The history of doom9 forum?

[shevegen]

Hey guys,

I registered here a long time ago. Back then I had a DVD collection of movies that I thought were important (to me). Some related to reallife too.

I had to do some ripping from DVD to a single file and Doom9 was useful here. I also learned about avisynth which was pretty cool. Back then I was using windows still.

Fast forward some ... 20 years or even more. My use cases have changed. Youtube also changed things a lot - I watch more stuff on youtube these days.

My old video collection is now mostly on external harddiscs, but I also collect fewer and fewer videos these days - lack of time, old age, you name it.

So while I have already become a casual user many years ago, I noticed that the things I "used to know" in regards to video and audio also changed. For instance, virtualdub - it is semi-dormant. Gordian Knot? Is it still active? Avisynth users still alive? I really don't know. I have been following other things.

It has been years since I actually had to encode a video too.

What has changed? Anything noteworthy?

Then there is doom9. As far as I understand, Swede is the current site operator. But that is not doom9 right? What happened to doom9? Did he vanish? Is he alive? What is he doing?

So it is basically just a lot of questions. Perhaps someone is bored and may want to add a bit of information. I am missing about like ... 15 years of context or so, give or take. And I am not super-active on the forum either, to word this nicely.

[Jamaika]

You need to ask yourself if a user has been defined here.
Then ask yourself what that user should be like.
Then, what happened that you weren't on the forum for 15 years and now you're here?
Then, how has the IT job market changed in the age of AI?
Is the slogan "Everyone can be an IT specialist" promoted on YouTube true?

Doom9 reminds me of the story of pirated movies like xvid and divx.
Avisynth should have been abandoned long ago because it's not an editor. It's a miracle it's still around, even with add-ons like these.
Before the pandemic I was interested in a whole range of new codecs. How they work and how they're created. After seven years, I realize I shouldn't care. I should be on a kinder planet but I'm not.
I understand that you are surprised that doom9 did not fail despite Facebook and Twitter.

Despite being on this forum for eight years, there hasn't been "pirated" x266, so-called "better than mainconcept."
Nobody has seen a free ac4 codec either. This philosophy has ceased to matter.
How can you compare an interlaced 320x240 image from the 1990s in old Sony camcorders with 16K 10bit film?
Can you afford the latest Intel processor for video editing on this forum? How many programming languages do you know?

[GeoffreyA]

As a teenager, I started in the DVD/DivX days too, stopped encoding for many years, and came back, finding a metropolis where the village had been.

The tools certainly changed and got better. VirtualDub2 is the successor of VD, which Avery Lee stopped working on years ago. GordianKnot doesn't seem popular any more, Handbrake taking its place. FFmpeg is celebrated, sometimes maligned, and underlies various software; it can do anything except make a cup of tea. AviSynth is alive and kicking with 64-bit power. VapourSynth, a newer frame-server, became associated with the anime community.

In terms of codecs, x264 became the best H.264 encoder, sending MPEG-4 ASP into oblivion. x265, its HEVC successor, became increasingly popular but never quite dethroned its venerable ancestor. In recent years, AV1 saw much interest but struggled, at first, with grain and detail, something afflicting most post-x264 encoders to a degree. There's even VVC, skating on thin ice.

[microchip8]

You need to ask yourself if a user has been defined here.
Then ask yourself what that user should be like.
Then, what happened that you weren't on the forum for 15 years and now you're here?
Then, how has the IT job market changed in the age of AI?
Is the slogan "Everyone can be an IT specialist" promoted on YouTube true?

Doom9 reminds me of the story of pirated movies like xvid and divx.
Avisynth should have been abandoned long ago because it's not an editor. It's a miracle it's still around, even with add-ons like these.
Before the pandemic I was interested in a whole range of new codecs. How they work and how they're created. After seven years, I realize I shouldn't care. I should be on a kinder planet but I'm not.
I understand that you are surprised that doom9 did not fail despite Facebook and Twitter.

Despite being on this forum for eight years, there hasn't been "pirated" x266, so-called "better than mainconcept."
Nobody has seen a free ac4 codec either. This philosophy has ceased to matter.
How can you compare an interlaced 320x240 image from the 1990s in old Sony camcorders with 16K 10bit film?
Can you afford the latest Intel processor for video editing on this forum? How many programming languages do you know?

