Startech USB3HDCAP opinions

[FLP437]

" I assume you mean the Proc Amp"

Yes sure, I mean Proc Amp not post process.

[FLP437]

Sorry I pressed the return key too soon

[FLP437]

“I feel like you're reluctant to try the Proc Amp adjustment, for some reason”

No , I´m not reluctant quite the inverse , I tried immediately only I have no tool to see and quantify the result . I´m probably doing something incorrect as I cannot activate the histogram in virtualdub I only get a black bar. I´m not comfortable with AviSynth I only use it trough a GUI like VideoCleaner for instance and I don’t known of other programs to do it.

If you can enlighten me I will be grateful

[FLP437]

Ok . Finally I got it , it was necessary to put in preview mode for the histogram to work. Beginners problems.

I assume the idea is to avoid values less then 16 to the left and higher then 235 to the right. As the histogram doesn´t show values but red zones , I think red zones are the ones out of this values and is necessary to adjust contrast and brightness in video proc. amp. to avoid having red zones is that correct ?

histograms and video proc. amp. attached

[Brad]

it was necessary to put in preview mode for the histogram to work. Beginners problems.

Not just beginners' problems! VDub's histogram is a pain in the ass to get working for all of us.

I assume the idea is to avoid values less then 16 to the left and higher then 235 to the right. As the histogram doesn´t show values but red zones , I think red zones are the ones out of this values and is necessary to adjust contrast and brightness in video proc. amp. to avoid having red zones is that correct ?

Yes, you got it exactly.

BUT very importantly, you can't trust the "below 16" measurement unless you crop all of the borders including the bottom (I would even crop the top a little just to be safe). It's normal for these borders to include blacker-than-black values while the video itself is normal video black, so you don't want to adjust based on VDub reading the value of the borders.

For the sample you posted, the black level looked good so I would just leave it at default. If you really want to adjust it for other tapes, I can try to explain how to use the Cropping filter of VDub's capture mode. The idea is that you activate it only while setting up the Proc Amp. Then you remember to turn it off before actually capturing... which is very easy to forget.

If you have Avisynth installed, using its histogram to check out a captured file is as simple as creating a text file with .avs extension instead of .txt containing:

AVISource("Your Capture Filename.avi")
Histogram()

Then you can open the AVS file in VirtualDub just as you would an AVI file.

[FLP437]

I still have the problem of how to select the best workflow to recapture my video8 tapes and after that my stock of VHS tapes. The remainder steps of the workflow ,video8 ,vhs players, etc will be the same , only the capture/decoder device will change.

1- Use My old XP PC (Pentium 4 , 3.4 Mhz ) with Asus PCI TV7134 WDM video capture( with Philips Saa7134 )
2- Use this new startech usb3hdcap connected to a usb3 equipped PC

Both work with Virtualdub and provide the same lossless capture files namely Huffyuv YUV2 4.2.2 8bits 720x576 (Pal ) interlaced. Both provide a flawless capture without dropped or inserted frames and with low jitter. Visually and for someone with limited experience the differences are not easily detectable .
Is there some output file quality criterion more quantitative then qualitative or some tools with which one can take a solid and supported decision ?

[Brad]

Well, you can use graphs and scopes like I showed to verify whether certain errors are present. But if they look about the same to you, maybe it's best to just pick whichever workflow is easiest and move along with the project.

If you want, you can always capture a short sample of the same material using both setups and post them here for me/us to comment on (I mean literally the same frames of content).

[FLP437]

I would very much like to have your opinion about the short samples I´m attaching, if you'd be so kind.

The workflow is the same with the capture device exception.

D8 -svideo- PC XP (Philips Saa7134 ) – lossless file HuffYUV 4.2.2 YUV2 8 bits 720x576 (PAL) interlaced
D8 -svideo- Usb3HDCap – PC usb3 -lossless file HuffYUV 4.2.2 YUV2 8 bits 720x576 (PAL) interlaced

The XP PC as 2 possibilities for capturing

Microsoft WDM Image Capture (win32) (VFW)
ASUS TV7134 Capture (VFW) (Direct Show )

Both work however it seems to me that the Asus TV7134 has better image . The sample was made with asus tv7134. The capture board as the same problem as the usb3hdcap blown-out brights. I readjusted both capture cards the better that I could in order that they don’t surpass the limit values. Also the D8 camcorder was with TBC on and DNR off in both captures

[TheThrillness]

I have some questions on this histogram.

