in five years' time...

[LoRd_MuldeR]

Not in 5, but maybe in 10 years we can start to paint our walls with hi-resolution OLEDs or E-INK instead of having small 50" LCD tvs like now.

However the field of vision is limited by the human eye. So there is some natural limit for screen size, unless you have a cinema hall at home ;)

[Caroliano]

I noted that animes are changing since 2000 for ever smaller characters in the screen, as you can see things more clearly in bigger HDTV screens. Maybe movies will adapt to use most of those monstrous resolutions and sizes. 1080p is far from being more resolution than our eyes can distinguish.

[zeeman_88]

Akupenguin

Thanks for the through answer. I admire your knowledge regarding temporal redundancy research.

I know there are several attempts to use 3D wavelet for exploiting temporal redundancy and I know they haven’t produce better results nor they are practical.

However, I am anticipating one day “in the future” we will see a video compression algorithm that doesn’t use trial and error or macro-blocks coding. Instead it will analyze, process and encode the whole GOP as one data chunk. Even if there is such algorithm today, it would not be practical, at least at the decoder side. However as more cores will be added to the CPU, such algorithms will be more feasible. Until then h.264 and its extensions will still be the video compression champion.

Now regarding h264 intra compression as the best lossy still-image format. I am not aware of that and have not compared it to JPEG2000 myself. Have you run such a test? I would be interested too see these results.

My expectation for such a test is that they would give comparable quality for low to medium compression. But at high compression ratio (80 and above), I doubt h264 Intra will be able to keep up.

:)

[Dark Shikari]

My expectation for such a test is that they would give comparable quality for low to medium compression. But at high compression ratio (80 and above), I doubt h264 Intra will be able to keep up.No, it trashes JPEG-2000 at pretty much any bitrate. This isn't surprising, since JPEG-2000 is a terribly designed format that has singlehandedly tarnished peoples' expectations of wavelet formats for an entire decade.

Its also not surprising since spatial intra prediction is impossible in an overlapped-block format.

[Sagekilla]

YV12 can never have full chroma resolution ;) It's 4:2:0. 10 bit would be a much better improvement IMO - that way we wouldn't have to hide banding with dither/noise.

~MiSfit


Full 4:4:4 would be handy though :)


@Zeeman: We could make x264 encode on a per-GOP basis, with it outputting ridiculously large amalgam of all the frames from the start to end of the GOP, which would constitute one huge data block ;)

[akupenguin]

Now regarding h264 intra compression as the best lossy still-image format. I am not aware of that and have not compared it to JPEG2000 myself. Have you run such a test? I would be interested too see these results.
My expectation for such a test is that they would give comparable quality for low to medium compression. But at high compression ratio (80 and above), I doubt h264 Intra will be able to keep up.

At high resolutions but very low bitrates, jp2k fares better than h264 since jp2k can reasonably efficiently discard whole subbands (which is almost as good as downscaling, though it still wastes a few bits), while h264 block size maxes out at 16x16. Nevertheless, in such extreme cases both jp2k and h264 benefit from downscaling the image and compressing the lower resolution version to the same filesize. And once you do that, h264 wins again. (Granted I had to tweak the resolution manually, as x264 doesn't optimize it for you.)

IOW, yes h264 fails at sufficiently low bpp, but low bpp is always avoidable.

[Dark Shikari]

At high resolutions but very low bitrates, jp2k fares better than h264 since jp2k can reasonably efficiently discard whole subbands (which is almost as good as downscaling, though it still wastes a few bits), while h264 block size maxes out at 16x16. Nevertheless, in such extreme cases both jp2k and h264 benefit from downscaling the image and compressing the lower resolution version to the same filesize. And once you do that, h264 wins again. (Granted I had to tweak the resolution manually, as x264 doesn't optimize it for you.)

IOW, yes h264 fails at sufficiently low bpp, but low bpp is always avoidable.And this is a general problem with the concept of wavelets.

The primary benefit of wavelets is more efficient coding of large-scale detail.

But, in most cases, almost all your bits are spent on small-scale, not large-scale, detail.

But what if you're encoding at a bitrate low enough that you don't store small-scale detail?

... then you might as well downscale anyways, and you lose the wavelet advantage again.

[Blue_MiSfit]

What about a hybrid mode? Wavelets for large-scale detail, and blocks for low-scale detail?

