Image quality in lossy image compression: PIK wins at small file sizes, FLIF at large file sizes https://encode.ru/threads/2814-Psychovisual-analysis-on-modern-lossy-image-codecs?p=54089&viewfull=1#post54089 …
-
-
Interesting, thanks! With d.7, hard to see any diff at 2x. Twitter converts to JPEG unless there's at least 1 transparent px.
-
I still could not run pik. Compiles fine but crashes with illegal instructions in image routines. Has it been tested on macOS?
-
No, we only run it on ubuntu and on recent intel cpus -- which cpu you have? If you'd like some results to look at, I can run them for you.
-
Thanks. Mine had just missed AVX2 and FMA. Planning to compare quality metrics with the modification of PSNR-HVS-M that I use.
-
I'm curious if you have interesting results with eyeballing -- or with your modification of PSNR-HVS-M?
-
In my random tests it correlates with Butteraugli but, of course, not that precise. Hope to find some time this week to do Error(Q) on JPEGs
-
There might be interesting performance middle-ground by taking the color model from butteraugli and then applying an ssim-like engine
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।