Coming Soon to HD DVD: Silicon Optix HD HQV
by Derek Wilson on February 8, 2007 1:25 AM EST- Posted in
- GPUs
Final Words
This is one of those times summing up is easy: both AMD and NVIDIA fail utterly at every test HD HQV throws at them. Advanced deinterlacing algorithms used on standard definition video are just not present when working with HD video. We are lacking inverse telecine and noise reduction as well. Watching interlaced HD content on a progressive scan monitor using a computer is not going to deliver an optimal experience until AMD and NVIDIA get with the program and start extending their image quality improving algorithms to HD interlaced content.
While watching 1080p content using a modern graphics card will not have many of the issues illuminated by HD HQV, noise reduction would definitely help. With higher resolution content, issues like noise are actually more noticeable and distracting. Ideally, movie studios would perform noise reduction on their content before encoding it for distribution, but this just doesn't happen right now.
With cable and satellite carrying interlaced video, watching TV and recorded content on the computer will be the more important benefactor of high quality deinterlacing support on graphics hardware. Of course, owners of 1080i HD DVDs and Blu-ray movies won't want to be overlooked. It could be argued that good deinterlacing is more important on cable boxes or TVs, but stand alone HD DVD and Blu-ray players are out there that provide higher quality deinterlacing and noise reduction than can currently be achieved on a PC. In fact, Silicon Optix not only makes the HD HQV benchmark, but their HQV line of video processors featured in high end consumer electronics devices is designed to pass all of the tests featured in their benchmark.
We certainly hope AMD and NVIDIA will pay some much needed attention to their HD video processing and address the issues we've seen here. As many of the techniques used in standard definition video processing can be extended to HD processing, we hope that future driver updates will make things more interesting. If and when we see something more interesting happen on PC graphics hardware, we will revisit the issue. We also plan on comparing the PC and consumer electronics hardware in the future as well.
This is one of those times summing up is easy: both AMD and NVIDIA fail utterly at every test HD HQV throws at them. Advanced deinterlacing algorithms used on standard definition video are just not present when working with HD video. We are lacking inverse telecine and noise reduction as well. Watching interlaced HD content on a progressive scan monitor using a computer is not going to deliver an optimal experience until AMD and NVIDIA get with the program and start extending their image quality improving algorithms to HD interlaced content.
While watching 1080p content using a modern graphics card will not have many of the issues illuminated by HD HQV, noise reduction would definitely help. With higher resolution content, issues like noise are actually more noticeable and distracting. Ideally, movie studios would perform noise reduction on their content before encoding it for distribution, but this just doesn't happen right now.
With cable and satellite carrying interlaced video, watching TV and recorded content on the computer will be the more important benefactor of high quality deinterlacing support on graphics hardware. Of course, owners of 1080i HD DVDs and Blu-ray movies won't want to be overlooked. It could be argued that good deinterlacing is more important on cable boxes or TVs, but stand alone HD DVD and Blu-ray players are out there that provide higher quality deinterlacing and noise reduction than can currently be achieved on a PC. In fact, Silicon Optix not only makes the HD HQV benchmark, but their HQV line of video processors featured in high end consumer electronics devices is designed to pass all of the tests featured in their benchmark.
We certainly hope AMD and NVIDIA will pay some much needed attention to their HD video processing and address the issues we've seen here. As many of the techniques used in standard definition video processing can be extended to HD processing, we hope that future driver updates will make things more interesting. If and when we see something more interesting happen on PC graphics hardware, we will revisit the issue. We also plan on comparing the PC and consumer electronics hardware in the future as well.
Facebook
Twitter
RSS
27 Comments
View All Comments
JarredWalton - Thursday, February 8, 2007 - link
*grumble* Should be "we've done HQV...."ShizNet - Friday, February 9, 2007 - link
big part of GPU driver problems are backwards compatibility [GF2-7, Rad.7-X1, DX6-9..];DirectX is totally new beast - why not draw the line and develop drivers from now on for Legacy devices and DX10+ ones?
this will keep old and new drivers in 'good' shape and there's no need for over bloated size files with old junk.