Or have I had too much damned kool-aid?

~Misfit

[*.mp4 guy]

Imo it should be noted that bpp sensitivity is directly related to transform size, big transform = more possible data removal, small transform = limited data removal, and large transforms don't work well for a variety of reasons, this includes wavelets, curvelets, magiclets or any other buzzwordlet transform. Or, put another way, extremely low bpp (extremely low being defined as a case where the compression standard being used does not in anyway utilize a large percentage of the available resolution) is retarded, you should always use a realistic resolution for your bitrate budget, a qcif blurry mess is just as good as a wuxga blury mess, assuming very low and equivelent datarates (better if you account for superior performance of standalone interpolation to wavelet subband interpolation.

@ds
1 what is so horrible about jpeg2k
2 give me an example of a significantly better wavelet image coder from the same time frame.

I'm not saying j2k isn't pretty useless, but I happen to think this is largely because wavelets are over rated.



.
.
.
.
(hooray parenthesis)

[Dark Shikari]

@ds
1 what is so horrible about jpeg2k
2 give me an example of a significantly better wavelet image coder from the same time frame.

I'm not saying j2k isn't pretty useless, but I happen to think this is largely because wavelets are over rated.I'm not very well-versed in JPEG2K, but from those who know more than me about it, I've been told almost universally that it was a badly designed standard--not just that wavelets are problematic, but that the standard was just poor to begin with.

You'd have to talk to someone who's more familiar with the specifics to know exactly why.

[*.mp4 guy]

What about a hybrid mode? Wavelets for large-scale detail, and blocks for low-scale detail?

Or have I had too much damned kool-aid?

~Misfit

That would unavoidably end up being an overcomplete representation (you have redundancy). You would also get all of the problems stacked together, wavelet ringing, dct ringing, blocking. Imo you would be better off with a hybrid wavelet, that used all of the dct basis functions as subbands (dct's can be described as wavelets), but then extend them to lower frequency ranges instead of ending abruptly. This would probably be very slow.

[Blue_MiSfit]

Ah, but of course slow is a relative term.

Computer power always increases ;)

~MiSfit

[zeeman_88]

On my spare time, I will run some test to compare them both, and I let you guys know my findings …

I think h264 Intra is faster and easier to implement, but wavelet methods like J2K has couple more advantage besides the one you mentioned (automatic downscaling which is why
J2K uses 4:4:4 mostly which lead to a better color quality at all rates).

Some of these advantages like, produce more acceptable artifacts (blurry versus blocky), Scalable streams, and may be I can add rate control is really easy.

[Dark Shikari]

That would unavoidably end up being an overcomplete representation (you have redundancy). You would also get all of the problems stacked together, wavelet ringing, dct ringing, blocking. Imo you would be better off with a hybrid wavelet, that used all of the dct basis functions as subbands (dct's can be described as wavelets), but then extend them to lower frequency ranges instead of ending abruptly. This would probably be very slow.Or you can just use a hierarchical lapped DCT, which allows you to in a sense get the best of both worlds.

[akupenguin]

automatic downscaling which is why J2K uses 4:4:4 mostly which lead to a better color quality at all rates
What does chroma subsampling have to do with bitstream peeling? If 4:2:0 is good for 720p, and 4:2:0 is good for 360p, then 4:2:0 is good for 720p with the HF subband removed. Alternately, you could peel luma and leave all the chroma, giving 360p 4:4:4. Or you could take a 4:4:4 input and peel it to 4:2:0. They're completely orthogonal choices.

[Shinigami-Sama]

and to continue the tradition
5 years from now we'll still be using 32bit and wondering why
we'll still be using mechanical harddrives and wondering why
x264 will become self aware and take over the world after dark tries to make some sort of predictive adaptive ME
HDTV will account for about 25% of basic TV packages
game consoles will still cost half a month's pay
the music industry wil finally start to realize they can't win after getting sodomized in court with a meathook
hollywood will still be wondering why no-one wants to buy a remake of a remake of a remake for the 3rd time this year...
the abortive x264vfw will still be around
dark will still be addicted to touhou
ATI will still have horrible drivers and support
another DNS hole will be exposed
wavelts will still just be toys to play with
youtube videos will still look like crap
we'll have GPGPU down to something somewhat useable
there will be 1tb laptop drives for under 150$
4gb ram will be standard
people will still complain x264 is to slow at 100fps on 1080p samples using sharktooth's insane profiles on a laptop...
I will still have this avatar
nettops(EEEPCs - similar) will have replaced PCs in coffee shops because they cost less than 100$/each and no-one cares if they loose it

[Quark.Fusion]

JPEG2000 uses Wavelet compression. There are video formats that use Wavelets, such as Dirac and Snow, but they did not prove to be superior to H.264 yet. Maybe they never will...


