or Connect
AVS › AVS Forum › Video Components › Home Theater Computers › HD 1080i Test Pattern to determine Vector Adaptive Deinterlacing + others icl. Ticker
New Posts  All Forums:Forum Nav:

HD 1080i Test Pattern to determine Vector Adaptive Deinterlacing + others icl. Ticker - Page 6

post #151 of 283
Thread Starter 
Rsaeire, try ATi AVIVO/X-Code MPEG Dekoder (part of the AVIVO Pack downloadable on AMD Catalyst site). This was the only decoder to play MPEG-2 Slices clean @ VA with your setup (HD3650 PCIe + Win7). H.264 Slices also VA but stuttering.
Overclocking the memory seems only to make sense if CCC only allows Weave + Bob.

as kevinqian wrote:
Quote:
I only changed the memory speed. It's actually 504/1008mhz to be exact. After changing it, the driver automatically enables VA. I have to do this every time after a reboot. With stock speed at 405/810mhz, Dxva Checker confirm that only Bob is enabled for up to 1080i.
post #152 of 283
Quote:
Originally Posted by blaubart View Post

Rsaeire, try ATi AVIVO/X-Code MPEG Dekoder (part of the AVIVO Pack downloadable on AMD Catalyst site). This was the only decoder to play MPEG-2 Slices clean @ VA with your setup (HD3650 PCIe + Win7). H.264 Slices also VA but stuttering.
Overclocking the memory seems only to make sense if CCC only allows Weave + Bob.

as kevinqian wrote:

Thanks for the reply blaubart.

The only download I saw on the site was the AVIVO converter, which I downloaded and installed. I googled "ATi AVIVO/X-Code, ATi X-Code, etc" but all I'm getting is mentions of the encoder/converter. I thought a new ATi decoder may have come with and I tried adding it in MPC HC under "external filters", but there was no new codec available.

In addition, I overclocked my graphics card to 825/1800 (stock = 725/1600), but that didn't work either; I'm still only able to get VA on the "man with the glasses" clips, so it appears that this ATi X-Code MPEG Decoder is my only hope.
post #153 of 283
So your memory is OC to 1800mhz effective? That should be plenty fast. Can you at least select VA in CCC? If DxVA checker crashes, d/l an earlier version. 2.1 should work.
post #154 of 283
Quote:
Originally Posted by kevinqian View Post

So your memory is OC to 1800mhz effective? That should be plenty fast. Can you at least select VA in CCC? If DxVA checker crashes, d/l an earlier version. 2.1 should work.

The graphics card's memory is 800 stock (1600 effective), but I've overclocked it to 900 (1800 effective). I can select all the deinterlacing options in CCC from weave all the way to VA in addition to having the pulldown detection option also; I leave the pulldown detection option ticked and the deinterlacing method set at VA.

I downloaded DXVA checker v2.1.0.0 before you posted and it works successfully. It states the following with my graphics card at stock and overclocked speeds:

ATI Radeon HD 3600 Series

[NV12]
VectorAdaptiveDevice: DXVA1/2, 720x480 / 1280x720 / 1920x1080

DXVA2_DeinterlaceTech_PixelAdaptive: Yes

As far as I can see, this confirms my graphics card supports VA up to 1920 x 1080?
post #155 of 283
Thread Starter 
Rsaeire, "ATi MPEG Video Dekoder" is part of the AVIVO converter package. So you should have installed it already. Remind, the infos I am giving are European infos of people who have Win7, I have XP. So it is a bit hard to transpose all that between different systems.

1) I tried it now in XP and "ATi MPEG Video Dekoder" does not work correct at 1080i in MPC-HC. I think it's the same in Win7.
2) The most used HDTV application in Europe is DVBViewer, and he plays MPEG-2 Slices in VA + "ATi MPEG Video Dekoder" + DXVA perfect. And he is the one the info comes from. Sorry for that, sometimes I forget to regard all the different systems in countries.


Quote:


The graphics card's memory is 800 stock (1600 effective), but I've overclocked it to 900 (1800 effective).

If 1080i is already active in DXVAChecker at 800 stock it may be useless to overclock.
post #156 of 283
According to DXVA checker you already have VA activated. Now run the cheese slice and post a screenshot. You may have VA but just not seeing it with trained eyes. Make sure to take sthe screenshot while the top graphs are moving initially.

