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 3

post #61 of 283
Quote:
Originally Posted by blaubart View Post

... Arcsoft is playing all fine. Cyberlink+VC-1 is not my way.
Do you have other interlaced VC-1 Video to check the judder?

The ArcSoft decoder might work better, but I don't trust it because it modifies the memory locations of the Windows Explorer as my Agnitum Outpost Firewall always reported to me when I started MPC-HC together with the ArcSoft decoder. I guess it tries to connect to the internet or maybe opens a backdoor to the filessystem. So, I prefer not to make any use of ArcSofts tricks.

Unfortunately, I don't have a really long VC-1 clip for you, couldn't find anything yet. And I'm still not allowed to post URLs. You can try this short clip:
wwwDOTalice-dslDOTnet/pkeunecke/testVC1i.m2ts
post #62 of 283
Thread Starter 
@tsanga
... but I like your style too


@Youg
buying new cards may help: people (have to) think about the innards of their PC in a very new way, rip out cards they forgot they were in, slim down other unnecessary, replace drivers, update softs, change this and that and HEEEY it works. But doing that all with the old card would have had the same effect...
Nothing but some silly statement, you know..


@peter
Thanks, all clips are welcome. And VC-1 int. rare to find!
Actually I asked the "Do you have other VC-1.." to hear if you have sth else to test than Slicies.

Ok, in your clip I have 2 stutters at the point the camera falling into the snow and 2 more at the point the rocks are passing by using your VC-1 Gabest+Cyberlink+VMR7(windowed). Arcsoft: no stutters with all renderers.
If you could try that on your setup?

"Agnitum Outpost Firewall always reported..."
Once I had a malware checker I could hardly convince not to delete my empty floppy drive from the system. Next he wanted to kill my empty trashcan. All dangerous...
I think those anti-softs often above all try to have influence on the security of the user in order to make him somehow paranoid and buy more and more anti-softs..
BTW, KIS did not complain.
post #63 of 283
@blaubart
ok, confirmed! The ArcSoft decoder plays all clips really smooth and yes indeed the CL VC-1 decoder stutters exactly at those positions that you mentioned. The ArcSoft decoder can even play some 60 FPS progressive clips which can't be played with the internal MPC-HC VC-1 decoder. The latter produces just a lot of artifacts and macroblocks unlike the WMVideo decoder, but that one has no DXVA support and the CyberLink VC-1 decoder won't connect to the ASF source filter. http://www.megaupload.com/?d=2KAAU7RR

So, I will try to install the ArcSoft "*.ax" files and some tweakers with regsvr32 manually now instead of using this suspect HD Packs installer to avoid any security risk as explained above. And I'm sure that the Agnitum Outpost Firewall had a good reason for the security warning and the network blocking of the modified Explorer, because my StartupMonitor also came up with a requester asking me for a connection service called "ACDaemon.exe". When I block that connection service then the Firewall stays quiet, but then MPC-HC scans the complete Windows directory before starting to play the clips with the ArcSoft decoder. Very strange, maybe caused by the HD Packs.
LL
post #64 of 283
Greetings peter.

You could try deleting the ACDaemon.exe ArcSoft put in the HKLM/.../ Run registry entry, and setting the Arcsoft Connect Daemon service to manual.

Best regards.
post #65 of 283
Thread Starter 
Ahh, that's what you mean, sorry peter, just forgot that service. Same thing with the demo, was the first for me to stop.

I will be back in a week,
Regards to all
post #66 of 283
blaubart may not be around....

What does it mean if the two 1p boxes don't flicker when they're stationary, but one's solid black and the other solid white? There's no combing visible anywhere, so it's not anything like weave. The ticker shows adaptive deinterlacing.
post #67 of 283
Thread Starter 
Back again.
"the two 1p boxes don't flicker when they're stationary, but one's solid black and the other solid white?"

That is a "doubling Skip Field" (read here). Only the odd or even field is used doubling the lines at half fps.
If the 1p boxes are both solid grey you have an "interpolated Skip Field".

