185KB
PNGGauntlet (Xtreme! / Auto filter) ...................... 208KB
IrfanView (PNGOUT 2 passes w/default) .................... 211KB
XnView (Export / 9 / Sub filter) ......................... 228KB
Photoshop CS1 (Save for Web / PNG24) ..................... 232KB
IrfanView (Standard save / 9) ............................ 240KB
XnView (Standard save / 9) ............................... 240KB
IrfanView (Save for Web / Best) .......................... 242KB
FastStone Capture v5.3 (16M / MAX/ Paeth filter) ......... 248KB
IrfanView (Save for Web / Optimal 256 / Xiaolin / Best) .. 81KB
Some fiddling with PNGGauntlet led to the best result, although it took ages to compress. FastStone is doing something interesting, as is Paint.NET because they get their results in a fraction of the time PNGGauntlet takes to crank its magic. I've separated those last two results for contrast (see how the Paeth filter seems to be causing a problem). Note also the palletized 256-color version from Irfan: this would normally be the option I would use, but I'm testing TrueColor results here.
Has anyone else got any suggestions that can beat the best results here and can anyone explain what the filters do because they seem to seriously affect the outcome depending on the image type.
Pingo claims to provide vast set of options, from fastest to strongest, hope would be tested: https://css-ig.net/pingo Also there was experimental ZPng - using PNG container with Facebook's Zstd codec. While file is not PNG it is still lossless and claims to get 66% of PNG size with 6% time
– Arioch 'The – 2019-04-26T11:52:50.690Even if the question is asked in a constructive manner (except for the "best" in the title maybe), it unfortunately has generated an open-ended list of answers, which is why I'm closing it now. – slhck – 2013-03-02T22:41:57.340
2For future reference, what would an acceptable form of this question be? For the record, I've found the answers exceptionally useful and constructive and therefore a good fit for this site. I would respectfully say that closing this one, in this particular case, is the wrong judgement. Natural up/down votes work fine here. If another tool comes along, you've denied us the opportunity to learn of it here. Open ended is of benefit here. – Charles Roper – 2013-03-03T10:31:30.297
Since 2009, a few things have changed. Stack Exchange has made it clear that open-ended questions aren't welcome anymore (e.g., "What's the best software for XZY?"), as the format just isn't made for this, and it isn't supposed to be a polling contest, but a Q&A site for actual problems with a closed set of answers. Now, your question isn't bad per se, but we definitely don't want to get a list of all PNG optimization programs out there, which by definition is not constructive. Wikipedia has you covered there.
– slhck – 2013-03-03T10:37:45.5431I understand that this question has been around for a long time and you found the answers useful—also, you've been a member of this community longer than me—but you have to understand that the policies just have changed drastically. As I said, your question isn't bad, you've even asked for an explanation of how the filters work, but it didn't seem to encourage the right kind of answers. – slhck – 2013-03-03T10:41:07.517