Need sage advice: capture cards, TBC, VCRs, pulling hair out!

[bilbofett]

Adv Color Bars attached (see post #60)

[bilbofett]

VHS movie examples 1-5 attached (see post #60)

[sanlyn]

Quote:

Originally Posted by bilbofett

Adv Color Bars attached (see post #60)

I took a look at the color bars and the samples. Are you sure DEFAULT and CALIBRATED aren't switched

DEFAULT: The darkest black bars are clipped at Y=16. Look at the "Minimum: line in the image below. The luma histogram doesn't go below Y=16.


CALIBRATED: Blacks on the lower left look darker than the default, but still clipped at Y= 16 anyway.


Why is one darker at the balck end? I think the clue is here in your notes on settings:

Quote:

1) ATI-USB600 DEFAULT ProcAmp settings.
ATI
brightness 110
contrast 32
hue 64
saturation 32
sharpness 2
*all are default*

8710
*all default*
....
.....

CALIBRATED ATI-USB600 ProcAmp settings.
ATI
brightness 108 <- Brightness is set darker than the default
contrast 32
hue 62 <- hue can change the luma value of some colors.
saturation 32
sharpness 1

8710
color +2 from 0 < increases overall contrast, darks darker, brights brighter
bright -2 from 0 < lowers black level
contrast +3 from 0 <- raises brights
all else default

These settings lower the blacks and add contrast. The ATI still clips at Y=16, and clips before the image ever gets to your ATI proc amp controls in VirtualDub.

Frame from Ex1.avi (darks seem to have been brightened, but dark detail still clipped):


You can brighten the darks, but you don't gain anything. It just greys-out.



If you're going to sue this workaround, you have to adjust darks at the 8710's proc amp, not the ATI's proc amp.

Meanwhile, I checked out the VC500, which isn't clipping.

[sanlyn]

I gave the VC500 a run last night and this morning. Turns out to be a pretty d ecent USB capture. I used VirtualDub for mall of these samples, adjusting the VC500's proc amp so that darks were as dark as I could get without clipping,

This is from a home VHS. You can't get a snappy image if blacks are muggy, murky, or shut down. This was played with the AG-1980. The black borders are at Y=0, no clipping in the cap device. [OOPS. Forgot to note, this was captured today with an ATI 9600XT AIW AGP. The images beaneth it are with the VC500),


One of the few retail VHS tapes I have left is Memohis Belle, kept because I keep fooling around with this horrible, horrible tape. The first 15 minutes is a muddy disaster. Crushed darks are even on the tape, forget about them ending up in a capture with even the best adjustment. These scenes are from the dim B17 interior, with lots of black jackets, shadows, etc. I played this tape with a Panasonic PV-S4670 (SVHS with no tbc), a DMR-ES10 for tbc pass-thru, and the AVT8710 for macrovision.

Frame 3600 with ColorYUV analysis readout and YUV histogram. I could have made darks brighter in the proc amp, but I wanted darks to be right on the edge of y=16-- which wasn't easy, I had to increase brightness anyway.


A dimmer shot of the interior:


Shot with very dark darks and bright brights:


All of these scenes can have shadow brought up a little more with filters later. Once blacks get crushed, though, forget about it.

The VC500 turned out OK. It's not an All In Wonder, but what else is?

[bilbofett]

Thanks again Sanlyn!

Did you note these examples?
CALIBRATED


CALIBRATED


CALIBRATED


CALIBRATED


Also, I guess the best example for a variety of blacks from the caps was "ex5.avi" (attached)