Also you must ask: Do we really need more efficient video compression than what we've got today?

Discs are becoming bigger and bigger, bandwidth is becoming cheaper and cheaper. Also I'm not sure if we will ever need much higher resolutions than "Full HD", simply because you can't integrate even bigger screens into your living room. Not everybody can have a separate room for TV and Beamer. Also look what happens to audio compression: Now that the HDD's and Flash's in mobile audio players become bigger and bigger, people go away from MP3&Co and go back to uncompresses Wave or FLAC...
I'm still waiting for high DPI displays and maybe movies will be at 4k resolutions.

[Sagekilla]

I want 300 DPI displays personally ;) Our current ones have maybe... 72 DPI at best?

[nm]

I want 300 DPI displays personally ;) Our current ones have maybe... 72 DPI at best?
Many current PC displays have 100 pixels per inch or more. Some laptops have 150 PPI displays, and with money to spend, you could have bought a 200 PPI LCD panel (http://en.wikipedia.org/wiki/IBM_T220/T221_LCD_monitors) over five years ago.

[vmrsss]

this is not directly relevant to x264, hope I can be forgiven for an incidental remark: i'm not quite sure why people as at the beginning of this thread insist that disc space is so cheap to kill dvds and other removable support. At the moment, I can buy a terabyte of reasonably reliable dvds for less that $30. I need twenty times as much for a terabyte of hard disc. Also, I can always buy another dvd and extend the space available to me, while there is a limit to how many discs a desktop will support. Finally, if a dvd breaks, you've lost 4GB, if your hard disc goes, then you've lost it all, which implies you really can't get away without systematic backups.

[Dark Shikari]

this is not directly relevant to x264, hope I can be forgiven for an incidental remark: i'm not quite sure why people as at the beginning of this thread insist that disc space is so cheap to kill dvds and other removable support. At the moment, I can buy a terabyte of reasonably reliable dvds for less that $30.13 cents per DVD-RW? Can you link me to where those are being sold, so I can buy a few thousand and resell them for four times the price? :pI need twenty times as much for a terabyte of hard disc.Uh... $600 for a terabyte hard disk?

[Sagekilla]

vmrsss: Where are you buying from that a 1 TB hard drive costs $600? I can buy one online for close to $100.

Downside to DVDs: After a few years (4-5 of use), will those DVDs still work well? ;) Or will you be able to find them all 200+ of them, especially if you have kids? Also, searching from 200+ discs for your movie is a pain IMO. I like being able to pull up an alphabetical list on my computer that shows 200+ all within the screen space.

Each method has it's ups and downs. DVDs are great if you take good care of them but most people don't.

[Blue_MiSfit]

My sentiments exactly. I did lots of DVD-> MKV burned on DVD backups several years ago, and most of those DVD-Rs now have CRC errors, even though they've been sitting in a binder and very rarely used! It's not a big deal since I can just re-rip from my original discs now (with x264 instead of Xvid - w00t), but it's a hassle!

Windows Home Server presents a compelling solution to those of us who have large, ever-growing media libraries and like to keep everything "always available". Specifically, its abstracted storage model, which allows you to add hard drives, and redistribute free space as necessary - and specify levels of redundancy. It's very cool! In fact, a 4tb WHS box is in my immediate future. That way, my big hungry overclocked Q6600 doesn't have to stay on all night so I can watch my stuff on the laptop while in bed ;)

Oh, and as icing on the cake it serves as a Terminal Services gateway (so you can remote desktop into your machines from anywhere on the intarwebs), and provides shadow copy / backup services as well.

~MiSfit

[Sagekilla]

Indeed, if you're looking to keep long lasting backups of your movies, the best way to go is aim for ~4 GB rips (Very easy for me, mine are regularly 3.5 - 4 GB @ crf 18) on a RAID 0'd data server. If you maintain it well (basically make sure it doesn't get dusty inside) then that data will always be there. If a drive dies on you, you pop in a new one. Or, if you have WHS like Blue_MiSfit said, you just pop in drives as you like and it reconfigures on the fly.

