Archival of Dark_Shikari's blog?

[Blue_MiSfit]

Anyone have a link to an archived copy of our old friend Dark_Shikari's blog "Diary of an x264 developer"? It was super insightful and I'd like to refer back to some older posts but the site is down and I can't find it on the wayback machine.

[sneaker_ger]

https://web.archive.org/web/20150416142328/http://x264dev.multimedia.cx/
?

[Blue_MiSfit]

Thank you! No idea why it didn't work for me the other day.

[Blue_MiSfit]

This was _SUCH_ a good post on encoder latency.

https://web.archive.org/web/20120829090519/http://x264dev.multimedia.cx/archives/249

[benwaggoner]

Lots of great posts in there. Probably some of the best educational material on real-world encoder design and implementation out there.

[Blue_MiSfit]

Another great one, this time about VP8

https://web.archive.org/web/20120801022030/http://x264dev.multimedia.cx/archives/486


6. Not supporting interlacing.

THANK YOU THANK YOU THANK YOU THANK YOU THANK YOU THANK YOU THANK YOU.

Interlacing was the scourge of H.264. It weaseled its way into every nook and cranny of the spec, making every decoder a thousand lines longer. H.264 even included a highly complicated — and effective — dedicated interlaced coding scheme, MBAFF. The mere existence of MBAFF, despite its usefulness for broadcasters and others still stuck in the analog age with their 1080i, 576i , and 480i content, was a blight upon the video format.

VP8 has once and for all avoided it.

And if anyone suggests adding interlaced support to the experimental VP8 branch, find a straightjacket and padded cell for them before they cause any real damage.


:D

Here we are today with kludgy interlacing support in HEVC :devil:

[manolito]

Sorry I don't remember which one of you encoding gurus posted this:
Foolproof VFR to CFR conversion
If no frame rate is available, or even if one is but it's nonstandard, I'd just open with DSS's convertfps=true with exactly double whatever the final framerate of the DVD is, weave the fields, and encode interlaced. That handles all VFR as optimally as possible without either blending or mo-comp.

More advanced analysis could reveal where progressive with soft pulldown is possible, but you have to parse the timecodes for that and pass the ranges on. Encoding interlaced just works.

So is it Encoding interlaced just works, or should should it be avoided by all means?

//EDIT//
It was foxyshadis who posted this...

[FranceBB]

Lots of great posts in there. Probably some of the best educational material on real-world encoder design and implementation out there.

Indeed! Most of the stuff that Dark was talking about 11 years ago are still very relevant today and offer a very important insight on how things work and why they were implemented the way they have been.

[benwaggoner]

Here we are today with kludgy interlacing support in HEVC :devil:
Victory!

The thing that kills me is that we almost got interlaced kicked out of ATSC 1.0 and could have been rid of it a couple of decades earlier. Intel had convinced everyone that doing SW deinterlacing was too hard for PC clients. The broadcasters were about to give up when Intel said "no, we figured it out." If they'd just shut up for another month...

[cogman]

What ever happened to Dark_Shikari?

[Andouille]

What ever happened to Dark_Shikari?

I think she works now out of the spotlights at a computer company with a fixed monthly income which is obviously welcome for anybody.
We don't either hear a lot about pengvado or other video compression legends.

[Blue_MiSfit]

Indeed :( If y'all ever read this, we miss you both!

[jpsdr]

Unless he changed sex, as far as i remember, Dark Shiraki is a "he".

[videoh]

He did change sex and is now Fiona.

[jpsdr]

Ah...
Ok, i didn't know, my mistake and apologies in that case, and thanks for the information.

[imcold]

We don't either hear a lot about pengvado or other video compression legends.

Skal works on compression project(s) at Google.

[chainring]

Skal works on compression project(s) at Google.

Went by "akupenguin" or something similar?

[imcold]

That was a reference to the "other video compression legends" part. Skal contributed to XviD.

[chainring]

That was a reference to the "other video compression legends" part. Skal contributed to XviD.Oops, yeah, I'm thinking of pengvado from the post one up from yours.

[TEB]

didnt some of old timers go to ateme?

[benwaggoner]

Here's another interesting one in retrospect: https://web.archive.org/web/20141128135416/http://x264dev.multimedia.cx/archives/360#more-360

An early HEVC test encoder took 8 hours to encode 10 frames on a then high end system!

Things have come a long way.

[FranceBB]

An early HEVC test encoder took 8 hours to encode 10 frames on a then high end system!

Things have come a long way.

2014... Yet history is repeating itself 'cause in 2018 it took me more than 2 days to encode a 3 minute test clip (a promo) in 8K on a 16c/32th Xeon in H.266 VVC, which is what we can consider a "then high end system".
On the other hand, the reference encoder has nothing to do with x266 and is far less optimized, so perhaps it's not a fair comparison.
Still, it makes us think: it was hard enough for monocore / early dual core CPUs when we switched from Xvid to H.264, it's been hard enough for 4c/8th CPUs when we switched from H.264 to H.265 and it's gonna be hard for 16c/32th CPUs when we're gonna switch to H.266...
Thankfully, though, development went on and the CPU market improved a lot in the past few years and incredibly accelerated thanks to AMD finally being a good rival for Intel after years of nothing/delusions (I'm looking at you 6core Phoenom and 8 core Bulldozer, you were slower than a 4c i5...). Anyway on the professional side, I've just ordered a brand new server with a dual Xeon Platinum for a total of 56c/112th which should be more than enough for the foreseeable future and 8K I hope, but time will tell.

[orion44]

What happened to the guy that hacked ms-mpeg4 v3 and created divx 3.11?

That video codec single-handedly opened the floodgates during the early '00s.

[SeeMoreDigital]

What happened to the guy that hacked ms-mpeg4 v3 and created divx 3.11?

That video codec single-handedly opened the floodgates during the early '00s.
Ah yes... Jérôme Rota (https://en.wikipedia.org/wiki/J%C3%A9r%C3%B4me_Rota). What a star :)

[FranceBB]

Oh those were the days... Although I was more inclined towards Xvid than DivX and I've been encoding SD stuff in Xvid only during my early days when I was just a student who was translating and subtitling stuff...
Still, I guess that if something like this happened nowadays (I mean reverse-engineering closed source patented technologies, modify them and then make a brand, a company and a business from it, there would be lawsuit after lawsuit... Especially if the "targeted" company was Microsoft).
Still, I wonder what happened to him too...
From a quick search on Wikipedia, it looks like he founded a Company based in San Diego and made business with it, so much so that it has bought Main Concept.
Those fella are very well known in Germany (and not only) for the closed source implementation of popular codecs and for the fact that they claim that their version are more parallelized than the original ones (like the MPEG-2 Encoder vs the FFMpeg one) and more complaint with the professional standards (like Sony / Panasonic XAVC and AVC Intra) in H.264 compared to the original x264 implementation...
So I think (and I'm just speculating) that he's still working there...

[benwaggoner]

I remember DS having written an extensive analysis of MPEG-4 part 2 ASP from her efforts trying to add psychovisual optimization to xvid. It was great! Lots of interesting insights into adaptive quantization, and the limits of the pt2's structure of having a frame QP and then allowing individual macroblocks varying by plus/minus even numbers.

I don't see it in the archive. Anyone remember it? Know where it may live?