I'd very much like to try what you're smoking or snorting :D

[Jamaika]

I'd very much like to try what you're smoking or snorting :D
Recently, there's been a lot of talk in my country about visa and espionage scandals in the IT industry. How many visas has my country issued, to whom, for what purpose, at what rates, and whether there were any takers. And honestly, I'm shocked.

[VoodooFX]

Is the slogan "Everyone can be an IT specialist" promoted on YouTube true?

Recently, I saw a video on YouTube where a guy was calling himself an IT specialist because… wait for it… wait… because he knows how to use Google search! :D

[Jamaika]

Of course it's a farce but...
There's advertising, and someone's paying for it. There are job fairs in Central Europe and beyond. Then there are TV programs about IT specialists in Mexico, Guatemala, Costa Rica, and so on, working in global companies that don't pay taxes in the US. "Can do"

[FranceBB]

Avisynth users still alive?
What has changed? Anything noteworthy?

It would be too long to state what has changed in Avisynth, but basically first we got support to more sampling other than YV12 and YUY2 with the introduction of YV16 and YV24 and RGB of course. Then, we went beyond 8bit with 16bit stacked in Dither Tools, a genius idea to go divide the image in MSB (Most Significant Bit) and LSB (Least Significant Bit) so that you had them one stacked on top of the other (which is also why it was called "double height"). Most filters got modified to support that so that you could index an 8bit video, bring everything to 16bit stacked where the first 8bit would be the video and the second 8bit would be completely green, then perform high bit depth filtering (and you could see the other 8bit populated) and finally dither down to 8bit before encoding using the Stucki, Atkinson or Floyd Steinberg error diffusion (or ordered dithering for that matter). Shortly after, the development of 16bit interleaved began as some people thought they could do a better job and make it more optimized than stacked. In the interleaved format you would have double width but instead of being 8bit + 8bit, you would have both interleaved together. Filters got updated again and we ended up in a situation in which some filters only supported old regular 8bit, some other supported 16bit stacked, some other 16bit interleaved and a few of them both. This required more RAM and the first Avisynth 64 builds started popping up, but some filters were only available at 32bit and some people decided to stick with 32bit. MPPipeline came around so that you could assign to each function its own process and therefore mix and match filters (i.e use a 64bit filter followed by a 32bit one followed by a 64bit one etc). Even on a 32bit installation, this was useful if one had PAE and much more RAM as you could assign 2GB per process (i.e per function, so per filter) and not have encodes crash. This all happened up until we got to version 2.6.1, after which things changed with the introduction of Avisynth+ for the future versions and the support for Windows98 and Windows 2000 got dropped (mind you, this was 2016). Stephen (qyot) and Ferenc (pinterf) picked up the project and kept going. Set's multithreading branch got merged and improved and we had SetMTMode() with several different multithreading modes that one could choose as well as the new Prefetch(). In theory, each filter is supposed to manage its own multithreading and create its threadpool, however some filters were not doing that and were single threaded, so the MT would process different frames at the same time and then append everything at the end, however this originally caused some issues with some temporal filters and even with TextSub(), so you could literally choose which multithreading mode you wanted and change it from function to function. After that, we've got proper planar high bit depth, so with ConvertBits() we could go to 10bit, 12bit, 14bit, 16bit planar and filters then started to get changed again to work with planar (i.e normal) high bit depth. We also got a lot more colorspaces with the various high bit depth Planar RGB and of course different sampling in YUV like 4:1:1. Then, we went beyond 16bit planar and into 32bit float to get as much precision as possible and allow for very detailed filtering. After that, we got more resizing kernel as well with the introduction of SinPowerResize() for downscaling which avoids overshooting and the introduction of ringing and even UserDefined2Resize() to allow the user to specify custom settings with very broad control. We even shifted from the old clip properties which were initialized at runtime and were static to the new frame properties so that each frame has its own properties and those can change dynamically mid stream. You can think about this as if a function could pass hints to the function coming after that and tell it what it's dealing with and what it changed without having to find alternative ways around that. Filters eventually started to get changed to support this and we're still in the middle of this transition. During the pandemic the work to merge all the various Avisynth forks once and for all began and this led to the merge of Nekopanda's fork to allow the use of NVIDIA CUDA accelerate processing on some specialized version of some filters with the OnCPU() and OnCUDA() functions. After that, it was the turn of AVXSynth (i.e the Linux fork of Avisynth). Although this was never actually merged into the codebase, the idea was the same: resolve all the problems we had and allow Avisynth to run on different operating systems, which led to the inclusion of Linux support and, later on, MacOS. Not only this allowed people to run this frameserver on different OS for the very first time, but also on different architectures, in fact we now also have ARM64 builds on top of the usual x86 and x86_64 builds, however most of the filters still have to be ported and re-compiled.