Just because a card puts data below 16 or above 235, why does this matter? Aren't the videos all going to be viewed on 0-255 monitors anyway so we don't actually lose that data on playback?

How does this correlate to 0-255 content? I have my PS3 set to Full and the driver configured for Full range and this is the result: http://1.bp.blogspot.com/-eNHwK8IjYgc/VUlvUxDZFzI/AAAAAAAAAmk/TjO8EBdLq8U/s1600/micoms...0255proper.png

As you can see, it has not clipped. If we then change the driver to 16-235, while providing a 0-255 signal, we get this: http://1.bp.blogspot.com/-oX4Tvw1AvgM/VUlvgH6C5EI/AAAAAAAAAms/kjq59EUG76A/s1600/clipci...6235bypass.png

As you can see, it now represents the full width of the histogram but clipped to hell. Obviously this is wrong to do such a thing but just thought I'd include it.

What I don't understand though is with the non clipped screenshot, am I not losing data from Full range since it only shows 16 to 235?

The only answer I can think of myself is that video should be within the 16-235 levels regardless of being Limited or Full because the media player you use expects 16-235 levels and then wants to map those values to 0-255 internally for playback. Thus it would disregard data outside 16-235 and it would be lost. Probably a wrong assumption but thought I would give it a bash anyway.

[Brad]

I would very much like to have your opinion about the short samples I´m attaching

Sorry for the delayed response. I typed up most of my reply the day of your post, but had to delay finishing it because of a medical thing.

You're certainly right that there's barely any visible difference. We're getting into serious nitpicking territory to see whether either offers a slight advantage for Video8 footage.

Neither card is showing hard clipping now, so that's great news. The clouds now have grey shading, giving them dimension.

As I alluded to before, it looks like you adjusted the Brightness so that the black borders are around 16. As a result, the black level of the video content itself is too high: their hats, pants, and the shadows are all variations of grey with no dips toward actual black.

The USB3HDCAP video is slightly sharper. I believe this is the artificial sharpening of the Proc Amp control, as there is a small white halo surrounding the left black border which isn't present on the TV7134 capture.

The histogram of the TV7134 video showed a small amount of banding even prior to my tweaks shown below. Really not an issue, but it's nicer not to have it, so a slight advantage there for the Startech.

Another very minor point: compared to the previous capture with the camcorder's TBC off, this TBC-on capture is shifted to the left slightly. The left black border remains in the same place, so the consequence is that you've lost a few image columns.

So then should you capture with the TBC off when using the USB3HDCAP? It appears so. The only place I see a difference in the line jitter correction between the two is in the head switching noise at the bottom of the frame, and here the USB3HDCAP seems to be doing a better job at it than the camcorder's TBC, with the exception of the bottom-right edge where it appears to run out of captured data to resize. (It's filled in with blackness that changes size every field.)

In summary, I would turn off the camcorder's TBC and capture with the USB3HDCAP - assuming this jitter finding holds true for your other tapes.



If you do use the TV7134, you should capture at 704x576, if possible. The way that it created this 720x file was by capturing the active 704x window and stretching it horizontally. The "tweaks" noted in the TV7134 screenshot are a quick attempt at matching it to the Startech image:

ColorYUV(off_y=-1,gain_y=9,cont_u=-10,cont_v=-10)
LanczosResize(704,height).AddBorders(8,0,8,0,color_white).SeparateFields()

If you want to post some VHS samples, I wouldn't mind taking a look...

[Brad]

Just because a card puts data below 16 or above 235, why does this matter? Aren't the videos all going to be viewed on 0-255 monitors anyway so we don't actually lose that data on playback?

