I tried a few simple filters and left my notes in the readme for how each worked or didn’t work
-
-
এই থ্রেডটি দেখানধন্যবাদ। আপনার সময়রেখাকে আরো ভালো করে তুলতে টুইটার এটিকে ব্যবহার করবে। পূর্বাবস্থায়পূর্বাবস্থায়
-
-
-
This. I like this. Nabbed.
-
Do you need 0.5GB of image data though? :-D
-
I am only complaining because my MacBook Pro literally has 1.5GB of SSD available and checking this in to my Perforce depot is going to make me actually have to clean house. ^_^
-
You could delete the images before upload I guess they aren’t needed for it to work
-
I will delete the images tonight so others won’t run into this issue
-
Are they used for any test cases? Or just purely example data? This is really a non-complaint; just seems unfortunate to have your nice tidy 0.5kloc code dwarfed by a pile of pixels i won't need. ;)
-
Legitimate complaint. The tester uses them
কথা-বার্তা শেষ
নতুন কথা-বার্তা -
-
-
Did quick test on images you provided. Results are disappointing, only 2,95% saved over PNG, which might be simply because of Zstd: https://file.io/RWfM6g
-
Ah darn need to find a few % somewhere one day :)
-
Maybe I'm doing something wrong? I used ZpngApp.exe on the Braid examples. But they won't get anywhere near 1/3 smaller. My previous link got deleted, here is a new link to the results:https://www.scribd.com/document/378109902/Quick-Zpng-Test …
-
I think because they are small and computer generated PNGCRUSH can do a better job with the small color palette. This is more for larger images
কথা-বার্তা শেষ
নতুন কথা-বার্তা -
-
-
Another suggestion: More color channel decorrelation. Proposal (encoder side): - g' = g [use green as luma approximation] - r' = r - g [red/green balance] - b' = b - avg(r, g) [yellow/blue balance]
-
Tried that one and got worse compression for some reason =(
-
Code here: https://gist.github.com/kajott/6771de6dfca0255039103e9c4c26e034 … (Not a proper codec, just a testbed for experimenting with various decorrelation and prediction schemes.)
-
Will open an issue and check it out
কথা-বার্তা শেষ
নতুন কথা-বার্তা -
-
-
Did you try 4:2:0 YUV subsampling?
-
Sounds lossy!
-
But probably reasonably lossy for photographic images, reasonably fast, and good compression gain!
-
Yeah for lossy compression it is used in just about every format
কথা-বার্তা শেষ
নতুন কথা-বার্তা -
লোড হতে বেশ কিছুক্ষণ সময় নিচ্ছে।
টুইটার তার ক্ষমতার বাইরে চলে গেছে বা কোনো সাময়িক সমস্যার সম্মুখীন হয়েছে আবার চেষ্টা করুন বা আরও তথ্যের জন্য টুইটারের স্থিতি দেখুন।