It's been a very long journey, but this awesome frameserver just keeps going and it's used by people all around the world. Also, some of the people who were around back then and used it for their own passions and to encode anime as fansubbers grew up, went to university, studied computer science and got a job in the industry and this is why several companies are relying on Avisynth for professional use. Instead of dealing with capped MPEG-2 .ts files from Fuji-TV, Tokyo TV etc, we're now dealing with mov and mxf mezzanine files in Apple ProRes, DNXHQX, XAVC Intra Class etc but we're still using the same concepts.


25 years and Avisynth just keeps on going and it still truly is the best frameserver in the world and a reference point for everyone.

[JarrettH]

I remember when we used to use stuff like custom matrix in XviD, when Real Player and DivX were hot, that GSpot utility, ac3filter, when doom9 cracked blu-ray encryption - it was in the news everywhere - 120 hz monitors are now common, so all that "smooth" video playback technology is no longer needed.

Lately I really enjoy MakeMKV, which modernizes disc ripping by encoding straight to an untouched mkv file, then you can easily work on your encode from there skipping the tedious process of IFO/VOB/glitchy media protection.

I still like visiting here, feels like stepping back in time. That lengthy Star Trek thread here https://forum.doom9.org/showthread.php?t=186417 reminded me of old discussions, which sometimes turned into discoveries :thanks:

I don't do as much encoding these days. The most recent small project was doing Are You Afraid of the Dark. I had to use x265 for the first time because I couldn't get x264 to preserve a reasonable file size while maintaining the heavy grain (I still used a light chroma noise removal to get rid of that ugliness and save some file size). Of course not every episode was interlaced the same - that was fun ;)

[Blue_MiSfit]

I got started here back in like... 2002 or something? I was in high school and I had just bought a DVD drive for my PC and was fascinated by how people on the Internet were able to compress DVDs down to manageable file sizes, like 700 MB to burn on a CD. I dove in, and stumbled my way around, learning old encoding tools like Flask MPEG and the like, orchestrating DivX 3.11 and later 4/5. I finally got familiar with AviSynth and XviD, then eagerly dove into x264 and had a blast chatting with the devs on here and on IRC (we miss you Dark_Shikari!)

It was all a hobby for a long time, until I took a job at VUDU back in 2008. They happened to use some of the cool tools from this world, and had a neat (if primitive by today's standards) toolchain to do their encoding using AviSynth and x264. We were streaming 8 Mbps 1080p x264 to their set-top box! It was awesome :)

This led me eventually to a company called Encompass where I learned lots about broadcast, satellite, cable, and VOD. I spent a long time there before eventually moving to Disney where I've been for almost 10 years, working on internal tools for asset management, content screening, and all kinds of specialized use cases.

doom9 has been a constant in my life for over 20 years, and has been an absolute fountain of knowledge. I love this community, even if I don't do any encoding for fun anymore. I check in basically every day to stay on top of developments in HEVC, AV1, and VVC. This place absolutely kick started my career!

AviSynth was a true love for a long time, and I've used VapourSynth professionally. Most of my work is in FFmpeg these days, though.

[Soulhunter]

