AVS Forum banner
Status
Not open for further replies.
1 - 6 of 6 Posts

· Registered
Joined
·
190 Posts
Discussion Starter · #1 ·
So like dozens of other people here I'm playing around with TT2 and the new nvidia codecs and I still don't understand exactly what VPP is. Yes I've read the official thread regarding the Nvidia codecs. In most of these threads people just want to know what settings to use. I'm more interested in exactly what this new filter does and why we should use it (i.e. I'm trying to get an answer that's more detailed than "use it because it looks better" or "use it because everyone uses it "). :D


Does the VPP option in TT2 just provide deinterlacing and colorspace conversion? If so, why do we need to replace the current deinterlacing algorithm in TT2? If the deinterlacing performance in TT2 isn't up to snuff, why not use Dscaler or something else? I understand that the colorspace conversion is useful since I'm using ffdshow (resize in ffdshow is done in YV12). But, is there any other reason I should use the VPP filter in addition to ffdshow?


Btw, I'm running a Radeon. I've heard that the vpp functions work in hardware on the new Nvidia 6600 cards (maybe NOT the 6800?). So, is it the case that one shouldn't bother with VPP if you don't have an Nvidia card?




Edit: ok, so I found this in a post from Karnis.


Video Post Processor

Advanced image processing film mode detection algorithm for better looking video playback. The ForceWare Multimedia video post processor detects, inverts, and reconstructs the 3:2 pulldown of incorrectly encoded DVD's (such as TV shows, documentaries, music videos, cartoons, bonus material, and special features) from NTSC 33ms video back to its original 42ms progressive film source. The result is crisp video playback for your entire DVD collection



Does this mean that for film dvds that we shouldn't even bother with VPP?
 

· Registered
Joined
·
4,720 Posts
Quote:
Originally posted by thollida
So like dozens of other people here I'm playing around with TT2 and the new nvidia codecs and I still don't understand exactly what VPP is. Yes I've read the official thread regarding the Nvidia codecs. In most of these threads people just want to know what settings to use. I'm more interested in exactly what this new filter does and why we should use it (i.e. I'm trying to get an answer that's more detailed than "use it because it looks better" or "use it because everyone uses it "). :D


Does the VPP option in TT2 just provide deinterlacing and colorspace conversion? If so, why do we need to replace the current deinterlacing algorithm in TT2? If the deinterlacing performance in TT2 isn't up to snuff, why not use Dscaler or something else? I understand that the colorspace conversion is useful since I'm using ffdshow (resize in ffdshow is done in YV12). But, is there any other reason I should use the VPP filter in addition to ffdshow?


Btw, I'm running a Radeon. I've heard that the vpp functions work in hardware on the new Nvidia 6600 cards (maybe NOT the 6800?). So, is it the case that one shouldn't bother with VPP if you don't have an Nvidia card?




Edit: ok, so I found this in a post from Karnis.


Video Post Processor

Advanced image processing film mode detection algorithm for better looking video playback. The ForceWare Multimedia video post processor detects, inverts, and reconstructs the 3:2 pulldown of incorrectly encoded DVD's (such as TV shows, documentaries, music videos, cartoons, bonus material, and special features) from NTSC 33ms video back to its original 42ms progressive film source. The result is crisp video playback for your entire DVD collection


Does this mean that for film dvds that we shouldn't even bother with VPP?
NO!


There is a difference between the framing rates of film and video cameras - 24 frames

per second for film and 30 frames per second for video.


A DVD is encoded at 30 frames per second - so there are duplicate frames [ actually

duplicate NTSC fields] thrown in to bring the effective framing rate of film based

material up to 30 fps.


These extra fields can fool some de-interlacers into weaving together parts of two

different film frames into a composite frame. This composite frame is full of aritifacts

because the two fields are from separate pictures.


To aid the de-interlacer in preventing this problem - there is a flag on the DVD which

is supposed to indicate if there are duplicate frames. Unfortunately, this flag is often

set erroneously - indicating the wrong sense.


Karnis is pointing out that VPP makes a determination if there are duplicate frames

[ "3:2 pulldown" ] independent of relying on the flag. So even if the DVD is mis-marked -

which is frequently the case - the DVD will play properly.


As with many functions, there are both hardware and software implementations of the

algorithm. If the card supports the algorithm in hardware - so much the better - since

a hardware implementation usually runs faster than a software implementation.


But just because your card doesn't support a function in hardware doesn't mean that

you don't need the software implementation of the algorithm for the best possible

playback. It just means that it can be done faster on certain video cards - and the

algorithm's work is done by the card's GPU instead of the system CPU. [ Depending

on what other algorithms are running on the CPU - ex. fddshow - one may run out

of CPU processing power. That's why off-loading the work to the GPU is a plus ]
 

· Registered
Joined
·
94 Posts
Quote:
Does it do anything else besides 3:2 pulldown?
There’s also the aforementioned color space conversion to YV12. Also, while the filter doesn’t have a properties page, it seems to have a few 'hidden' options configurable straight from the registry under the following key:

HKEY_LOCAL_MACHINE\\Software\\NVIDIA Corporation\\Filters\\vpp\\

EnableNoiseReduction
EnableInv32Detection
DeinterlaceMode
MarkDeinterlacedFrames


I found these values with Regmon, but haven’t had the time to experiment with their effects (if any).


Cheers.
 

· Registered
Joined
·
190 Posts
Discussion Starter · #5 ·
Quote:
Originally posted by Scotty L
Does it do anything else besides 3:2 pulldown?
Well, the real question is does it handle deinterlacing and 3:2 pulldown BETTER than Theatertek? Right? Personally, I thought the codecs that came with TT already did a pretty good job.
 
1 - 6 of 6 Posts
Status
Not open for further replies.
Top