A few questions
1) Y=16, is that 16-235? (as opposed to 0-255?) So wouldn't that be ideal for the spec? I'm confused
2) I understand you're giving best recommendation to capture so that Y=0 (but as close to 16 as possible?)
3) When you say "brighten darks", are you saying, increase whites, or diminish blacks (or do something with grey?). "Brighten dark" almost sounds like an oxymoron to me
4) So I shouldn't do *any* ATI ProcAmp setting changes at the capture stage? Or, I need to do it at the 8710 level *prior* to the ATI ProcAmp?
5) Is it possible to adjust black on the 8710 so that Y=0 going into the ATI ProcAmp, and *still* calibrate the ProcAmp to have pleasant looking levels, without clipping/crushing the black? If I leave everything on the 8710 and the ProcAmp at default, it looks like washed-out, de-saturated greyish garbage (IMO).
I don't want to have to adjust levels on my TV everytime I load these up (or hand them out to friends/family)
6) Is your main point that calibration on the ProcAmp is worthless, and needs to be done on the 8710 only? What if I didn't have an 8710? Aren't long time veterans like lordsmurf and others calibrating the ProcAmp and changing settings on it?
7) I spent hours upon hours calibrating my monitor, the ProcAmp, and the 8710 using a professional calibration disc that I've used on several TVs for years now. I shouldn't do that?

A bit of my understanding lies with countless posts where lordsmurf, msgohan (maybe?) and yourself (maybe others) all praised the USB600, and everyone said it was great, have done hundreds of caps with it, wrote tons of how-to's for it, uploaded driver upon driver, but no one ever noticed it crushes/clips blacks all these years until I posted? Sorry, that sounds like a complaint..its not, its a bit of confusion and disappointment. Again, thank you everyone for all the awesome help. If before I was drowning, at least now I'm treading water (I hope)

[lordsmurf]

I actually have a VC500 that I want to sell. It was $100 about 8 years ago.
Decent, but not my favorite.

[bilbofett]

a note about the black bars in this first picture:



The 3rd black bars on the outer side represent the maximum amount of black allowed in the picture. I was instructed by the calibration disc to turn brightness down as far as it could go *just* until the 3rd black bars *almost* dissapear, but are barely visible.
I also noticed when I set my monitor to Limited RGB range (16-235), I cannot even see those 3rd black bars.. they're just gone. "Clipped", possibly.
When I set the monitor to Full RGB range (0-255), they re-appeared.

Now, the eyes are one thing, but actual numbers in ColorYUV are another.

In either calibrated or non-calibrated settings, the black bars are still there.
Here's the non-calibrated:



Is the black-level in both pics Y=16 instead of 0?

[sanlyn]

y=16 in YUV become RGB=0 in RGB. Video is stored as YUV but it displays as RGB on monitors and tv's, which use a different luma and gamma curve. The storage system was devised to meet bandwidth limits. 0-255 takes more bandwidth. Some of the bandwidth in transmitted video is used for other data -- one of the engineers in here can give you a graduate course in that area (LOL!). I know what it means in general, but don't ask for numbers.

When capturing as done here, you want to adjust input to fit within the y=16-235 range. You can spill a few pixels on either side for unimportant material but don't open the floodgates into the unsafe zone beyond 16-235. That spillover will show up as black or super-white in RGB.

If a capture device is already clipping at y=16, you'll have nothing to retrieve below that point. If you darken the input further, as you did with the samples, then video that started out as y=16 is now a lower value, say y=10 or y=6. That would get clipped, too. The video samples I posted involved raising the y value at the dark end to prevent crushed blacks (the brightness control), and lowering hot values to prevent clipped brights (contrast control). You might encounter tapes made in such a way that it requires different adjustments. If you start out with a luma range that extends lower than y=16 and higher than y=235, you have to raise the level of darks and lower the level of brights -- i.e, "shrink" the incoming signal to fit in that 16-235 window. It might not look so terriffic, but you're going to tweak anyway later. The idea is to prevent loss during capture.

Images of the The color and gray scale panels I posted earlier came from T3-Adv Color Bars-DEFAULT.avi and T4-Adv Color Bars-CALIBRATED.avi, near the end. For those images in the B17 I would, in practise, up the darks just a bit -- but since very little was darker than y=16, I could also have retrieved part of those few really darkest of pixels below y=16 in post processing, as long they're not totally cut off during capture.