It would be too long to state what has changed in Avisynth, but basically first we got support to more sampling other than YV12 and YUY2 with the introduction of YV16 and YV24 and RGB of course. Then, we went beyond 8bit with 16bit stacked in Dither Tools, a genius idea to go divide the image in MSB (Most Significant Bit) and LSB (Least Significant Bit) so that you had them one stacked on top of the other (which is also why it was called "double height"). Most filters got modified to support that so that you could index an 8bit video, bring everything to 16bit stacked where the first 8bit would be the video and the second 8bit would be completely green, then perform high bit depth filtering (and you could see the other 8bit populated) and finally dither down to 8bit before encoding using the Stucki, Atkinson or Floyd Steinberg error diffusion (or ordered dithering for that matter). Shortly after, the development of 16bit interleaved began as some people thought they could do a better job and make it more optimized than stacked. In the interleaved format you would have double width but instead of being 8bit + 8bit, you would have both interleaved together. Filters got updated again and we ended up in a situation in which some filters only supported old regular 8bit, some other supported 16bit stacked, some other 16bit interleaved and a few of them both. This required more RAM and the first Avisynth 64 builds started popping up, but some filters were only available at 32bit and some people decided to stick with 32bit. MPPipeline came around so that you could assign to each function its own process and therefore mix and match filters (i.e use a 64bit filter followed by a 32bit one followed by a 64bit one etc). Even on a 32bit installation, this was useful if one had PAE and much more RAM as you could assign 2GB per process (i.e per function, so per filter) and not have encodes crash. This all happened up until we got to version 2.6.1, after which things changed with the introduction of Avisynth+ for the future versions and the support for Windows98 and Windows 2000 got dropped (mind you, this was 2016). Stephen (qyot) and Ferenc (pinterf) picked up the project and kept going. Set's multithreading branch got merged and improved and we had SetMTMode() with several different multithreading modes that one could choose as well as the new Prefetch(). In theory, each filter is supposed to manage its own multithreading and create its threadpool, however some filters were not doing that and were single threaded, so the MT would process different frames at the same time and then append everything at the end, however this originally caused some issues with some temporal filters and even with TextSub(), so you could literally choose which multithreading mode you wanted and change it from function to function. After that, we've got proper planar high bit depth, so with ConvertBits() we could go to 10bit, 12bit, 14bit, 16bit planar and filters then started to get changed again to work with planar (i.e normal) high bit depth. We also got a lot more colorspaces with the various high bit depth Planar RGB and of course different sampling in YUV like 4:1:1. Then, we went beyond 16bit planar and into 32bit float to get as much precision as possible and allow for very detailed filtering.

Fascinating... Thanks for this write-up!

[Soulhunter]

As old-timers share their stories here, I’ll do the same...

I’ve never told this in full before, partly because it involves some "illegal" activities,
but I think those should be beyond the statute of limitations by now. So here we go!

My father was a movie pirate in the late 1970s and early 1980s. Growing up in the 1950s, he escaped the grim post-war reality by going to the cinema every weekend. He was obsessed with Westerns, a passion that continued into the era of "spaghetti Westerns" like The Good, the Bad and the Ugly and A Fistful of Dollars. In the late 1970s, he befriended someone who worked at a film duplication house, which prepared copies of Hollywood movies for distribution to European cinemas. This connection gave my dad access to private "pre-screenings" of the latest films.

Around the same time, VCRs emerged—not yet VHS, but Betamax. These machines were expensive, costing around 1500-2000 bucks (late 1970s money, so roughly 5k in 2025 USD). Video rental stores didn’t exist yet, and buying a movie on tape could cost over 100 bucks (about 400 USD in 2025 money...). Also video cameras became available. They were bulky, lacked built-in tape recorders or batteries, and required a power outlet and a VCR to record. These cameras cost around 3000-4000 bucks (9k in today's money).

Despite the high investment costs, my dad’s friend at the duplication house saw an opportunity to profit by transferring new cinema releases onto tapes using these new technologies. At the time, it could take a year or more for a movie shown in theaters to become available for purchase (or later, rental). So there was a high demand for "1 year earlier access + half the price of normal retail" copies. The friend acquired a VCR and a video camera to record films, while my dad invested in two VCRs and a stockpile of blank tapes to make copies for sale. Their operation ran smoothly for a couple of years until a customer snitched, leading to a police raid.

My dad’s equipment was confiscated, and he was arrested. Fortunately, the copyright law at that time only covered written works and audio recordings but hadn’t yet been updated to explicitly address private movie copying, as home video was a novel technology. As a result, my dad was found not guilty and got released. It took over a year for him to recover his equipment and the tapes, though the police “forgot” to return a dozen tapes—all of which happened to be porn movies.

After this scare, and with my birth around that time, my dad decided to stay lawfully. But his love for movies never faded, and he shared that passion with me. Every weekend, he’d rent the latest films from the video store, and we’d watch them together. He also took me to the cinema often, despite my mom’s disapproval.

One of my fondest childhood memories is seeing Terminator 2: Judgment Day in theaters at age 10 or 11. I’ll never forget the opening scene: The camera gliding across a ground littered with human skulls, stopping, and the Terminator’s foot stomps one of the skulls, CRUSH!, then panning up to reveal the endoskeleton firing plasma rifles amid a war-torn backdrop, PEW! PEW! BOOM!. My eyes widened, my jaw dropped, and I was hooked until the credits rolled. It was my first “real” big-budget, non-kid-friendly movie, and it left a lasting impression.