Both deinterlacing modes are often used by decoders that cannot use DXVA (for the moment or always). The black+white vers. is the rarer, more troubled one.
post #68 of 283
Thanks for the explanation.

The thing is DXVA is on (CPU usage <10%). But I had noticed "skip field" before (I think) - when I pause the material and advance frame by frame, each one is shown twice, i.e. I'm actually seeing 30 fps at 60 Hz. I believe the decoder (MS DTV decoder) is doing this. I'll need to do more investigation to see if this happens with all video formats, and if this only happens during the non-moving part of your test pattern.

Also the behavior changes when I change driver version, so maybe just software limited. As renethx indicated, my GPU deinterlaces well with XP versions of the driver.
post #69 of 283
Thread Starter 
Did you ever have success running any interl. video? Maybe it's useful to describe first the go's to better understand the no go's. And maybe you should try some more decoders.
post #70 of 283
I have had success before. I duplicated renethx's setup and results in Windows XP and deinterlacing works very well with the Cyberlink MPEG2 decoder. It's the Vista/Win7 drivers that are trouble (also what renethx reported).

EDIT: Using your test pattern I discovered that MS DTV-DVD decoder does field skipping in 7MC x64, but in MPC x64 it does adaptive de-interlacing - video driver version 190.65. Different from results earlier with different version driver. So it appears at least some improvement was made (earlier driver only did weave), but now the performance doesn't translate to Media Center.
post #71 of 283
Alright, I hate updating my own post, but I might be onto something.

Using Nvidia driver version 190.62, GeForce 9400:

- Windows 7 Media Center x64 does skip field if resolution is 1080p, but adaptive if resolution is lower (CPU util < 10% in all cases)
- 32-bit WMP inside x64 Win 7 does adaptive regardless of resolution setting (CPU util < 10% in all cases)

So the driver (or decoder?) is different in 32-bit or 64-bit mode.
post #72 of 283
Thread Starter 
... or the player. MPC-HC e.g. is well known showing vital 32-64-bit differences. I hate to say this but in your case I would think about wether it's healthy now to lose patience and invest some bucks in a card that is well known to support Vector Adaptive via DXVA..

BTW if your CPU is one of the really super fast it is possible to have about 10% util. NOT using DXVA. So did you ever have more than 10%?
post #73 of 283
Quote:
Originally Posted by blaubart View Post

... or the player. MPC-HC e.g. is well known showing vital 32-64-bit differences. I hate to say this but in your case I would think about wether it's healthy now to lose patience and invest some bucks in a card that is well known to support Vector Adaptive via DXVA..

I did.

Quote:
Originally Posted by blaubart View Post

BTW if your CPU is one of the really super fast it is possible to have about 10% util. NOT using DXVA. So did you ever have more than 10%?

I don't think E7400 = superfast. I took your advice and always run the test now with the CPU util graph running in the background.
post #74 of 283
I stumbled onto something strange - not sure if this is a decoder or driver issue (or something else).

Setup:
Windows 7 RC x64 Media Center w/ MS DTV DVD decoder
HIS H467QS1GH Radeon HD 4670
Catalyst 9.8

- Vector adaptive deinterlacing works fine tested with cheese slices, "man with glasses" SD clip, and vertrezmotion from w6rz.net
- Play some other clip, e.g. AVS HD 709 basic settings mp4, and use trick play controls (e.g. skip forward/backward)
- Replay cheese slices and/or "man with glasses", and deinterlacing fall back to only adpative (VA is broken)
- Exit MC and relaunch, and everything back to normal, i.e. VA works again

Any ideas? Or just need new decoder/driver?
post #75 of 283
Thread Starter 
Maybe the other clip's format already forced MC to use adaptive but you could not notice that since it's not a VA test video? And after this shocking experience MC is unable to regain VA?
post #76 of 283
blaubart,

I think this might be exactly what's happening. I have been trying to figure out exactly what kind of video (format, encoding, etc.) knocks it into adaptive and gets it stuck. But I haven't arrived at any definitive conclusion.
post #77 of 283
Does anyone know where I could get a copy of the HQV disc?