You can also run cheese slices within DXVA checker. Click Directshow filters button, then select the video file and then click on the MS DTV-DVD decoder>Play>EVR. If the player shows VectorAdaptive, then it's being used.
post #157 of 283
Quote:
Originally Posted by blaubart View Post

Rsaeire, "ATi MPEG Video Dekoder" is part of the AVIVO converter package. So you should have installed it already. Remind, the infos I am giving are European infos of people who have Win7, I have XP. So it is a bit hard to transpose all that between different systems.

1) I tried it now in XP and "ATi MPEG Video Dekoder" does not work correct at 1080i in MPC-HC. I think it's the same in Win7.
2) The most used HDTV application in Europe is DVBViewer, and he plays MPEG-2 Slices in VA + "ATi MPEG Video Dekoder" + DXVA perfect. And he is the one the info comes from. Sorry for that, sometimes I forget to regard all the different systems in countries.



If 1080i is already active in DXVAChecker at 800 stock it may be useless to overclock.

That's no problem. I couldn't get DXVA checker to install earlier on, so I went ahead and overclocked. Now that I see the overclock doesn't assist with VA, I'm back to stock speeds.

Quote:
Originally Posted by kevinqian View Post

According to DXVA checker you already have VA activated. Now run the cheese slice and post a screenshot. You may have VA but just not seeing it with trained eyes. Make sure to take sthe screenshot while the top graphs are moving initially.

You can also run cheese slices within DXVA checker. Click Directshow filters button, then select the video file and then click on the MS DTV-DVD decoder>Play>EVR. If the player shows VectorAdaptive, then it's being used.

I've taken two consecutive screenshots during playback of the "Slicies MPEG2 PAL 1080i 25.ts" clip, so you can see how the grid lines are not solid when in motion and also how in the VA section, as pointed out by blaubart in his original post, the diagonal lines don't remain lines when in motion.

In addition to the above, I've attached a screenshot of the "man in glasses" 576i clip, just so you can see VA in use on my system.

All images were taken in MPC HC using the Microsoft MPEG-2 Video Decoder and the EVR CP.

I tested the Slicies MPEG2 PAL 1080i 25.ts clip in DXVA checker by selecting "Microsoft MPEG-2 Video Decoder => Play => EVR" and it states below the video "Processor Device: VectorAdaptiveDevice ()". While that might indicate that VA is being used, judging by the tell-tale signs, as mentioned at the start of my post, something is definitely amiss.

I also just wanted to say thanks for all the help so far you two, I'm learning a lot more about deinterlacing than I thought.
LL
LL
LL
post #158 of 283
Congratulations you have VA at 1080i.
post #159 of 283
Quote:
Originally Posted by kevinqian View Post

Congratulations you have VA at 1080i.

That's odd, as I thought playback of the 1080i cheese slices clip would look practically identical to the 1080p cheese slices clip, but it doesn't.
post #160 of 283
Thread Starter 
Yes you have VA. To see that there's a really big difference exit MPC-HC, in CCC switch to MA and start Slicies again.
post #161 of 283
Quote:
Originally Posted by blaubart View Post

Yes you have VA. To see that there's a really big difference exit MPC-HC, in CCC switch to MA and start Slicies again.

I should have taken screenshots when I originally posted as VA quality was the same as Bob quality when I started this debacle; how I'm sure is because I compared what I saw with the examples in your original post. The one thing that remains the same, however, is those grey lines. They're still not as clear as the progressive cheese slices clip when using VA.

Anyway, I thought the point of VA was to provide the cleanest deinterlacing method that would match a progressive version of a file? I don't see that on my system, do you see it on yours? What I'm getting at is that if you test the 1080i file on your system with VA, does the quality match the 1080p clip?
post #162 of 283
You will never be able to achieve the 1080p demo results. The point of this test is stressing deinterlacers beyond their capabilities and so far none can do it. But worry not, that's as good as you will get in any of today's technology. Most tv sets today are only capable of MA and i bet you never noticed.
post #163 of 283
Thread Starter 
Quote:
Description:

First, don't get shocked - the pattern will NEVER look totally clean and "peaceful". I had to build in points (noise), lines and movements that can not be handeled by today's deinterlacers. And some that will be almost deinterlaced. But just in these parts the different video processors and deinterlacing options show determinable effects.

post #164 of 283
Quote:
Originally Posted by kevinqian View Post

You will never be able to achieve the 1080p demo results. The point of this test is stressing deinterlacers beyond their capabilities and so far none can do it. But worry not, that's as good as you will get in any of today's technology. Most tv sets today are only capable of MA and i bet you never noticed.