The workaround is to Use the brightness control to raise black levels in the AVT so that the ATI doesn't clip so close to black. That trick was recommended in another forum. You might still have to make some adjustment to superwhites in the ATI proc amp, depends on the video.

Adjust your TV for every movie? I guess you've adjusted your TV with the calibration disc by now. What you do on a PC with histograms and post processing is to match your video for output to that kind calibration standard on your TV - which means 16-235 for capture. Your PC monitor could use some of that, too. You need a colorimeter to really do it right, but the TV setup disc can get you fairly close.

No, I don't mean that capture card proc amp adjustments are worthless. Use your AVT to trick your ATI into clipping at a lower input level (by raising blacks with your AVT), then tweak it with the ATI proc amp. Every tape will be different (a couple of your samples from tape had completely different level problems), but making the input brighter to begin with from the AVT should work with similar settings most of the time unless you run into a really dark or overly bright tape. Make the darks brighter, then fit them into the capture histogram with ATI's controls.

I realize several newer ATI UDB and PCIe devices are recommended and I don't object to things staying that way. Everyone ends up using their preference. At these low prices (my AIW's were costing some c-notes, and that was 14 years ago), it's worth a trial or two.

BTW, you mentioned "calibrating" a proc amp. Almost every project will require different adjustments, more or less. That applies no matter what capture device you use. Tape is just a damn outlaw, inconsistent and unpredictable. I've made thousands of digital recordings off tv and never adjust them. But those old VHS retail and tapes off cable? They're the road to insanity. Ain't you glad they'll be tidied up and gone one day?

[lordsmurf]

Quote:

Originally Posted by bilbofett

A bit of my understanding lies with countless posts where lordsmurf, msgohan (maybe?) and yourself (maybe others) all praised the USB600, and everyone said it was great, have done hundreds of caps with it, wrote tons of how-to's for it, uploaded driver upon driver, but no one ever noticed it crushes/clips blacks all these years until I posted? Sorry, that sounds like a complaint..its not, its a bit of confusion and disappointment.

You're starting to get into the aspect of video that will make your head spin.

Morpheus, to Neo: "This is your last chance. After this, there is no turning back. You take the blue pill, the story ends. You wake up in your bed and believe whatever you want to believe. You take the red pill, you stay in Wonderland, and I show you how deep the rabbit hole goes."

Most people want to blue pill: "Buy the ATI 600 USB, use it at default settings, be happy"

But the truth is that two major factors are also at play: the signal (VCR/tapes) and chipsets. Both can cause a wide variance in IRE and gamma values.

For example, the AVT-8710 TBC. The (mostly) green ones from pre-2011 are fine, the (mostly) black ones post-2011 have issues. But such differences are rarely as obvious. While I've not seen one, ATI 600 USB chips could vary. The beginning of a hardware run is almost never the same by the end of the run some years later.

Another example: Your tapes may be recorded with darker/lighter values, due to the cameras and VCRs.

And VCRs differ. The JVC and Panasonic behaves differently, but even models vary (JVC SR10U vs 9800U), and even the same model (1980 vs 1980). The VCR could crush black before the capture device ever gets it.

So before you can test hardware, and prove that it has issues, you have to have a wide baseline. This is why I often dismiss Amazon reviews as noise, and take so long to test hardware. It's not a matter of plugging it in, but plugging it into dozens of different items in different workflows. I need to see an error in multiple situations before I declare an error.

Tapes often give me a headache. Values are all over the place. Testing, testing, and more testing, and assembling a cooperative workflow, before an actual capture can commence.

Or just take the blue pill. They're Smurfy!

[sanlyn]

Thanks for more wise words, lordsmurf.

@bilbofett, a couple of those tapes fom your samples look really dark. Low-level clipping aside, do they look that way on TV? Just because they're retail is not a guide. My old copy of Fantasia looked just goofy, dark and no contrast, some sequences come up green, other problems. Give the dark tape a brightness boost with AVT, then tone it down with the ATI proc amp and histogram view to see if the dark bottom looks better. I realize some of it is supposed to be film noirish, but they're really too dark to begin with. I've been thru plenty of those myself. It's possible they don't have good detail anyway, like the earlier sample. At least the newer samples didn't have discolored patches in the shadows. I think it was just a flaw with the way that tape was made.

