Comb Filter Testing :: 1 < 2 > 4

  • Reply
  • Print
althor1138's avatar
RE: Comb Filter Testing

I've updated post #2 to include a link to the lossless file of the pluge test. To my eyes it looks identical to the x264 version.

Last edited on April 28, 2013 at 5:38 PM by althor1138

Luke threw twice....maybe.

althor1138's avatar
RE: Comb Filter Testing

So what was the verdict on the pluge test?  Has anybody checked it out yet? It would be interesting if somebody would post the same test with another card just to see a real world comparison.

Luke threw twice....maybe.

drngr's avatar
RE: Comb Filter Testing

The ATI TV Wonder HD 750 USB does not have a 3D comb filter, by the way. Only the PCIe card.

Last edited on June 6, 2013 at 2:48 AM by drngr
poita's avatar
RE: Comb Filter Testing

I will be back home next week and will take a look at your pluge test and run through all the various comb filters I have here.

What software and OS are you using to capture with the PCIe 750?

althor1138's avatar
RE: Comb Filter Testing

poita said:

What software and OS are you using to capture with the PCIe 750?

I am using Win7x64 and 32bit Virtudaldub to capture.

Luke threw twice....maybe.

poita's avatar
RE: Comb Filter Testing

I have one of those cards here, had no luck under Windows7 64 bit and virtualdub, it bluescreened me everytime I hit capture. Maybe you could drop me a PM as I would like to try the card without having to drop back to XP.

 

 

AntcuFaalb's avatar
RE: Comb Filter Testing

I'll be testing the Hauppauge HVR-2250 (NXP SAA7164 chipset) over the weekend.

I plan to capture the Snell & Wilcox Zone Plate, that pattern we settled on for AGC testing, and the Stormtroopers-looking-for-Leia clip.

I'll be connecting my Panasonic DVD-S29 to it via composite, so the aforementioned captures should demonstrate the quality of the SAA7164's 3D comb filter.

Is there anything else ya'll would like me to capture?

"And I'm shocked at you Moth3r for being off-topic, Because if people off-topic you say "stay on-topic, STAY on-topic, STAY ON-TOPIC", and we are not in the Off topic section of OT.com, now are we?" –pat man

"Look again." –Moth3r

poita's avatar
RE: Comb Filter Testing

Do you have the Video Essentials on laserdisc as well?

 

AntcuFaalb's avatar
RE: Comb Filter Testing

poita said:

Do you have the Video Essentials on laserdisc as well?

Yes, I do, but both of my LD players are with Duncan, so I can't spin up the VE LD at the moment.

The NXP SAA7164's 3D comb filter is way too strong. There's significant checkerboarding and ghosting at sharp shot-changes!

... and by "ghosting" I mean real temporal-NR-looking ghosts.

Yuck.

Last edited on May 27, 2013 at 5:03 AM by AntcuFaalb

"And I'm shocked at you Moth3r for being off-topic, Because if people off-topic you say "stay on-topic, STAY on-topic, STAY ON-TOPIC", and we are not in the Off topic section of OT.com, now are we?" –pat man

"Look again." –Moth3r

poita's avatar
RE: Comb Filter Testing

I'm yet to find a 3D comb filter that doesn't do that unfortunately to varying degrees.

 

Last edited on May 27, 2013 at 11:59 AM by poita
AntcuFaalb's avatar
RE: Comb Filter Testing

poita said:

I'm yet to find a 3D comb filter that doesn't do that unfortunately to varying degrees.

I've never seen one this bad.

The moving circle in the Snell & Wilcox Zone Plate pattern left easily-visible ghosts of itself. Yuck!

The checkerboarding wasn't as bad as the DMR-ES10, but I've never seen the DMR-ES10 leave temporal-NR-looking ghosts.

Last edited on May 28, 2013 at 1:41 PM by AntcuFaalb

"And I'm shocked at you Moth3r for being off-topic, Because if people off-topic you say "stay on-topic, STAY on-topic, STAY ON-TOPIC", and we are not in the Off topic section of OT.com, now are we?" –pat man

"Look again." –Moth3r

AntcuFaalb's avatar
RE: Comb Filter Testing

poita: What do you think of the ATI Theater 750 HD's 3D comb filter?

althor1138's samples don't show any of the 3D-comb-filter-artifacts I'm accustomed to seeing.

I wish we had a solution for the dynamic AGC problem...

"And I'm shocked at you Moth3r for being off-topic, Because if people off-topic you say "stay on-topic, STAY on-topic, STAY ON-TOPIC", and we are not in the Off topic section of OT.com, now are we?" –pat man

"Look again." –Moth3r

drngr's avatar
RE: Comb Filter Testing

AntcuFaalb said:

The NXP SAA7164's 3D comb filter is way too strong. There's significant checkerboarding and ghosting at sharp shot-changes!

... and by "ghosting" I mean real temporal-NR-looking ghosts.

How can you tell that it's a result of the comb filter vs. an actual noise reduction filter feature of the chip?