Quote:
Originally Posted by blaubart View Post


No problem and thanks again to both of you, I'm glad it's all sorted.

The funny thing about the new deinterlacing options available to me is that DVDs now look even clearer than before; which I didn't expect as I thought the deinterlacing options I was using, mainly bob, were the best I could get. I take a lot of screenshots of DVD, HD DVD and Blu-ray and when compared in Photoshop, the DVD quality is better than I originally thought possible, which means I'll have to redo my DVD screenshots! Not necessarily a bad thing, as now I have a better basis for comparison.
post #165 of 283
Thread Starter 
Ok, edition 275d of Post #1, (not) extra for you Rsaeire . Hope some points are better to be understood now?

Please pardon me for sleeping sometimes in English class years ago.
post #166 of 283
Quote:
Originally Posted by blaubart View Post

Ok, edition 275d of Post #1, (not) extra for you Rsaeire . Hope some points are better to be understood now?

Please pardon me for sleeping sometimes in English class years ago.

blaubart, your guide is excellent. There was confusion on my side as I was looking at cheese slices over the course of two days and changing a multitude of settings and getting nowhere. I still don't know exactly what I did to resolve the VA issue, but it was definitely resolved after posting here and due to your and kevinqian's advice.

Now I'm just trying to figure out which MPEG-2 decoder to use when watching DVDs, as I'm unsure whether there's any picture quality difference between v7.3/8/9 of the Cyberlink Video/SP decoder.
post #167 of 283
Cheese slices referenced a second time.
http://www.anandtech.com/video/showdoc.aspx?i=3738&p=3

Apparently 5670=4670 for deinterlacing performance. 5570 still doesn't quite hold up.
post #168 of 283
Quote:
Originally Posted by tsanga View Post

Cheese slices referenced a second time.
http://www.anandtech.com/video/showdoc.aspx?i=3738&p=3

Apparently 5670=4670 for deinterlacing performance. 5570 still doesn't quite hold up.

Quote:
Originally Posted by amdinput View Post

b) Regarding the "Enable smooth playback" option, when it is checked, the drivers will detect the capabilities of the GPU and adjust the post processing algorithms enabled to make sure that there's no dropped frames, which is always the number one priority. This is typically on entry level products. With mid-range and higher, the option has no effect, as everything is enabled.

Yet, knowing that many like to have full control on what to enable/disable, unchecking the "Enable smooth playback" option, allows to control which post-processing is on or off.

I wonder how the cheese slices favor on the 5570 when you disable "Smooth Video Playback" and get all of the post-processing features back with VA deinterlacing.

The 5570 hardware should be powerful enough to handle it and it sounds more like a driver issue of turning off too much stuff. What exactly I do not know.
post #169 of 283
Thread Starter 
Since 2xxx - 4xxx series it was always the same, wait long enough and somebody got sth. to work, others following...

But this is nothing AnandTech/Ryan Smith should consider. New card, new driver - doesn't work right - so shame on you AMD! In this way it's more kind of a political controversy
post #170 of 283
Quote:
Originally Posted by renethx View Post

GeForce GT 230M is a discrete graphics (in that it is not integrated on the NB die). This GPU looks fine at least in specs (equivalent to GT 220, that supports VA).

Hi renethx,

do you have a GT 220 and have you verified that the GT 220 does VA? I'm asking because I'm thinking about buying one.

2nd question, not directed just at renethx, has anyone tested Nvidia ION? Does it support VA?

Thanks.
post #171 of 283
Quote:
Originally Posted by boiler11 View Post

Hi renethx,

do you have a GT 220 and have you verified that the GT 220 does VA? I'm asking because I'm thinking about buying one.

2nd question, not directed just at renethx, has anyone tested Nvidia ION? Does it support VA?

Thanks.

GT 220: Yes (tested)
ION = GeForce 9300: No (perhaps; GeForce 9300: tested, Adaptive at best)
ION2 = GeForce G210 (codename GT218): No (perhaps; GeForce G210: tested, MA with heavy stuttering)
post #172 of 283
Quote:
Originally Posted by tsanga View Post

Cheese slices referenced a second time.
http://www.anandtech.com/video/showdoc.aspx?i=3738&p=3

Apparently 5670=4670 for deinterlacing performance. 5570 still doesn't quite hold up.