[lordsmurf]

Quote:

Originally Posted by sanlyn

My old copy of Fantasia looked just goofy

I thought it was Mickey?

[sanlyn]

Shoulda seen that one coming, LOL!

[bilbofett]

thanks for the replies gentleman. I like the Matrix, and I'm a red-piller.
I'll make this short and sweet: I have no idea how to use the avisynth COLORYUV plugin, nor the Vdub Spectral Analyzer.
Is there a simple(r) way to use them to monitor live captures so I can live-adjust the brightness/contrast to get black levels not to crush? I could then adjust it per-tape, one time, using some of the darkest/blackest scenes, and maybe some of the brightest scenes?
"That trick was recommended in another forum." Do you have a link handy please?

[sanlyn]

ColorYUV has functions that adjust contrast and several other brightness and color properties. If you go into "All Programs" or "Apps" or whatever name replacement Microsoft came up with during one of their over-medication spells, open the Avisynth program group and click on "Avisynth documentation". In the upper left corner is a small grid with the letters of the alphabet from A to Z. Click the letter C and you get a list of functions and commands that start with C. And so on. Or Google the phrase "Avisynth ColorYUV" and go to the older online wiki page (http://avisynth.org.ru/docs/english/...s/coloryuv.htm) or the newer page that's twice as long with tables, pictures, and stuff (http://avisynth.nl/index.php/ColorYUV).

Either way, you'll find that there isn't a Vdub spectral analyzer, but a ColorYUV function called "analyze". Use the statement "ColorYUV(analyze-true)" and all those numbers get overlayed on the image. Avisynth has a ton of documentation, maybe a bit too much for some of us -- like the details comparing ColorYUV() with Tweak() functions and the line that says "A saturation of 0.8 gives for example: cont_u = cont_v = - 0.2 * 256 = -51.2". OK, fellas, I'll take your word on that. I won't be getting that complicated with it.

Meanwhile, people use stuff like "Analyze" to examine problems. Nobody goes through that all the time. The histogram you see on the right-hand side of the images is put there by the Avisynth Histogram filter that gives you a choice of several specialized graphs to play with (http://avisynth.nl/index.php/Histogram). VirtualDub capture has a levels histogram that I hope you've been using. If you have difficulty with it, let us know. It takes only a minute or two to check adjustments.

What forum did I learn some tricks in? No way to remember where most tricks come from after all that time. Sad to say, there's no one-stop classroom for picking through all this material. Best way is to submit specific problems for solutions, but equally effective is browsing threads about video problems that aren't yours just to get a better understanding of how these processing elements come together. I save a great many samples, suggestions, and avisynth scripts or VDub filter tips.

The dark videos you picked for samples were a good idea, Probably the contradictory settings you made with the AVT and the ATI proc amps cancelled each other out. Remember that Brightness controls black levels, Contrast controls brights. It's a bit silly to use control names that don't do what they sound like they do, but that's the way those controls work. I'd try the darkest of those same samples again but with brighter AVT settings, and see what you get. Let us know.

[msgohan]

Quote:

Originally Posted by bilbofett

Is there a simple(r) way to use them to monitor live captures so I can live-adjust the brightness/contrast to get black levels not to crush? I could then adjust it per-tape, one time, using some of the darkest/blackest scenes, and maybe some of the brightest scenes?

You can use VirtualDub's capture histogram (with all borders cropped off). It is also possible to feed the input of a capture device to Avisynth with the aid of GraphEdit in order to use its waveform monitors instead, but unless you've already learned what the displays mean, there's no point investing in the effort of getting them to show up right now.