For individual hundreds of DVDs, if a disc fails, yes you just lose one disc but you have to re-rip every time that happens. Over time that can really accumulate, and nothing beats having all your media a click away ;) I rip my movies because I don't want to deal with discs.

[vmrsss]

13 cents per DVD-RW? Can you link me to where those are being sold, so I can buy a few thousand and resell them for four times the price?

Sorry, I meant DVD-R, which I guess we can all buy for about 15 cents?

:pUh... $600 for a terabyte hard disk?

Admittedly I am talking of over one year ago, yet I paid for a lacie firewire 1T disk about £265, which was roughly $530. From what I read from you, probably not the best of choices, or perhaps prices went down quite considerably since...

Certainly if 1TB hard disc can be gotten for about $100, I'll have to reconsider my DVD-R burning strategy... Indeed, as people have suggested, to keep track of hundreds of discs and index their contents in some reasonable and reliable way it is a serious burden a would gladly do without...

About DVDs going wrong after a while, luckily it's never happened to me. It surprises (and worries) me to hear of DVD-R with CRC errors even if never used... In fact, my "hollywood" DVDs are on average older than 6yr, and still going strong. Are they technical different from self-burn discs?

[Dark Shikari]

Sorry, I meant DVD-R, which I guess we can all buy for about 15 cents?Newegg says 35 cents, not counting shipping. And don't suggest a local store to avoid shipping: they're twice the price.
Admittedly I am talking of over one year ago, yet I paid for a lacie firewire 1T disk about £265, which was roughly $530. From what I read from you, probably not the best of choices, or perhaps prices went down quite considerably since...Mine was $220... over half a year ago. And it was an external.

[Blue_MiSfit]

About DVDs going wrong after a while, luckily it's never happened to me. It surprises (and worries) me to hear of DVD-R with CRC errors even if never used... In fact, my "hollywood" DVDs are on average older than 6yr, and still going strong. Are they technical different from self-burn discs?


Yep :) Very much so! Pressed discs are hugely more reliable than burned discs, especially cheap burned discs!!

Here's a Samsung F1, 1TB, for $115 with free shipping..
http://www.newegg.com/Product/Product.aspx?Item=N82E16822152102

~MiSfit

[Sagekilla]

For $115, you can get:
1) 1 TB drive
2) 1.5 TB on ~328 discs if you're not paying shipping @ 0.35/disc.

Now tell me, what's quicker (assuming all the movies are perfectly 4.5 GB): Copying all your data over to a new drive, or backing up 328 DVDs? ;) We might have fast DVD-burners but 328 discs still takes a loooong time.

[*.mp4 guy]

If you consider dud discs, and disc death whith equivelent usage patterns of discs and HD's, I don't see how the dvd's can come out ahead, even assuming everything else is optimal.

[Sagekilla]

On an interesting note, I happen to have one of the infamous IBM "death star" drives running in my system still (Yup, a Deskstar 75GXP don't ask, I don't even know why I do) and it still hasn't died on me some how. I have quite a few DVDs and even with good care they've died over the years. Of course, my Deskstar that still runs is rather anomalous and I don't think that's normal in the bit for a drive with it's reputation ;)

Still, I'd sooner trust my movies on a single drive rather than DVDs. And if people are really concerned with losing data, that's where RAID comes in.

[Soulhunter]

Regarding DVD-Rs vs HDDs: Around 1/4 of my (supposed to be high quality) Verbatim/TJ discs got CRC errors over the years, even the ones I never used! >_< But my HDDs still run fine, and they run 24/7...

http://img507.imageshack.us/img507/1449/pc03sw7.png

[Industrial_One]

Heh! Sept 2003 was the time I first signed up. Oh the memories... [...], wishing videos wouldn't take up so much space... To be honest, I didn't give a **** about digital video to wonder how it'll develop in 5 years time. Them video standards never impressed me up until I discovered H.264, but even that is only an extension and practically nothing new.

I dunno about 5 years, but we may finally step out of the matrix in 10 years time -- ditch the retarded 2D+3Dtime macroblock-based concept and move onto 3D+4Dtime compression where you'll save 20x the space.