Quote:
Originally Posted by dwaleke View Post

I wonder how the cheese slices favor on the 5570 when you disable "Smooth Video Playback" and get all of the post-processing features back with VA deinterlacing.

The 5570 hardware should be powerful enough to handle it and it sounds more like a driver issue of turning off too much stuff. What exactly I do not know.

dwaleke,

You're right. It was a driver issue that ATI has addressed:
http://www.anandtech.com/weblog/showpost.aspx?i=669

5570 is good enough afterall. Heck, the article even suggests 5450 may be enough if VA deinterlacing is all you want, but given ATI's poor track record with driver flakiness, it may be a bit too marginal.
post #173 of 283
blaubart, you should write to AMD, demanding royalty for your hard work.
post #174 of 283
Thread Starter 
Ok, AMD do you here that? Then send the dollars to Haiti, NOW!
post #175 of 283
Now that I have VA deinterlacing working correctly, I have been updating my existing DVD screenshots and have noticed something odd upon closer inspection.

Here is what I am using to playback DVDs in Vista with a Radeon HD 3650:

MPC HC => Cyberlink Video/SP Decoder (PDVD8) => EVR CP.

If I zoom in on a DVD screenshot, taken when VA was in use, in Photoshop, I notice that the pixels seem misaligned, as if the deinterlacing wasn't done correctly. I can notice this "supposed" misalignment when zooming in anywhere between 600 - 1600%. I have included a cropped section of a screenshot from Blade II zoomed in 800% in PNG and have included both Bob and VA examples. While I'm aware that Bob and VA are different deinterlacing methods, I have included both for comparative purposes.
LL
LL
post #176 of 283
Thread Starter 
But the misalignment may be caused by lots of things happening during screenshooting and zooming. Why is *hi zooming* so important for you? Even scaling from SD to 1080 is hardest work for a HTPC and the quality differences enormous depending on your soft and hardware.
Try to set the output of your HD 3650 to the original resolution of the DVD video (in CCC). So the TV will do the scaling. TV's do a good, may be better scaling then a HTPC. I'm saying this carefully, hoping the AVS-pro's will not beat me.
post #177 of 283
Quote:
Originally Posted by blaubart View Post

But the misalignment may be caused by lots of things happening during screenshooting and zooming. Why is *hi zooming* so important for you? Even scaling from SD to 1080 is hardest work for a HTPC and the quality differences enormous depending on your soft and hardware.
Try to set the output of your HD 3650 to the original resolution of the DVD video (in CCC). So the TV will do the scaling. TV's do a good, may be better scaling then a HTPC. I'm saying this carefully, hoping the AVS-pro's will not beat me.

It's not that I'm looking to use high-zoomed in screenshots, I was just comparing image quality between Bob and VA deinterlacing methods. The screenshots are taken while the DVD is paused and at the DVD's native resolution, i.e. 720 x 576, and then resized to 1024 x 576 in Photoshop using the Bicubic filter, to accommodate the image's correct aspect ratio; and yes, the misalignment is present before resizing.

So, do you think something is amiss with the VA deinterlacing, as shown in my screenshots, or am I merely seeing the results of how the MPEG-2 codec compresses visual data?
post #178 of 283
Thread Starter 
Quote:


or am I merely seeing the results of how the MPEG-2 codec compresses visual data?

Decompression, yes and in addition you're analyzing the software genlock of the player at work. Did you notice that the paused video is always a bit more grainy then the moving? And if you switch 1 frame ahead or back the grains are somewhere else? That's it.
post #179 of 283
Quote:
Originally Posted by blaubart View Post

Decompression, yes and in addition you're analyzing the software genlock of the player at work. Did you notice that the paused video is always a bit more grainy then the moving? And if you switch 1 frame ahead or back the grains are somewhere else? That's it.

I hear what you're saying. Do you really think that's the answer? I was thinking it might have been the pixel structure and that's why it's blocks of pixels aren't exactly perfectly aligned, as the MPEG-2 codec works in blocks of 8 x 8 pixels, but that still doesn't appear to be it.
post #180 of 283
Thread Starter 
Sorry I could not help you more. I had to be at your place to see what you mean. And we have no references to depend on. In this way we may speak and speak but mean completely different things. As a consequence of watching many endless discussions leading to nowhere I built Slicies for *everybody*. Amazing enough that this worked.
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Home Theater Computers
AVS › AVS Forum › Video Components › Home Theater Computers › HD 1080i Test Pattern to determine Vector Adaptive Deinterlacing + others icl. Ticker