Welcome to Doom9's Forum, THE in-place to be for everyone interested in DVD conversion. Before you start posting please read the forum rules. By posting to this forum you agree to abide by the rules. |
3rd February 2013, 10:11 | #17341 | Link | |
Registered Developer
Join Date: Mar 2010
Location: Hamburg/Germany
Posts: 10,346
|
Quote:
People always talk as if 4:2:2 is the worst thing ever, but its already twice as much chroma information as 4:2:0, and converting to that as good as possible is still better then doing a crap job at it.
__________________
LAV Filters - open source ffmpeg based media splitter and decoders Last edited by nevcairiel; 3rd February 2013 at 10:13. |
|
3rd February 2013, 10:26 | #17342 | Link | |
AV heretic
Join Date: Nov 2009
Posts: 422
|
Quote:
I got your point btw. "HQ chroma processing" is really no big deal, correct? C'mon, tell us the truth ) Last edited by Qaq; 3rd February 2013 at 10:50. |
|
3rd February 2013, 11:28 | #17345 | Link | |
Registered User
Join Date: Mar 2002
Location: Sofia, Bulgaria
Posts: 661
|
Quote:
I guess it changed power states and when it was in the P0 it was enough to handle. Just barely but enough. Can you check with GPU-Z if I'm correct? If I'm not asking too much can you also check the GPU and Memory controller load and share your results?
__________________
Z370M Pro4 | i3-8100 | 16GB RAM | 256GB SSD + 40TB NAS NVIDIA GTX 1060 6GB (385.28) | LG OLED65B7V Win 10 64bit 1803 + Zoom Player v14 |
|
3rd February 2013, 14:17 | #17347 | Link | ||
Registered User
Join Date: Jul 2008
Posts: 157
|
Quote:
Quote:
I am guessing if you're on XP(and this isn't directed to you Prinz, but the XP users with the similar problem) and having the problem with stuck windows its a different kind of bug, perhaps as noted related to programs using hardware acceleration. I only used flash as example, because it uses hardware acceleration, but perhaps more what is effecting it for you guys is the animations these programs have with dxva is enabled when closing/opening. Perhaps changing some advance settings around in mad vr can diminish this problem for you? On the subject of YCbCr 4:4:4, aren't Blu Rays and DVDs (& HD-DVDs) encoded in YCbCr? Does this mean, (mad vr aside) that all pc software converts: DVD & HDDVD YCbCr 4:2:0 too RGB Blu Ray YCbCr 4:2:2 or 4:4:4 too RGB And if you set your video driver to YCbCr: DVD & HDDVD YCbCr 4:2:0 too RGB too (video driver)> YCbCr Blu Ray YCbCr 4:2:2 or 4:4:4 too RGB too (video driver)> YCbCr So this means on PC, there is no definitive solution, such as standalone blu ray players or even the PS3 to playback formats truly natively, providing of course your display supports it. Last edited by Dodgexander; 3rd February 2013 at 14:21. |
||
3rd February 2013, 14:22 | #17348 | Link | |
Registered Developer
Join Date: Mar 2010
Location: Hamburg/Germany
Posts: 10,346
|
Quote:
Anyhow, your TV displays RGB, so at some point it needs to be converted, might as well be the PC. The only limitation here is that many TVs do their internal processing only in 4:2:2, so if you use any mode that does processing (most do), it'll use 4:2:2 as an intermediate step. PS: You get that wrong in your post, all commercial content is more or less 4:2:0, Blu-rays, (HD-)DVDs, Broadcasts .. only some satellite broadcasts use 4:2:2, but its not that common.