Wwhat - Sunday, February 11, 2007 - link
Since DX10 is vista-only and vista uses a whole new drivermodel it is obvious and inevitable that there are separate drivers developed for post-DX10 heh.So why are you asking for something that everybody already knows is going on and sees happening? Have you not heard about the issues concerning vista and the issues the graphics companies have/had releasing drivers for it?
Plus since ATI-nay-AMD has lots of X1- cards only stuff, it's clear that they also separated their drivers in that sense already.
kilkennycat - Thursday, February 8, 2007 - link
I'm sure that Silicon Optix would only be too happy to quickly develop a hardware HDTV silicon-solution for nVidia and ATi/AMD or their board-partners as manufacturing-option for their graphics cards.. No doubt Silicon Optix developed the HD-HQV tests both to weed out the under-performers AND encourage the widest possible use of their silicon......... Would save nVidia and ATi the bother of even more driver-complication and possible tweaks to their GPU hardware (for mucho, mucho $$) for the few that want the highest-quality HD replication ( regardless of whether the source is 1080p or 1080i or even 720p) from their PCs... The same few would probably be only too willing to shell out the $50 extra or so for the "High-quality-HD" Option-version of their favorite video card.abhaxus - Thursday, February 8, 2007 - link
I use either VLC or DScalar to watch 1080i on my PC. I've got an X800XL so I don't have the ability to use avivo. Would be interested to see how this disc fairs on those two solutions, I've always liked VLC's X method deinterlacing.RamarC - Thursday, February 8, 2007 - link
The testing seemed to focus on de-interlacing issues. HD DVD (and Blu-Ray) are intended to store progressive (non-interlaced) content. Some early titles (and crappy transfers) may be stored as 1080i, but by the middle of this year, 95%+ off all HD titles will be 1080p and de-interlacing will be non-issue.ShizNet - Thursday, February 8, 2007 - link
why focus on interlaced content?______________________________________
can you say TV-broadcasting?
same 95%+ of 'stuff' you'll be watching is TV/Cable/Dish [which are 1080i] and not [HD]DVDs nor IPTV for next 5 years+
even when all TV stations will go digital it's only 540p, don't confuse it up w/ HDTV - 720/1080[i/p]. only BIG ones with deep pockets will go HDTV full time.
autoboy - Thursday, February 8, 2007 - link
You guys are missing the point of this test. Broadcast TV is almost all 1080i content and deinterlacing is very important. The HD-DVD is simply the source of the benchmark but should be able to test the playback capability of PCs for broadcast HD as well as interlaced HD-DVD if it exists. Playing progressive scan images is easy and the only thing that should affect it is the noise reduction which I don't use because it ussually reduces detail.Still...this article left me with more questions than answers.
What decoder did you use for the ATI and Nvidia Tests? Nvidia purevideo decoder or purevideo HD?
Did you turn on Cadence detection on the ATI and Inverse Telecine on the Nvidia card?
What video cards did you use? You ussually use a 7600GT and x1900pro
What drivers did you use?
What player did you use?
Is this test only for HD-DVD decoders or can you use any mpeg2 decoder which would make this a much more relevant test since 1080i HD-DVD is rare and Broadcast HD is what really matters here.
What codec does the HQV use? Mpeg2? VC-1? H.264? Because most VC-1 and H.264 are progressive scan anyway and Nvidia does not claim to support purevideo with anything but mpeg2.
Did you turn on Noise reduction in the Nividia control panel?
Why does Nvidia claim HD Spacial Temporal Deinterlacing, HD Inverse Telecine, HD Noise Redution, etc in thier documentation but cannot do any of the above in reality? Is this h.264 and not supported?
hubajube - Thursday, February 8, 2007 - link
Well this settles whether or not I build an HTPC for HD movie play. This combined with needing a fast CPU (read expensive) as well as a HDCP capable video card pretty much kills a HTPC in the short term. I'll just get a standalone player for now.cjb110 - Thursday, February 8, 2007 - link
Could you get more hddvd players and push them through this test?!?!Also include the DVD results too, as its no good if it can only do one format correctly.
tbh I think it is pretty atrocious that only recently with the Denon 5910 and the Oppo players that we have a dvd player that actual plays dvd's 'properly'.