You basically answered yourself at the end of your post, but to expand slightly: the standard for video is that black is 16 and 235 is white. If your monitor and video card are set up for 0-255, your standard media player by default will assume that the content is using video levels: it will clip superblacks and superwhites, put 16 at 0, put 235 at 255, and expand everything in between to fit. The same goes for hardware players outputting RGB, which are less configurable if you want them to do something different.

How does this correlate to 0-255 content? I have my PS3 set to Full and the driver configured for Full range and this is the result: http://1.bp.blogspot.com/-eNHwK8IjYgc/VUlvUxDZFzI/AAAAAAAAAmk/TjO8EBdLq8U/s1600/micoms...0255proper.png

Assuming this is video game content that contains values of 0-255, as your next screenshot implies, the driver is compressing down to video levels.

As you can see, it has not clipped. If we then change the driver to 16-235, while providing a 0-255 signal, we get this: http://1.bp.blogspot.com/-oX4Tvw1AvgM/VUlvgH6C5EI/AAAAAAAAAms/kjq59EUG76A/s1600/clipci...6235bypass.png

As you can see, it now represents the full width of the histogram but clipped to hell. Obviously this is wrong to do such a thing but just thought I'd include it.

I'm not sure I agree this is wrong when your source is "Full". But yes, it will mess things up if you try to play it back like ordinary video content.

What I don't understand though is with the non clipped screenshot, am I not losing data from Full range since it only shows 16 to 235?

You're missing certain in-between values. A direct comparison using a gradient should show it fairly clearly, I would think. But the driver likely also adds dithering.

The only answer I can think of myself is that video should be within the 16-235 levels regardless of being Limited or Full because the media player you use expects 16-235 levels

Better yet, just use Limited so that you keep video levels from the source to the capture.

[TheThrillness]

Thanks, makes sense.

Better yet, just use Limited so that you keep video levels from the source to the capture.

Sometimes tough to do this depending on your setup but agree yeah. E.g. Monitor being full or some capture cards only being able to select full. Mostly only a problem with game capture ofcourse.

[Saturn2888]

Is it just me or has anyone else noticed this is a 1080i59.94 device instead of 1080p60 as advertised?

I'm taking HDMI output from my 60fps camcorder, and I'm assuming the output during non-recording is actually 1080p60, but the USB3HDCAP itself is showing 1080i59.94. Plugging into other devices like the AVerMedia ExtremeCap U3, it runs at 1080p60.

EDIT:
Seems there are some quirks getting the input as 60fps w/ the AVerMedia ExtremeCap U3. The camcorder's HDMI-out is probably 59.94fps which is why it's showing that. Still, I'm only getting 540 lines instead of 1080 lines when looking at the input in the drivers. That would be interlaced video wouldn't it?

[newpball]

Just because a card puts data below 16 or above 235, why does this matter? Aren't the videos all going to be viewed on 0-255 monitors anyway so we don't actually lose that data on playback?

Once would expect so right, especially in the 21th century?

Nope does not work, at least not with YUV, video really needs all that overhead below 16 and above 235 (really, trust brilliant engineers on that!).

You could use all the bits by capturing RGB but the problem is that RGB video is effectively blacklisted by most media players.

[FLP437]

I will post some vhs captures sanples with the usb3hdcap in a week or so when I will have a JVC SVHS player do do it

[FLP437]

As member vaporeon800 as expressed interest to see a VHS capture ( standard tape not SVHS) c with the USB3HDCAP I´m attaching such a sample. The capture as been made with a combo Minidv / SVHS deck from JVC model HR-DVS3EU. Digital 3R, Digital TBC/DNR and video stabilizer were all off.

If possible I welcome your comments as before related to this capture.