The problem is, there is no reliable mathematical definition for depth yet. Even though you look at a 2D image, you can tell how far an object is in the pic, but can't really explain. What is a pixel with more or less depth? Darker? Brighter?

Currently, high-dynamic range videos can be manipulated in intriguing ways using such technology (http://gizmodo.com/5038183/photographs-enhance-video-in-absolutely-unbelievable-ways), as well as restoring quality. But it only works with static scenes, which *****.

But as I said, 10 years from now, they'll fix that problem, and you'll be able to convert all your **** low-quality DivX encodes to high quality and bring 700-4700 megs down to 50. So feel free to encode to low bitrates; you can compensate for it later.

As far as the movie industry goes, CGI films will be distributed in closed-source .EXE with its procedurally-generated textures and graphics. Perfect lossless quality and resizable to any resolution. Maybe this way they could utilize a practical lockout chip to prevent piracy like some games do, and we will just use a screen recorder and continue to distribute 'em in crappy, lossy multimedia formats.

And 50 years from now, we'll all have video/audio cards with exabyte-sized libraries and Kolmogorov codecs that'll losslessly compress UHD-quality 3DTV movies down to a couple megs. The final limit. Mark my words.

[Raptus]

I dunno about 5 years, but we may finally step out of the matrix in 10 years time -- ditch the retarded 2D+3Dtime macroblock-based concept and move onto 3D+4Dtime compression where you'll save 20x the space.
Not if we're still encoding 2D material.

But as I said, 10 years from now, they'll fix that problem, and you'll be able to convert all your **** low-quality DivX encodes to high quality and bring 700-4700 megs down to 50. So feel free to encode to low bitrates; you can compensate for it later.
No you can't. You might try to extrapolate some of the lost information to fake some detail, but you'll never be able to restore the original data.
As for huge improvements in compression efficiency, we are actually getting diminishing returns for increasing complexity, so the efficiency curve is asymptotic. But that is nothing to worry about as storage capacities are still growing fast.
Probably the largest gains that still can be made is from a better exploitation of psychovisual effects.

[*.mp4 guy]

There is a lot of redundancy that we juct can't get at the way we are doing things now (in video coding). If we will ever get close to using all of this redundancy is debatable, but it is there, unlike with image compression, where we could thoretically get close to an optimal solution in 5 or 10 years.

[pcordes]

Heh! Sept 2003 was the time I first signed up. Oh the memories... [...], wishing videos wouldn't take up so much space... To be honest, I didn't give a **** about digital video to wonder how it'll develop in 5 years time. Them video standards never impressed me up until I discovered H.264, but even that is only an extension and practically nothing new.

I dunno about 5 years, but we may finally step out of the matrix in 10 years time -- ditch the retarded 2D+3Dtime macroblock-based concept and move onto 3D+4Dtime compression where you'll save 20x the space.

The problem is, there is no reliable mathematical definition for depth yet. Even though you look at a 2D image, you can tell how far an object is in the pic, but can't really explain. What is a pixel with more or less depth? Darker? Brighter?

Currently, high-dynamic range videos can be manipulated in intriguing ways using such technology (http://gizmodo.com/5038183/photographs-enhance-video-in-absolutely-unbelievable-ways), as well as restoring quality. But it only works with static scenes, which ****.

But as I said, 10 years from now, they'll fix that problem, and you'll be able to convert all your **** low-quality DivX encodes to high quality and bring 700-4700 megs down to 50. So feel free to encode to low bitrates; you can compensate for it later.


I wrote a big long reply on the assumption you weren't joking/exaggerating for fun. 'cause it didn't sound that way...

Not without magic. Once you destroy information, it's gone. You might be able to reconstruct 3D from even a blocky or blurry video, but all the textures of everything would still be blurry and artifacted. That link you posted is about adding new information about the original source (the high quality still photo) to each frame of the video. You _need_ the HQ still of the source. The equivalent for a full-length movie would be HQ still pictures of every scene. So if you have that, then you can make a low-qual encode and fix it later. But if you can get a frame of every scene from e.g. a BluRay, why not just rip the BluRay and forget about your divx encode? Even with HQ stills of each scene, it wouldn't work as well for dynamic scenes, because it would have to figure out what was going on and what detail to put where. New stuff that came on screen (e.g. an explosion) wouldn't have any information that could be applied to it.