It is discontinued and the only one I could find was used and $150. HQV did not reply to my email.
post #78 of 283
Quote:
Originally Posted by rdunnill View Post

Does anyone know where I could get a copy of the HQV disc?

It is discontinued and the only one I could find was used and $150. HQV did not reply to my email.

You might look at the Spears and Munsil disc.
post #79 of 283
Quote:
Originally Posted by -peter- View Post

@blaubart
ok, confirmed! The ArcSoft decoder plays all clips really smooth and yes indeed the CL VC-1 decoder stutters exactly at those positions that you mentioned. The ArcSoft decoder can even play some 60 FPS progressive clips which can't be played with the internal MPC-HC VC-1 decoder. The latter produces just a lot of artifacts and macroblocks unlike the WMVideo decoder, but that one has no DXVA support and the CyberLink VC-1 decoder won't connect to the ASF source filter. http://www.megaupload.com/?d=2KAAU7RR

So, I will try to install the ArcSoft "*.ax" files and some tweakers with regsvr32 manually now instead of using this suspect HD Packs installer to avoid any security risk as explained above. And I'm sure that the Agnitum Outpost Firewall had a good reason for the security warning and the network blocking of the modified Explorer, because my StartupMonitor also came up with a requester asking me for a connection service called "ACDaemon.exe". When I block that connection service then the Firewall stays quiet, but then MPC-HC scans the complete Windows directory before starting to play the clips with the ArcSoft decoder. Very strange, maybe caused by the HD Packs.

I also have a VC1 file that arcsoft plays ok but WMVideo decoder can't handle. Which of the *.ax file(s) do I need to register so I can use it in MPC or another directshow app?

thanks.
post #80 of 283
Thread Starter 
I'm not at my place now, can't tell you the ax at the Moment. But simple registering won't work anyway. Depends also on the version.
Since TMT3 the use of the filters in third party applications is disabled. You need TMT2.
Read down from here for more.
post #81 of 283
Quote:
Originally Posted by blaubart View Post

More service again, see edited bottom of topic..

onboard HD3200 will definetly not, that's for sure.

Quote:
Originally Posted by renethx View Post

To be precise,

- HD 3200/3300 with a HT 2.0 processor: Only Bob and Weave work.

- HD 3200/3300 with a HT 3.0 processor: Motion adaptive works up to 1920x1080; only Bob and Weave work beyond that (i.e. under dual monitor).

HD 3200/3300 with HT 3.0 is the best IGP (as far as HD deinterlacing is concerned).

not true.

I have had VA on a HD3200 with 5050e HT2 cpu. GPU speed at 1000 and bus at 225 made all options available (and they work) for 1920*1080.

same gpu and 240e HT3 at original speeds also have all options working
post #82 of 283
Quote:
Originally Posted by peeteebee View Post

not true.

I have had VA on a HD3200 with 5050e HT2 cpu. GPU speed at 1000 and bus at 225 made all options available (and they work) for 1920*1080.

same gpu and 240e HT3 at original speeds also have all options working

Information on your test configuration will be helpful (to reproduce your results):

1. OS (if Vista/7, Aero on or off? Don't expect ordinary users to turn off Aero.)
2. Driver version
3. Player (source filter, decoder, video renderer) used to play back the "Cheese Slices" test file (MPEG-2, VC-1, H.264 respectively)
4. Motherboard model number
5. Cool'n'Quiet on or off?
6. GPU core 1000MHz?? That's highly overclocked. Use the default clock 500MHz.
7. Use the default base clock 200MHz instead of overclocked 225MHz.

You shouldn't expect ordinary users to overclock the system (not every system has oc options; even if oc is possible, no guarantee of stability for every system). Saying "Hey, my overclocked system passed the test. Your results are wrong." does not sound good.

Just because I don't believe your results (contradict to the experience of every HD 3200/HT 1000MHz user I am aware of).
post #83 of 283
At stock speeds you are right, it doesn't work, I know. But this doesn't mean it will definetly not, like stated by blaubart.