__________________
LAV Filters - open source ffmpeg based media splitter and decoders Last edited by nevcairiel; 3rd February 2013 at 14:24. |
|
3rd February 2013, 14:44 | #17349 | Link | |
Registered User
Join Date: Jul 2008
Posts: 157
|
Quote:
For those confused like me, here is a good article to read. |
|
3rd February 2013, 15:53 | #17350 | Link | |
Registered User
Join Date: Nov 2012
Posts: 3
|
Quote:
4:4:4 is full resolution of both chroma and luma. 4:2:2 is full luma, full vertical chroma resolution, but only half horizontal chroma resolution. 4:2:0 is full luma and half chroma resolution. Most videos are 4:2:0 to lower bitrate but the display have to upscale the signal to full resolution 4:4:4 to display it. Last edited by Zarper; 3rd February 2013 at 15:56. |
|
3rd February 2013, 16:19 | #17351 | Link | |
Registered User
Join Date: Jul 2008
Posts: 157
|
Quote:
|
|
3rd February 2013, 17:15 | #17353 | Link | |
Registered User
Join Date: Mar 2006
Posts: 58
|
Quote:
Thanks for the reminder |
|
3rd February 2013, 17:39 | #17354 | Link | |
Registered User
Join Date: Mar 2006
Posts: 58
|
Quote:
"Whats the difference between setting YCbCr and RGB and which should be used with MadVR?" as relates to the Pixel Color Format setting in CCC. You can have it set to YCbCr 4:4:4 and still get a picture using madVR, but it might not (probably won't?) look very good. That was my experience anyway, after updating the CCC + drivers. Sometimes AMD/ATI updates reset the Pixel Color Format, which is annoying. I don't see a setting in madVR related to RGB levels OTHER THAN the setting I named. And that is how madVR outputs to the display device, unless I'm seriously confused (a distinct possibility). So, where in madVR, other than that option, can I set some RGB option? In LAV Video I have it set to "PC (0-255)" for "RGB Output levels (for YUV -> RGB conversion)". |
|
3rd February 2013, 18:36 | #17355 | Link | |||
Registered User
Join Date: Jul 2008
Posts: 157
|
Quote:
Quote:
When you mentioned setting madVR to expect YCbCr, you meant just the video levels of 0-255? There isn't a way to tell MadVR to use YCbCr over RGB, which is why your post confused me, I was wondering if I was missing something! Quote:
Why have extra colour spaces when theres no source material to take advantage of it. For a PC I understand why this conversion takes place, but why on a PS3 or Blu Ray player? I thought "deep colour" and "extra white and black" was a feature of HD and that it would be utilized when playing back a BRD. So it turns out the Colour is just the same quality as DVD?? |
|||
3rd February 2013, 19:21 | #17357 | Link | ||
Registered User
Join Date: Mar 2006
Posts: 58
|
Quote:
Quote:
"First the video needs to be converted from 4:2:0 to 4:2:2, then to 4:4:4, then to RGB, and finally it can be fed to the display controller. This is the same process no matter what display technology is being used, whether LCD, DLP, plasma, or CRT." ... ...and on from there. So, if I understand that section correctly, 4:4:4 (or even RGB?) from the player *may* be better than having the display do the conversion. And from what others have said over the past day or so, displays aren't so great at that. So am I correct if I say... if one sets their video card to the Pixel Color Format (AMD name) RGB 4:4:4 (Full RGB), the card is doing the conversion and outputting the converted signal to the display, and the display is not doing any conversion. In the end, I thought this paragraph summed it up pretty well: "Unfortunately device makers tend not to reveal the exact processing steps they use internally, or the algorithms they use to convert various color spaces to RGB. Some use different algorithms depending on which color space is fed in. The bottom line is to assume nothing, and test every combination." So I tested and settled on what looks best. Setting my TV's HDMI input to the "PC" label disables too many options (ratios, as well as Advanced stuff), and I couldn't see a difference, at least when watching US cable TV. Sorry if this is hijacking the thread. Thanks for indulging. I'll shutup now |
||
4th February 2013, 02:18 | #17358 | Link | ||||
Registered User
Join Date: Apr 2009
Posts: 1,019
|
Quote:
The game/graphics modes still display 24p correctly, but disable MotionFlow interpolation (which is very good at reducing judder but not introducing a sped-up look) and put the set into a lower contrast mode that is optimized for latency rather than image quality. The local dimming no longer turns backlight zones off, and drops contrast to about 3,500:1. To improve latency, it also decouples the backlight array from the LCD panel, with about a frame of latency between them. This is not really a problem at only 3,500:1 contrast, but would have been if it allowed the local dimming zones to be turned off, as it does in the Theater mode. That said, the benefits of madVR are still visible when the set processes in 4:2:2, and you can still tell the difference between a good and bad chroma scaling algorithm (e.g. aliasing, loss of saturation) but it's certainly not as obvious as it is in the Game/Graphics modes. (where I do all my testing/evaluation) Quote:
And PC sources are all RGB native, as are console games. Quote:
With Blu-ray that means a 1920x1080 Luma channel, with a 960x540 Chroma channel. With DVD you have a 720x480 Luma channel and a 360x240 Chroma channel. (or 720x576 and 360x288 if it's PAL) So both being encoded using 4:2:0 does not make them "the same". Quote:
Computers render natively in 0-255 RGB. So by setting your video card to output 0-255 "Full" RGB you avoid any further conversions in your PC. If your output is Full/Limited YCC, or Limited 16-235 RGB, then the video card is performing a conversion on the output - and in most cases, that should be avoided, as video cards typically do not do it with enough precision. Once it ends up at your display, you have no idea what it is doing with the image really. But typically displays will be using more internal precision than your video card does when performing these conversions, and there's nothing you can do about it anyway. What seems to happen these days, is that most displays convert everything to 4:2:2 YCC internally for processing (grayscale, gamma, CMS etc) which then must end up as RGB because pretty much all displays are made up of RGB subpixels. |
||||
4th February 2013, 08:51 | #17359 | Link | |
AV heretic
Join Date: Nov 2009
Posts: 422
|
Quote:
|
|
4th February 2013, 11:19 | #17360 | Link | |
Registered User
Join Date: Apr 2009
Posts: 1,019
|
Quote:
It does negate some of the benefits of madVR, but doesn't eliminate them. If you have plans on using a PC hooked up to the display for other reasons though (e.g. browsing the web, gaming) or connecting up video games consoles, I would not want a display that can't display RGB at full resolution, because those sources are RGB native. I tried taking some photographs to illustrate the differences between various scaling algorithms when displayed at 4:2:2 and 4:4:4, but unfortunately they don't show it as clearly as I had hoped. What I did find though, was that with my TV operating at 4:2:2 - at least with the sample that I tried - there is zero difference between Jinc 3 AR and Lanczos 3 AR Chroma. They're not just "very close" or "similar" - I can layer the images on top of each other in Photoshop and use the "difference" layer blending tool, and nothing shows up. 4:2:2 chroma really does seem to serve as an "equalizer" and remove a lot of the differences between algorithms. But you can still tell the difference between a soft chroma algorithm and a sharp one. It's more difficult to see differences in aliasing though. Comparing Nearest Neighbor scaling probably shows off the differences between 4:2:2 and 4:4:4 the clearest - it removes a lot of the aliasing. (it's still visible in person - the photo makes it look like it just disappeared) But that's not a good thing - that shows just how much is being thrown away with 4:2:2, and how much it's blurring chroma as a result. I already mentioned that it makes Jinc 3 AR and Lanczos 3 AR look identical, making Jinc pointless for chroma if your set only displays 4:2:2. I wish I had though to test it, but I suspect that there would be no difference between Bicubic 75 AR and Lanczos 3 AR either, even though there are clear differences between the three of those at 4:4:4. When you look closer at Jinc in 4:4:4 vs Jinc in 4:2:2 for example, you notice that it is not as sharp, edges look fuzzy, and the finer lines are not as bright. If you look at the left of the A here, you see that it is not uniformly lit at 4:2:2 These things are much easier to see on the display itself rather than these photographs though. Keep in mind that only chroma scaling is affected - luma scaling should be no different in 4:2:2 or 4:4:4, and that is going to be much more noticeable than chroma scaling anyway. I think this is good news though - it shows that upscaling chroma to 4:4:4/RGB with madVR does indeed make sense. I've seen many people argue that there's no point for a television to display full 4:4:4/RGB resolution at all because video sources are 4:2:0 native anyway - well this clearly shows that there is a benefit to it. That's not to say it's a massive difference and that you should avoid buying a display if it won't pass full RGB resolution, but it's nice to know that it can make a difference. |
|
Tags |
direct compute, dithering, error diffusion, madvr, ngu, nnedi3, quality, renderer, scaling, uhd upscaling, upsampling |
Thread Tools | Search this Thread |
Display Modes | |
|
|