The need for the HQ still photo is not something that will go away, ever. Information theory says you can't get more information out of something than is there in the first place. Of course, you're arguing that future magic will be able to figure out how all the sharp detailed textures are supposed to look from the less-detailed information left in your video. Not impossible, in theory, if the magic algorithm can assume that it's a video of a real-life 3D scene containing the sorts of things humans usual have around. So if it thinks something is a face, it could make it look like a highly detailed face. But that's just making up details, not recovering the information you threw away in low qual encode.

Anyway, I'm going stand by my claim that there's no way you'll ever get a crappy old 1CD divx to look as good as a 2CD x264 rip, even with whatever post-processing you want on the divx and none on the x264. (I'm talking about a typical feature film here. And I suppose this depends on your definition of good... And of course the postprocessor isn't allowed to have any information from the original source to add detail back in to the video.)


As far as the movie industry goes, CGI films will be distributed in closed-source .EXE with its procedurally-generated textures and graphics. Perfect lossless quality and resizable to any resolution. Maybe this way they could utilize a practical lockout chip to prevent piracy like some games do, and we will just use a screen recorder and continue to distribute 'em in crappy, lossy multimedia formats.

Yeah, video on the Internet (esp. p2p) is going to keep sucking as long as most people don't know they could be making better encodes than the default settings on whatever they use. And various other things like that. [...], and it will continue to be mostly horrible, both low vid qual (mpeg1 and 2 will never die...), and not very erotic: unrealistic with bored-looking actors.

About the only thing with any decent quality is HD movie rips (which is quite a pleasant surprise, considering how crappy quality everyone seems to be willing to accept in everything else). [...], but for anything but the most popular/cult movies, mostly made when the movie came out on DVD. That's ok for current stuff, since x264 is starting to mature, but not many people encode older movies. Although a lot of people make encodes with non-optimal x264 settings. I wish the x264 defaults included maybe 3 b-frames and 2 reference frames. e.g. [...] There is a ton of strobing going on, but the x264 string in the bitstream looks like mostly default settings: ref=1. Which is really bad for videos with e.g. flash photography or stuff like that going on. Not to mention the stage light-show. It did use 2 bframes, which is maybe why it was watchable at all.

(Is this kosher with the forum rules regarding discussion of illegal filesharing? I'm trying to be oblique, and definitely not mention any specific movies. Everything I've said applies equally to Creative-Commons licensed DVD releases, etc.)


And 50 years from now, we'll all have video/audio cards with exabyte-sized libraries and Kolmogorov codecs that'll losslessly compress UHD-quality 3DTV movies down to a couple megs. The final limit. Mark my words.

lossless? Not that small. Noise prevents lossless from being that small. With storage and bandwidth as cheap as they probably will be in 50 years, what's the incentive to make some kind of AI codec that 3D-models the things in the video... (Figuring out what's going on and storing just geometry and textures is probably the most efficient way, but it would require absurd amounts of CPU time, and like I said, probably some AI.)

Technology isn't magic, and it's possible to put some hard limits on what's theoretically possible. I know video is pretty highly redundant, but it's just not _that_ redundant. Even the script (i.e. written by the screenwriter) would be a few percent of a couple MB when compressed with e.g. LZMA or PPM (which I think I read gets close to the entropy limit on most text, i.e. almost 8 bits of information per byte of compressed output).

I could believe a couple GB for UHD, but still not for lossless. There's always going to be too much entropy in the noise. Hopefully by then we'll be using more than 8 bits per channel, so there's more "information" in the noise for the same amount of noise. (noise being random, you can't gain anything from losslessly compressing it.)

[Industrial_One]

Not if we're still encoding 2D material.

I am talking about 2D material. A 2D image has a visual (3D) depth, and by exploiting the redundancies with a codec that can model the objects, you can do way better, save a shitload of space.

As for huge improvements in compression efficiency, we are actually getting diminishing returns for increasing complexity, so the efficiency curve is asymptotic. But that is nothing to worry about as storage capacities are still growing fast.
Probably the largest gains that still can be made is from a better exploitation of psychovisual effects.

That's cuz you are thinking in only one direction, fixated on one concept.