You don't have to believe me, but I know what I had and saw using this HW, even if it contradicts to the experience of every HD 3200/HT 1000MHz user you are aware of. You have to update you list because now you do know of a user that had VA on an hd3200.

can't reproduce since I have replaced my 5050e by an 240e. Its not just changing cpu, my system is Passively cooled, so I have to take it apart.

To be honest, I was surprised to have all options available in CCC (bob, weave, va and others)

I deselected auto and selected VA. Using the test pattern .ts (cheese thing) I found my system was using VA, but since I don't like to overclock I put it all back to stock speeds and ordered a better (read HT3) cpu.

my system specs during that test:
gigabyte ga-ma78gm-s2h, rev 2
ams 5050e (at 2900+, clock 225 * default multiplier 13)
cpu at 1000 (also tried 800 and 900, they both work too)
win7 (had good results using vista too)

settings:
aero was off (and when it was on, I got a popup that it was switched off
during playback)
cnq was on
1920*1080 via hdmi

software/drivers:
ccc used: can't remember, have used 9.7 .. 9.11. Last tests were with 9.8
because newer versions had problems
7mc, mediaplayer, powerdvd 9


with new cpu and everything and stock speeds, I also have VA on my HD3200 integrated gpu.
post #84 of 283
Quote:
Originally Posted by peeteebee View Post

At stock speeds you are right, it doesn't work, I know. But this doesn't mean it will definetly not, like stated by blaubart.

You don't have to believe me, but I know what I had and saw using this HW, even if it contradicts to the experience of every HD 3200/HT 1000MHz user you are aware of. You have to update you list because now you do know of a user that had VA on an hd3200.

can't reproduce since I have replaced my 5050e by an 240e. Its not just changing cpu, my system is Passively cooled, so I have to take it apart.

To be honest, I was surprised to have all options available in CCC (bob, weave, va and others)

I deselected auto and selected VA. Using the test pattern .ts (cheese thing) I found my system was using VA, but since I don't like to overclock I put it all back to stock speeds and ordered a better (read HT3) cpu.

my system specs during that test:
gigabyte ga-ma78gm-s2h, rev 2
ams 5050e (at 2900+, clock 225 * default multiplier 13)
cpu at 1000 (also tried 800 and 900, they both work too)
win7 (had good results using vista too)

settings:
aero was off (and when it was on, I got a popup that it was switched off
during playback)
cnq was on
1920*1080 via hdmi

software/drivers:
ccc used: can't remember, have used 9.7 .. 9.11. Last tests were with 9.8
because newer versions had problems
7mc, mediaplayer, powerdvd 9


with new cpu and everything and stock speeds, I also have VA on my HD3200 integrated gpu.

Here is my attempt of reproducing your results. Unfortunately they were not reproducible. My conclusion is, Extreme overclocking + Unpractical assumption (Aero off) + Non-reproducible performance = Your claim of "not true" is totally meaningless. (I am not sure where you came from, but remember that this forum is not for overclocking competitions.)

HyperTransport 2.0

Test setup:

- GIGABYTE GA-MA78GM-S2H
- Athlon X2 4850
- Windows 7 32-bit, Aero off
- Catalyst 9.8 & 9.11
- HD 3200 @1000MHz, HT @1125MHz (225MHz x 5), C'n'Q off, Memory @400MHz (DDR2-800).

The best deinterlacing method available is Adaptive.

(If you are able to post a similar screenshot of your claimed CCC along with Cheese Slices, I will believe you. )

HyperTransport 3.0

Test setup:

- GIGABYTE GA-MA78GM-S2H
- Athlon II X2 240
- Windows 7 32-bit, Aero on
- Catalyst 9.8 & 9.11
- MPC HomeCinema 1.3.1378.0 with the internal MPEG TS souce filter, CyberLink Video/SP Decoder (PDVD9) & Microsoft DTV-DVD Video Decoder for MPEG-2, ArcSoft Video Decoder for VC-1, EVR.
- HD 3200 @500MHz, HT @2000MHz (200MHz x 10), C'n'Q on, Memory @400MHz (DDR2-800). That is, everything is at default.

