Very impressive new lossless image compression codec: FLIF

It appears to beat, in compression every other existing lossless image codec in every scenario: wow!
It supports alpha channel, up to 16 bit per channel, is interlaced by default for progressive loading and decoding.


Kind of compression performance to expect:
26% smaller than brute-force crushed PNG files.
35% smaller than typical PNG files.

For lossless, that's huge gains.

#supercurioBlog #image #compression #codec

FLIF – Free Lossless Image Format
FLIF – Free Lossless Image Format. FLIF is a novel lossless image format which outperforms PNG, lossless WebP, lossless BPG and lossless JPEG2000 in terms of compression ratio. According to the compression experiments we have performed, FLIF files are, on average: …

Source post on Google+

Someone made an extended comparison set between Fabrice Bellard new BPG and JPEG (Mozjpeg), JP2K-Kakadu, WebP and Original images

The images are very well chosen and their variety give a pretty great idea of the image codecs performance in practice.
Spoiler: BPG entirely hammers everything else available today, including WebP.

One thing tho:
WebP, if I understand correctly is still based on VP8 and its keyframe image encoding is not that impressive as you can see on Daala comparison here:

Thing is, this second comparison tend to show that VP9 keyframes image codec performs admirably, and looks to me at least as good or better than x265.

Consequently, VP9 surpasses by very far current WebP/VP8.

Now I think that now I want a new WebP based on VP9. It'll be best in class in terms of image quality and not patent-encumbered patents.
Sounds good, right?
Can someone at Google please upgrade WebP? 🙂

#supercurioBlog #image #compression #codec #comparison

BPG Image Comparison
Medium images were first encoded with bpgenc at 19 CRF filesizes. Medium is 60% of Large. Small is 60% of Medium. Tiny is 60% of Small. Everything else was matched to +/- 5% filesize. BPG, JPEG 2000, and WebP decoded in javascript when needed. This page is based on’s Daala comparison …

Source post on Google+

Fantastic Fabrice Bellard released a new image format: BPG

It sports lossless and lossy compression and is available today for the web.

Available in browsers already? How!
Well this comes from the man who wrote a PC Emulator in Javascript 3 years ago 🙂
So he implemented a small Javascript decoder for BPG: problem solved!

My favorite features are that BPG is future-proof

Finally a lossy compression format not limited to 8 bits colors
It matters because it's impossible to represent a smooth gradient on 8 bits without resorting to dithering.
Dithering destroys compression ratios or is destroyed by lossy compression.
Now that we're evolving towards larger color gamuts (wide gamut displays are out here for 5+ years), 10 to 12 bits sampling rate is even more of requirement.

Can be hardware-accelerated
Because it's based on a subset of HEVC video compression, it can be decoded and encoded by dedicated chips available on your next mobile SoC.
Well, hardware HEVC encoding not quite yet but decoding arrives soon.
Today, JPEG encoding and decoding is accelerated system-wise (including for browsers) on every mobile system.
BPG design allows to get the same speed and power efficiency.

My least favorite feature is: oh noes software patents, because HEVC.

I look forward to an updated comparison between BPG and a WebP based on VP9.

Discussion ongoing on Hacker News

#supercurioBlog #image #compression #codec

BPG Image format
BPG Image format. Introduction. BPG (Better Portable Graphics) is a new image format. Its purpose is to replace the JPEG image format when quality or file size is an issue. Its main advantages are: High compression ratio. Files are much smaller than JPEG for similar quality.

Source post on Google+

I'm listening a lot of music from Play Music All Access lately, at the maximum bitrate which is 320kbit/s LAME mp3

And when I stumble upon a song or an album I've been used to listen to in lossless/FLAC, I'm like hmm no that's not quite it.

I wonder if I would be able to discern random samples using the same codec & bitrate and lossless in ABX test.
Maybe it makes a difference when you know a song very well in it's original lossless format already.

I would describe the main difference being in:
– impact of drums, typically snare drums and cymbals
– some flanger effect and loss of temporal resolution in vocals
– loss of resolution in synthetic instruments like those found in electronic music like trance or dubstep, or practically everything shaped like a square or triangle wave.

I still think All Access is probably the best sounding streaming service, and content discovery is great but boy I would like to be able to stream lossless instead. It's not the same experience.

Also I don't recommend to spend days listening to samples of every available audio compression codecs at various bitrates in order to find the transparency sweet spot for them.
Because you can't un-train your ear after that, too late ^^

#supercurioBlog #audio #streaming #compression #codec

Source post on Google+