Not without magic. Once you destroy information, it's gone. You might be able to reconstruct 3D from even a blocky or blurry video, but all the textures of everything would still be blurry and artifacted. That link you posted is about adding new information about the original source (the high quality still photo) to each frame of the video. You _need_ the HQ still of the source. The equivalent for a full-length movie would be HQ still pictures of every scene. So if you have that, then you can make a low-qual encode and fix it later. But if you can get a frame of every scene from e.g. a BluRay, why not just rip the BluRay and forget about your divx encode? Even with HQ stills of each scene, it wouldn't work as well for dynamic scenes, because it would have to figure out what was going on and what detail to put where. New stuff that came on screen (e.g. an explosion) wouldn't have any information that could be applied to it.

That's the problem. It needs some smart temporal interpolation modelling scheme in addition to the spacetime-fusion technique or whatever to compensate for the artifacts that would arise in dynamic scenes. However, this will improve in the future. Just think how your own mind works. Can you visualize an HQ version of an LQ video just by looking at it? Would an HQ photo help? Do you now have a general idea of what's infesting the quality? Cool, you should now be able to imagine what the next kickass scene with the nukes would look like.

The need for the HQ still photo is not something that will go away, ever. Information theory says you can't get more information out of something than is there in the first place. Of course, you're arguing that future magic will be able to figure out how all the sharp detailed textures are supposed to look from the less-detailed information left in your video. Not impossible, in theory, if the magic algorithm can assume that it's a video of a real-life 3D scene containing the sorts of things humans usual have around. So if it thinks something is a face, it could make it look like a highly detailed face. But that's just making up details, not recovering the information you threw away in low qual encode.

True. Feeding the app HQ source photos will be mandatory. So you would have to touch the photo up yourself, or hire a photoshop professional. I doubt that having a photo for every scene will be necessary, as there should be some interval of the amount of parameters the program needs to get an idea how to fix the whole video, e.g. you got a sitcom that takes place in only one location throughout the whole series -- you just take photos of a couple scenes that encompass what the general show looks like.

[...]

lossless? Not that small. Noise prevents lossless from being that small. With storage and bandwidth as cheap as they probably will be in 50 years, what's the incentive to make some kind of AI codec that 3D-models the things in the video... (Figuring out what's going on and storing just geometry and textures is probably the most efficient way, but it would require absurd amounts of CPU time, and like I said, probably some AI.)

We won't be shooting film by that time. Everything will be procedurally constructed CGI, hence no noise. And if we wanted some noise into the film for extra texture, that can be stored as a tiny 8-byte pseudorandom seed. Also, it is in everyone's interest for multimedia to be high quality and as small as possible. Stuff that takes days/hours to distribute is not what I call "distributable." The real future of P2P is instant sharing.

I could believe a couple GB for UHD, but still not for lossless. There's always going to be too much entropy in the noise. Hopefully by then we'll be using more than 8 bits per channel, so there's more "information" in the noise for the same amount of noise. (noise being random, you can't gain anything from losslessly compressing it.)

Try 32 bits.

P.S. Notice how I have the class to completely hijack this thread and **** about 10-50 years predictions like some LSD-binging nerd. So I'll get back on topic and give my 10 cents on what I predict in 5 years time:

H.265 will be available, somewhat less dominant than H.264 is today, but popular. People [...] as the 50% efficiency increase will make sub-HD possible at bitrates below the megabit range.

Retards will be releasing encodes with 8-channel 192 KHz audio at like 10 Mbps and brag about their "ultrahigh quality encodes."

Avery Lee will find some time to upgrade Avidemux, turn it into a professional, reliable application like VirtualDub that can actually EDIT content contained in MP4/MKV without crashing/producing corrupt video/errors or otherwise **** up in any way. Then all the [...].

Joost will replace BitTorrent (yo, BT sucks) for video content, and 3rd generation P2P apps will dominate current 2nd gen apps, but will still be popular in 3rd world countries where people can't afford the multiplied bandwidth of their traffic being routed thru proxy peers. Stealthnet and Tesla will be big, Freenet will develop a user-friendly GUI and maybe become more popular.

Ultrareal CGI films will be more common and less expensive to produce. There will be more credible movies than Flatland completely created by one person. Indies in general will be more common, especially with oldschool 8/16-bit games.