Naturally, I inherited my dad’s love for movies and started collecting them as a teenager on VHS. Then, in towards the end of the 90s, DVDs became a thing. As I owned a PC since 1995, I realized digital video on discs should allow for lossless copying, just like I have already done with CDs.

So I saved up for a DVD drive and a new PC. My first DVD was a “flipper” version of Starship Troopers, you had to flip halfway through because dual-layer DVDs weren’t common yet. To my frustration, copying the files directly didn’t work—they wouldn’t play. After researching online, I discovered DVDs were encrypted with CSS. At some time research led me to Doom9, a forum dedicated to digital video, where I dove into the technicalities of cracking encryption and copying DVDs, sparking a new fascination with the technology behind digital video.

Over time, I became more interested in the technical aspects of processing, filtering, and encoding video.

The rest is history, documented here...

[Swede]

Then there is doom9. As far as I understand, Swede is the current site operator. But that is not doom9 right? What happened to doom9? Did he vanish? Is he alive? What is he doing?

Yup, still hosting this. Been searching my old posts and emails trying to find the exact time we moved the forum to my private servers but can't seem to find any date. It must've been somewhere around 2006, 2007 or so, although I hosted the www since 2002 and more or less been the webadmin of it all since then (all of it for free without annoying ads or so).

doom9 is still lurking around but mainly has other priorities IRL nowadays.

I guess all of us 'oldtimers' has shifted our priorities during the last 20 years, myself I've just had my fourth grandchild but, hey, still trying to keep the forum running.

/Swede

[manolito]

I think it's time for a BIG Thank You to Swede, he is doing a phantastic job maintaining this forum... ;):D

Cheers, and to many more years for this place!
manolito

[benwaggoner]

Yeah, it has been a while here? This forum started with all these young cryptopunk pirates, and we've grown up into the graybeards of the digital media industry. Some of the best compressionists I know today started out here ripping DVDs.

I think the biggest legacy of Doom9 might be the community that built around x264, which become the default professional H.264 encoder. So many people trying crazy new parameters, sharing the results, engaging directly with the developers, iterating fast, and working with content like anime that was largely ignored by the professional encoder industry. Every time we see a modern encoder with CRF or equivalent, we see the long legacy of x264 and Doom9. It's crazy how many retroactively "obvious" encoder features were invented for x264. CRF, MBTree, long lookahead rate control, presuming QP and PSNR weren't great quality metrics, trellis entropy coding, and a lot of things I'm forgetting.

My story was kind of funny; I had read Doom9 on and off for several years, but working in the "legitimate" content industry didn't have an account under my own name because I didn't want potential clients to worry I might be less than scrupulous about keeping their content secure. But then I joined Microsoft, who encouraged me to formally join Doom9 to talk about HD-DVD, VC-1, Windows Media, and other things no one cares about anymore. And it was a great community, even when we disagreed or when I got dunked on for working for the "evil empire" as Microsoft was often considered, given many assumed that we were plotting to remove the ability to play back media files without DRM and such.

Being the guy who could answer questions about VC-1 and HD-DVD and to some degree Blu-ray, I was asked to become a Doom9 moderator around...2007 maybe? I'm 13.5 years gone from Microsoft now, but am still here and still deleting crypocurrency spam messages for y'all.

And wow, this will be my 5098th post here.

[benwaggoner]

And wow, this will be my 5098th post here.
And sheesh, I realize now that I only joined Doom9 17 years after I made my first digital video file (in MacroMind Director Animator, November 1989). So it was only two years ago that I'd spent the majority of my career post-Doom9 than pre-Doom9.

The idea of DVD piracy itself would have been so funny the first decade and a half. When CD-ROM first game out, those 650 MB were bigger than most internal hard drives and a single CD-R was maybe $18, so copy protection was largely "like you have anywhere to copy it to, silly head." But in 1994 my video startup bought a 4 GB RAID ($5K for for the drive, $1K for the NuBus SCSI card). We we were able to copy all of Myst or The Journeyman Project to a folder and run in massively faster than anyone else could.

I started the 90's in college with this weird academic obsession with video compression but without anything practical I could do with it; I didn't touch QuickTime until fall 1992 I think, after getting my BA in Neuropsychology. I had written a screenplay with a two page scene of people flirting over compression ratios in anticipation of QuickTime, however ;)! I suppose I was always meant to be a compressionist...