I have not yet decided if I will do my analog video8 recaptures ( about 60 ) and VHS first time captures( about 200-250 ) with the usb3hdcap device .Here at videohelp The Thrillness expressed a quite favorable opinion ( however perhaps more related to digital and games captures ), vaporeon800 as also expressed a favorable opinion even for analog capture . However at DigitalFaq forum on a thread about a different matter but that turned out discussing also capture devices at the end, the opinion was not as favorable namely for analog capture and a lot of defaults were pointed out. In that forum there is a list of recommended capture cards but the list is not updated as it is almost impossible to buy any of the recommended cards and in the European market is even worse to get any of them.

As so, I´m in doubt if I will do my captures with this device or if I must buy a new one. I have a lot of old legacy capture devices but I don’t think any of them will give better results than the usb3hdcap (Philips Saa7134 PCI capture decoder, Dazzle DVC100 usb, Hauppage Win/tv 878/9 and one more recent the terratec G3 ) all of them allow lossless capture. If I search amazon I can see a limited offer (Hauppauge WinTV HVR-2250,Hauppauge Colossus, BLACKMAGIC DESIGN INTENSITY PRO ,Osprey 100e Video Capture Card, elgato video capture usb, diamond vc 500 usb, roxio usb and some more, but none of the recommended ones) and i didn't found good reviews for any of them .

I have saw some posts by vaporeon800 that analyzed and tested several recent captures cards / devices perhaps you can enlighten me if I can use one of my old legacy devices , use the usbb3hdcap or by a new one and which one to get good results in analog capture. As I have a lot of work ahead to do I don’t want to commit myself to the bulk work before I have decided the capture device, if not I have to recapture again as is is already the case with my video8. The other elements of the workflow are stable only the capture device is still not decided.

[Brad]

Great, thanks for posting it!

I'm a bit disappointed that the head-switching skew is uncorrected, as with your SLV-T2000 sample on DigitalFAQ. But the rest of the image looks stable.

You've set the brightness a bit too high, again. If you address that, I don't see any reason not to go ahead and do your captures with the USB3HDCAP.

[LMotlow]

That seems OK for VHS. But what about the analog-8 tapes? Seems like playback with those tapes gets his capture setup spooked. I started to think it might be a tape or two, but all of those v8 tapes can't have the same problem. Maybe the original camera was loco.

[Brad]

Besides the unadjusted proc amp, I'm not seeing a problem with the Video8 samples posted here.

[LMotlow]

Besides the unadjusted proc amp, I'm not seeing a problem with the Video8 samples posted here.

Oh. And to think that for all these years I've been misled into believing that bad line twitter and aliasing were problems, along with some subtle field shifts. I stand corrected, I now see that these noisey artifacts are natural. But I don't see them in the VHS captures. How would the O.P. add this noise to the VHS caps? If the other captures have no problems, it seems something is missing from the VHS caps.

[Brad]

Classy. Some images would be more elucidating than snark.

[LMotlow]

Post #38 already has images.

[FLP437]

Related to other possibilities that can boost my analog captures and increment quality from the level I can get from this usb3hdcap I think 3 brands could eventually provide better results matrox, black magix and Aja kona .

The first two have more accessible prices however matrox mx02 HD is not available any more and I don’t know if any other card could be recommended and if we can still get it, Black magic, I saw a recent thread from vaporeon800 and he was not happy with the intensity pro 4k namely the driver. And the normal intensity pro didn´t have good reviews to.

Related to black magic design intensity pro 4k, as this card any serious advantage in what analogic is concerned versus the usb3hdcap ?

Related to the Aja Kona, this is the most expensive but perhaps the only one I can buy if I want to ( there is a limited offer at amazon ). Is expensive, but if I considered it in terms of a per cassette basis, not so much ( as I have a large number ) and I could always try to resell it after captures .However are the captures sufficiently distinctive to justify buying it. And does it work on a PCIe bus on a windows 8.1 environment without drivers problems, can we still use virtualdub or are we obliged to use special software , are there other hidden costs software, accessories and how about the learning curve is it easy to use or a complicated piece of hardware.

[Brad]

Post #38 already has images.

Where?

Related to black magic design intensity pro 4k, as this card any serious advantage in what analogic is concerned versus the usb3hdcap ?