Tapestry media and HVDs will quickly replace Blu-rays, and Blu-rays will become the standard casual media format as DVDs are today. Ultrahigh-definition will be prominent, but won't replace HD. HD will be more common as machines get more powerful and make processing high resolutions a casual activity. SD will still be popular as 720x576 for all intents and purposes is enough, but any lower-res rips will be rare.

These faulty [...] LCDs will start to be replaced by LEP displays (that you can fold like paper) which take way less power, have a way higher contrast ratio and don't distort from other angles. Virtual retinal displays will be developed and finalized, but not yet affordable or widely available.

[...]

And I'll be in Japan makin' phat cash.

[/10cents]

[squid_80]

Avery Lee will find some time to upgrade Avidemux, turn it into a professional, reliable application like VirtualDub that can actually EDIT content contained in MP4/MKV without crashing/producing corrupt video/errors or otherwise **** up in any way.Not a chance in hell of this happening.

[LoRd_MuldeR]

Avery Lee will find some time to upgrade Avidemux, turn it into a professional, reliable application like VirtualDub that can actually EDIT content contained in MP4/MKV without crashing/producing corrupt video/errors or otherwise **** up in any way.

VirtualDub is still stuck in the VFW era of the early 1990's and you call it a "professional" application? :D

If Avery Lee had some free time to spend, I think he'd better kick all the VfW code out of VirtualDub and add support for encoder plugins. Allowing decoder plugins already was a first step into the right direction ;)

And before you further bash on Avidemux, please name us one fully-fledged OpenSource video editor with fully working H.264 + MP4/MKV support or what you have contributed to create one :rolleyes:

[Industrial_One]

VirtualDub is still stuck in the VFW era of the early 1990's and you call it a "professional" application? :D

If Avery Lee had some free time to spend, I think he'd better kick all the VfW code out of VirtualDub and add support for encoder plugins. Allowing decoder plugins already was a first step into the right direction ;)

I should've said: There will be a professional, user-friendly app that will let us edit MP4/MKV streams as easy as AVI today.

And before you further bash on Avidemux, please name us one fully-fledged OpenSource video editor with fully working H.264 + MP4/MKV support

There is none, that's the whole problem. But I'm not gonna turn this into a VfW lecture [...] and I grow weary of wasting my time educating them of the benefits of x264vfw.

what you have contributed to create one :rolleyes:

[...]

Now give me one good reason why I should ever spend a second programming an MKV editor for free.

[squid_80]

I don't even know why I still post here after that.I don't know why you do either, this forum doesn't support any of the things you've supposedly "contributed." Since it seems to have brought you so much grief maybe you should consider a new (legal) hobby.

[LoRd_MuldeR]

Now give me one good reason why I should ever spend a second programming an MKV editor for free.

Because you obviously don't hesitate to use existing OpenSource tools for free as long as they serve you well.

Maybe you feel like giving something back to the community one day? :confused:

(If we exclude all the software that was developed and contributed by people for FREE, the video world would be a really sad place - no Avisynth, no ffmpeg/libavcodec, no x264, no LAME, no Xvid, no ffdshow, no VirtualDub, no Avidemux, no MPlayer, no Media Player Classic, no VLC Player, no DG(AVC)Index - just to name a few)

[Sharktooth]

deleted by mod
so, your contribution to the "community" was spreading illegal material all over the world along with your stupidity. what can you expect from an illegal activity?
get a life and since you're so unrespectful of other ppl (legal) work, please REFRAIN to use ANY opensource software. if you want one, make your own... and ensure it will NOT be opensource.
your second option is to pay for softwares that do what you want... if they will ever exist (good luck!!!).
remember: NO ONE OWE YOU ANYTHING. that's a lesson i hope you will learn...

[Industrial_One]

Since this is originally a thread where we smoke chronicles/guess the future and is going off-topic, don't reply to this post, just read.

Because you obviously don't hesitate to use existing OpenSource tools for free as long as they serve you well.

Maybe you feel like giving something back to the community one day? :confused:

[...]

so, your contribution to the "community" was spreading illegal material all over the world along with your stupidity. what can you expect from an illegal activity?
get a life and since you're so unrespectful of other ppl (legal) work

[...]

please REFRAIN to use ANY opensource software.

Ok, can I still use closed-source freeware? ¯_¯

[foxyshadis]

Way off any kind of topic, devolving into a flamewar laced with profanity, and discussion way out of line for this forum. This is supposed to be a professional forum, act with a professional decorum. Closed.