AntcuFaalb's avatar
RE: Comb Filter Testing

drngr said:

AntcuFaalb said:

The NXP SAA7164's 3D comb filter is way too strong. There's significant checkerboarding and ghosting at sharp shot-changes!

... and by "ghosting" I mean real temporal-NR-looking ghosts.

How can you tell that it's a result of the comb filter vs. an actual noise reduction filter feature of the chip?

I explicitly disabled noise reduction in the Capture Filter Settings and checked several times to make sure it was disabled prior to writing that post.

(Interestingly, the card gives you the ability to disable the 3D Comb Filter from the same dialog.)

I also viewed it in a dumb DirectShow capture program that has no noise reduction feature of its own (WM9Cap).

"And I'm shocked at you Moth3r for being off-topic, Because if people off-topic you say "stay on-topic, STAY on-topic, STAY ON-TOPIC", and we are not in the Off topic section of OT.com, now are we?" –pat man

"Look again." –Moth3r

althor1138's avatar
RE: Comb Filter Testing

AntcuFaalb said:

I wish we had a solution for the dynamic AGC problem...

 

I haven't found any instances in capturing my laserdiscs where the agc causes any negative issues.  Of course I might just not be seeing it. 

Last edited on May 31, 2013 at 11:21 PM by althor1138

Luke threw twice....maybe.

poita's avatar
RE: Comb Filter Testing

AntcuFaalb said:

poita: What do you think of the ATI Theater 750 HD's 3D comb filter?

althor1138's samples don't show any of the 3D-comb-filter-artifacts I'm accustomed to seeing.

I wish we had a solution for the dynamic AGC problem...

Mine just keeps crashing the system with a BSOD under windows 7.

I installed 32 bit XP, it worked okay until I rebooted, now I get the BSOD again.

Every time I go to virtualdub and hit capture AVI it bombs out.

drngr's avatar
RE: Comb Filter Testing

Suppose we figured out the best 3D comb filter and then captured everything with the best 3D comb and a second time with the best 5-line 2D comb.

Could we come up with an algorithm to pick the best frame?

Last edited on June 3, 2013 at 5:09 AM by drngr
poita's avatar
RE: Comb Filter Testing

Yes, I have been discussing this recently. A script that took the first two frames from each scene change from a 2D capture and inserted them into a 3D capture would solve a lot of the worst of it.

AntcuFaalb's avatar
RE: Comb Filter Testing

The simplest solution is to use the SCSelect() Avisynth function.

"And I'm shocked at you Moth3r for being off-topic, Because if people off-topic you say "stay on-topic, STAY on-topic, STAY ON-TOPIC", and we are not in the Off topic section of OT.com, now are we?" –pat man

"Look again." –Moth3r

drngr's avatar
RE: Comb Filter Testing

Snell & Wilcox 4:3 Pattern 1: Sony DVP-NS50P -> Panasonic DMR-ES15 -> Diamond VC500

It looks virtually identical to the performance of the DMR-ES10, unsurprisingly.

For quick comparison:


V1 = AVISource("Comb Filter - Panasonic DMR-ES10.avi").AssumeTFF().Bob()
V2 = AVISource("Comb Filter - Panasonic DMR-ES15 to Diamond VC500 [128-128-128-140-0].avi").CropPad(8,0,0,0).Trim(0,30).AssumeTFF().Bob()

Interleave(V1,V2).Histogram()

function CropPad(clip c, int x1, int y1, int x2, int y2, int "color") {
   color = Default(color,color_white)

   c.Crop(x1,y1,-x2,-y2).AddBorders(x2,y2,x1,y1,color)
}


Meanwhile, I've tried two different DVD players as source, and with both I get a 15-frame cycle of partial rainbowing on the second Snell & Wilcox pattern. The Sony comes out considerably worse than the other one.

Snell & Wilcox 4:3 Pattern 2: Sony DVP-NS50P -> Panasonic DMR-ES15 -> Diamond VC500

AntcuFaalb's avatar
RE: Comb Filter Testing

Thanks for contributing, drngr!

The DMR-ES15 uses the same in-house-developed A/V chip as the DMR-ES10.

It's the Panasonic C1AB00001979 according to the service manual.

"And I'm shocked at you Moth3r for being off-topic, Because if people off-topic you say "stay on-topic, STAY on-topic, STAY ON-TOPIC", and we are not in the Off topic section of OT.com, now are we?" –pat man

"Look again." –Moth3r

_,,,^..^,,,_'s avatar
RE: Comb Filter Testing

I found this comparison really interesting, take a look at it, it's worth time to read it all: http://notonbluray.com/blog/comb-tb-tests/

   "IT!. . . COULD! . . . WORK!!" 

 

released: SW:ANH [OT) | SW:ANH ['97) | Jurassic Park | The Lost World: JP | The Thing |

drngr's avatar
RE: Comb Filter Testing

Here is a test of the "XCard" PCIe board I mentioned some months back in the cap card thread, put together by a VideoHelp member called danno78. It's based on an AVerMedia mini-PCIe capture card that does 3D Y/C built into some laptops. According to a member who was pissed off at danno78 for obscuring its origins, it uses the SAA7136 v2.