Now Vector Adaptive is available in CCC, but this is deceptive, only Motion Adaptive is actually working!! Here are screenshots of Cheese Slices and HD HQV Benchmark:

Cheese Slices MPEG-2 1080i

HD HQV Video Resolution Loss Test (compare with these pictures)

In summary, my previous statements are perfectly correct!

Quote:
Originally Posted by renethx View Post

To be precise,

- HD 3200/3300 with a HT 2.0 processor: Only Bob and Weave work. (Update: Adaptive may work under certain extreme conditions [deinterlacing is still crappy])

- HD 3200/3300 with a HT 3.0 processor: Motion adaptive works up to 1920x1080; only Bob and Weave work beyond that (i.e. under dual monitor).


LL
LL
LL
post #85 of 283
Quote:
Originally Posted by peeteebee View Post

To be honest, I was surprised to have all options available in CCC (bob, weave, va and others)

Just having the options available in CCC does not mean they work. Did you try playing blaubart's samples to verify that VA was working?

My 4550 allows me to choose VA in CCC, but it does not work. At least not with the last drivers I tried, somewhere around 9.6.
post #86 of 283
HD 4200 is slightly better than HD 3200 with HT 2.0 processors. Notably all deinterlacing options are available in CCC 9.8 (of course deceptively!).

HyperTransport 2.0

Test setup:

- GIGABYTE GA-MA785GM-US2H
- Athlon X2 4850
- Windows 7 32-bit
- Catalyst 9.8
- MPC HomeCinema 1.3.1378.0 with the internal MPEG TS souce filter, CyberLink Video/SP Decoder (PDVD9) & Microsoft DTV-DVD Video Decoder and EVR.
- HD 4200 @500MHz (default), HT @1000MHz (200MHz x 5; default), C'n'Q off, Memory @400MHz (DDR2-800).

When Aero and all the Avivo post processing (except for deinterlacing) are off (a highly unpractical assumption!), MA works. Otherwise MA works but playback stutters heavily. (BTW even under extreme overclocking conditions, VA does not work.)

HyperTransport 3.0

The same as HD 3200: Up to MA works with Aero on or off.

Summary: If you want to get the most out of AMD 780G/785G/790GX IGPs, use a HT 3.0 processor. (The cheapest dual core HT 3.0 processor is Athlon II X2 240.) If you want Vector Adaptive deinterlacing for HD contents, your only choice is a good discrete card (e.g. Radeon HD 4670 or higher, GeForce 9500 GT or higher, GeForce GT 220 or higher).
post #87 of 283
just a quick reply since I reinstalling win7 and can't post results right now (will do so later)

I have had loads of occasions with only bob/weave and adaptive possible, but every now and then (don't know why or what I changed) the other options were there too and the testresults looked good.

Anyway, will post results asap (with 240e cpu). At this moment I think I should have bought the cpu since I'm thinking about buying an dedicated Nvidia card, without all the ATI do-donts problems
post #88 of 283
double post, ignore this one
post #89 of 283
I recently upgraded my system from a 780g based board (GA-MA780GPM-DS2H) to a 785g board. With this change VA is now available in CCC, and 1080i material does not stutter anymore (previously it did stutter, when I tried to enforce the VA setting)

Here's the spec:
Motherboard:A785GMH/128M
CPU: Phenom 8750 BE (mild overclock to 2.6 Ghz)
GPU: Clocked to 900Mhz
OS: Windows 7 x64
CCC: 9.8

A screenshot using Käsescheibchen looks like this


The screen shot was taken with MPC-HC (on 1080i H.264, but I get the same result for MPEG2, and also with the native decoders in WMP)

I'm kind of confused whether this is VA or not.
One word regarding the GPU overclock, I also get stutterfree playback with 700Mhz, but 900MHz makes the system a bit more fluent, e.g. switching between playback and EPG)
LL
post #90 of 283
Is the screen shot taken during the moving portion of the video, or the initial static phase?
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