I ended the 90's having become a published author and expert on digital media, having moderated a 3000 attendee session at NAB, Authored the first DVD ever made 100% on a PC, designed the video infrastructure for LEGO Mindstorms 1.0, and having joined Terran Interactive to launch their consulting and services business. Terran made the wonderful Media Cleaner Pro, which was the first good GUI desktop compression tool.

Media 100 bought Terran and then imploded, so I did consulting until joining Microsoft end of 2005. I had my own series of tutorials for video encoding tools, was on retainer for 5 of the 6 major compression tool vendors, and created a number of early HD proofs of concept in 2004-2005. Consulting was fine, then I accidentally got hired by Microsoft.

My wife and kids went away to summer camp for a long weekend, and I had more than a few consecutive hours for myself for the first time in years. So I decided to lean in HARD on self-indulgent fun. Yes, I went on AVS Forum. AmirM, the VP of Microsoft's Digital Media Division, was weirdly active on AVSF, getting into lots of debates with people about VC-1 and HD-DVD. That weekend someone had told him that it was impossible to make a good HD-DVD due to bitrates and other limitations. And I remembered that Apple's lastest update to their DVD creation tool supported making a limited HD-DVD image (basically DVD features but supporting 1080i and 1080p).

So I took some rights-free 1080i source I had from a prior project, encoded it up in ProCoder, made a <650 MB .dmg image of it (so it could be written to CD-R, not requiring a DVD drive), and shared in on AVS Forum saying "see, it's not that hard. Burn the disc yourself and try it out. It'll work in any Mac."

Microsoft started recruiting me and I started within 2-3 months after that. Weird it retrospect that Apple had launched HD DVD support a year before the format itself had launched, but worked out well for me!

At Microsoft I worked on HD-DVD, VC-1, Windows Media, and Silverlight. And after 6+ years of every single project I touched being cancelled, I was done with that.

[Blue_MiSfit]

I had written a screenplay with a two page scene of people flirting over compression ratios in anticipation of QuickTime


Next level, Ben! :D

[Doom9]

Maybe I should have written that book when a deal was offered in the early days. It certainly would've helped preserve some of the history. In the early days, this forum was on ezboard until eventually that became too painful and I went for vbb.
In the early days I used freemail services that eventually went defunct or I migrated away from them. e.g. ifrance. And I don't have any archives, so it's hard to reconstruct what happened when exactly. There was excelland, my first hoster if I recall correctly, and the url being go.to/doom9 and then in September 2000 I first registered the doom9 domain (I think based on some outside encouragements) and made a forum on http://pub28.ezboard.com/bdoom9smpegforum. Oh and all that while I was an exchange student in another in another country, and being abroad meant an analog modem only at home (I had cable before, though with a traffic limit) for the first half of the exchange year so I had to do most things from when I was at school.

So.. as I type this we're just reaching the 25th anniversary of this forum (October 1st 2000 was the launch).

I'm also happy to see that MeGUI is still alive. Who would've thought. I haven't done any video encoding related programming in a looong while, but still contributing to other projects besides my day job (also still have a completely non-IT hobby where I created a tool to make my life a lot easier).

who encouraged me to formally join Doom9 to talk about HD-DVD, VC-1, Windows Media,well, that's interesting because the content industry sure never liked us. Went as far as to have lawyers threaten potential hosts back in the day. Luckily, it never went beyond that.

[SeeMoreDigital]

So.. as I type this we're just reaching the 25th anniversary of this forum (October 1st 2000 was the launch).


Happy anniversary ;)

[Soulhunter]

So.. as I type this we're just reaching the 25th anniversary of this forum (October 1st 2000 was the launch).


Oh wow...
Big Boss is also still around!?
Happy anniversary! And THANK YOU for creating this forum!

[FranceBB]

Happy anniversary. It's so nice to you see you online! :D

[DTL]

The new ideas for second-half of 202x for AVS+ are at least:

1. Continue motion to processing in linear RGB even for HDR. The possible optimal way is usage of FP16 format to save 1/2 of caches and RAM in comparison with float32 and to get 2x more performance in processing more samples with new AVX512_FP16 instructions set. Expected in desktop CPUs to the end of 202x and present in industrial intel CPUs already. With addition of the FP16 float plane formats for data exchange in between filters.

2. Optimize full or parts of AVS+ filtergraph for L2 (L3) CPU cache size for better performance. This requires new filter connections and data transfer methods with single plane of frame and part of plane of frame processing modes and again all (compatible) filters some redesign (as was with transition from 8bit to 2x8 stacked and to planar 16/32 bits).