Snell & Wilcox 4:3 Pattern 1 - three tests

  • Philips DVP642 - CVBS - Panasonic DMR-E20 - YC - XCard
  • Philips DVP642 - CVBS - Panasonic DMR-ES15 - YC - XCard
  • Philips DVP642 - CVBS - XCard

 

Negative effects of DVD recorder passthrough:

Both have added noise and smear the chroma resolution a little.

The EMR-E20 shifts the image <0.5 pixels to the left, and causes the bottom line to jitter.

The DMR-ES15 mattes off ~2.5 pixels along the right edge. Mine brightens the image, which may be fixable by changing the "Lighter/Darker" setting in the Setup menu, but I can't find a universal remote that will give me that button.

I don't know how they affect the left edge of the image, because my DVD players only output black there.

Comb filter comparison:

The DMR-E20 has a small amount of dot crawl in the 1.0 and 1.5 MHz chroma patches. Other than that, the two Panasonic recorders seem to function equivalently.

They both have a small amount of dot crawl near the top of the image not present with the XCard. Meanwhile, the XCard has additional rainbowing in the center third of the moving zone plate (both horizontally and vertically).

All three require the zone plate to be still for 4 frames to be completely clean. The 3rd frame has slightly less rainbowing with the XCard.

Here is a quick comparison of a shot change on an LD to show checkerboarding: DMR-E20 vs XCard. The XCard has a bit less.

 

Another card that does 3D comb filtering is the GOTVIEW X5 3D Hybrid PCI-E (the one without "3D" in the name does too, but uses a hardware MPEG-2 encoder). Some PAL testing of that hardware encoder was done by PCtuner.ru, without the SW plate. The ADC is CX23887, also used in the Hauppauge WinTV-HVR-1250 (software) and WinTV-HVR-1800 (hardware). But Hauppauge's comparison table claims that the HVR-1250 doesn't have a 3D comb filter. I guess they forced it into 2D mode...?

Last edited on August 25, 2013 at 9:54 PM by drngr
drngr's avatar
RE: Comb Filter Testing

I also did a bunch of searches and found which Panasonic DVD recorders use which chips. Presumably the ones with matching video processors have the same comb filter performance.

C1AB00000791:
DMR-E20 (3 in)
DMR-E30 (3 in)
DMR-HS2 (HDD, 3 in)
DMR-T3030 (HDD, 3 in)
DMR-T3040 (HDD, 3 in)

C1AB00001979:
DMR-E65 (3 in)
DMR-E75 (VCR, 2 in)
DMR-E85 (HDD, 3 in)
DMR-E95 (HDD, 3 in)
DMR-E500 (HDD, 3 in)
DMR-EH50 (HDD, 3 in)
DMR-EH55 (HDD, 3 in)
DMR-EH60 (HDD, 3 in)
DMR-ES10 (3 in)
DMR-ES15 (2 in)
DMR-ES20 (2 in)
DMR-ES25 (2 in)
DMR-EZ17 (2 in)
DMR-EZ27 (2 in)
DMR-T6070 (HDD, 3 in)

C1AB00002083:
DMR-EA38V
DMR-EH75VP
DMR-ES30VP
DMR-ES35V
DMR-ES40VP/VPC
DMR-ES45V (VCR, 2 in)
DMR-ES46V (VCR, 2 in)
DMR-EZ37V
DMR-EZ475V
DMR-EZ47V
DMR-EZ48VP/VPC
DMR-EZ485VP
SA-HT830VP/VPC
SA-HT833V

C1AB00002084 (PAL only?):
DMR-ES30VGN/VEC/VEE (VCR, 2 in)
DMR-ES35VEG/VEC (VCR, 2 in)
DMR-EZ48VEB (VCR, 2 in)

C1AB00002379 (PAL only?):
DMR-EH55EC/EP
DMR-EH56EG
DMR-EH65EE/GC/GCS/GN (Russia, etc.)
DMR-ES15EB/EBL/EP/EC/EG (UK, Ireland, other EU countries)
DMR-EX75EB/EC/EG (UK, other EU countries)
DMR-EX79EG
DMR-EX85EB/EG (UK, Germany)
DMR-EX768EB/EF/EC/EP (UK, France, other EU countries)

Last edited on August 26, 2013 at 8:03 AM by drngr
AntcuFaalb's avatar
RE: Comb Filter Testing

Can we assume that C1AB00002379 > C1AB00001979?

Is there a list online somewhere of every DVD recorder Panasonic ever made?

"And I'm shocked at you Moth3r for being off-topic, Because if people off-topic you say "stay on-topic, STAY on-topic, STAY ON-TOPIC", and we are not in the Off topic section of OT.com, now are we?" –pat man

"Look again." –Moth3r

Members reading this topic: None