Quote:

Originally Posted by sanlyn

QTGMC also requires YV12.
...
ChomaShift and nnedi3 work in YUY2, but RemoveDirt variants and warpsharp won't.

General info for other scripts:

QTGMC does support YUY2, though at some point I managed to break this on my system. I assume when I upgraded to Avisynth 2.6, since the plugins are oriented toward YV16 for 4:2:2.

aWarpSharp2 supports YV16 in Avisynth 2.6.

[bilbofett]

sanlyn lordsmurf msgohan,
I incurred a very serious hospital stay, now I'm recouping.
I'm not even able to lift the S-VHS player (much less the laserdisc player)
so I can't run any capture tests or proceed forward for a few weeks.
Sorry I've been away.
I may be able to get a friend to come over and lift stuff and help me plug things in, then I can resume where this was left off (speaking mostly to the helpful sanlyn here, but everyone's helpfulness is much appreciated of course!).
It's taxing to even sit here and type this. I'm ok though. Oh well.

Sanlyn, my last cap test (a few weeks ago) using vdub's histogram showed I could basically go about it two ways:

a) leave the ATI-600 USB ProcAmp settings near default, and adjust the AVT-8710. I noted that one tick down on the 8710's brightness resulted in instantaneous clipping of black.
b) calibrate the ATI-600 USB ProcAmp live using the histogram, increasing the brightness but lowering the contrast; this seemed to provide *WAY* more headroom to settings on the 8710, ie., for some strange reason, no matter how low I set the brightness on the 8710, I could never get it to clip black?! Weird.

Anyone have any opinions on the ATI-600 PCI-E card and clipping blacks? I have that also, but never set it up.

[sanlyn]

Sorry to find you're out of sorts for while, but there's no rush. I'll be away for a week helping with family. Better to enjoy your recoup. Capturing is already enough frustration.

It would be better to raise blacks a little with your AVT before the signal gets to the capture device. The AVT won't need a huge adjustment, maybe only 15% or so -- check the results in the capture histogram. Then use the VDub histogram and the ATI proc amp to keep levels from blowing away. You'll have to tweak anyway after capture.

[msgohan]

Sorry to hear that, and yeah, there's no rush from us.

Quote:

Originally Posted by bilbofett

increasing the brightness but lowering the contrast; this seemed to provide *WAY* more headroom to settings on the 8710, ie., for some strange reason, no matter how low I set the brightness on the 8710, I could never get it to clip black?!

I believe it's simply clipping at 30, for example, instead of 16. A large amount of clipping would show as a big spike at the bottom of the histogram, regardless of how far away from red it is.

[bilbofett]

ok, I'm back.
informal poll for sanlyn + msgohan + lordsmurf (and anyone else)

should I just abandon any further testing with the Ati 600 USB?
I've not heard one person say "It does NOT clip blacks".

I don't want to spend any more time on it if there's no way around it.

To give an idea; I started my project on March 20th and I *still* have not done one single official capture yet, because I'm dinking around in the prep stage with all this testing. That's not progress, its pathetic.
I'd rather spend my time getting another analog cap card that doesn't clip blacks.

Please tell me your vote (with any recommendations for a similar card)
I still have the ATI 600 PCI-E, does anyone know if that clips blacks? It's a different chipset

Thanks for the help guys, its really appreciated.

[lordsmurf]

I'm having a hard time following this thread.

What I can say is this: I've not noticed an issue with my own ATI 600 USB card.

I can re-test it sometime, but not for a while. This was a card that I used on location, on a decently-calibrated IPS laptop, and it performed as needed. When not traveling, I'd use the ATI AIW cards. Those were better. Also realize that the tapes I dealt with are way worse that anything you're dealing with. On the off chance that is does clip (which I don't think it does), black levels were the least of my problems.

Since 2012, my ATI 600 has mostly gathered dust in a drawer. No more traveling.

My advice: If you're wanting a high-quality card, get the ATI AIW, and quit trying to find something that works in a post-XP world. Trying to limit yourself to non-XP PCIe or USB is too limiting.