Based on my experience with it, all I see are disadvantages. For example, you stated that you haven't experienced dropped frames with your USB3HDCAP. The Intensity Pro 4K did drop frames when I fed it an unstable VHS signal.

I've never seen anyone here report on using the MXO2 or AJA Kona for analog tape capture.

[FLP437]

What as worried me have been the negative opinions on digitalfaq related to video8 captures however perhaps original camera and source tapes are in the origin of some problems and not the usb3hdcap itself, also it could be difficult to distinguish between source related problem and intrinsic problems associated with the device .My problem is i don't want to start over , if tomorrow I find a better device and have to recapture everything again ( 250-300 tapes )

So in your opinion, and so far, my best bet is to go ahead and use this device as there is nothing currently on sale in this price range ( until 250 usd , even more if we speak of matrox and black magic ) that can provide better results and also it is very difficult or impossible to grab one of the specials legacy ATI all in wonder 6XX series.

As i still have some problem doing a correct adjustment, do you know of a iso file to burn a dvd or files with video patterns that I can use to better try to adjust the video proc. amp. thanks in advance

[LMotlow]

Post #38 already has images.

Where?

With two video captures and hundreds of frames, you can't see any images with bad line twitter and aliasing? That being the case, I don't see the purpose of making up 2x and 3x images for something that should be obvious, especially to more experienced members here.

Related to black magic design intensity pro 4k, as this card any serious advantage in what analogic is concerned versus the usb3hdcap ?

Based on my experience with it, all I see are disadvantages. For example, you stated that you haven't experienced dropped frames with your USB3HDCAP. The Intensity Pro 4K did drop frames when I fed it an unstable VHS signal.

I've never seen anyone here report on using the MXO2 or AJA Kona for analog tape capture.

I've seen where Aja can be used for losseless VHS capture, but paying $100 for VHS capability and $1100 for features that will never be used doesn't seem reasonable. I also see that it requires a ton of associated software and cryptic settings. For that amount of money, you could get or build your own XP or Win7 PC and find any of several capture cards that would be compatible, and still have cash for more.

What as worried me have been the negative opinions on digitalfaq related to video8 captures however perhaps original camera and source tapes are in the origin of some problems and not the usb3hdcap itself

That is possible. Perhaps the original camera was at fault or wasn't tracking properly.

I saw those digitalfaq posts myself (not a paid member there, so I decline to post but have visited since 2004). I suppose the usb3hdcap is useful, but I'd say you're limited to ultimately outputting progressive video, because attempts to deinterlace or separate fields for cleanup will look worse when those fields or frames are reassembled. I wouldn't try to use a capture card to deinterlace, none of them do it well. I'd suggest that you connect your playback camera to a TV, if you can find one that will accept its output cable, and see how those tapes play directly from the camera. You could also connect the camera to a DVD recorder's inputs and see if that playback would have the same field and interlace problems. The camera and usb3hdcap do output nice color, even if red is a bit pumped and the overall effect tends toward plasticky DV, but the color overall isn't bad.

[FLP437]

LMotlow you said one phrase that catch my attention “ the original camera could be faulty or wasn´t tracking properly” . At least the last 11 tapes from a total of 60 have serious tracking error as apparently the tape guides got loosened and I got the footage serious compromised .

However in the original camera the tapes played normally and I never remarked special problems at the time ( 20-25 years ago ). Perhaps even before this more serious problem that got worse with time the camera had already some minor problem . However the camera doesn´t work anymore it was 26 years old and had only composite as output and I have disposed it about 20 years ago.

I still have a TV with a svideo input I will try it with the D8 camera and see the result . I also have a HDD/DVD recorder that as also DV input but the dvd recording part is not working properly , I have send it for repair , I hope it can be repaired it is about 10-11 years old and spares are already limited. If I can get it repaired I will try to.

How can I get a capture with a progressive output with virtualdub . All my captures have resulted in interlaced files.

[LMotlow]

Always capture interlaced as interlaced. Deinterlacing is for post-processing. Remember that deinterlacing does have a cost, even it is subtle, and it doubles the frame rate. 50 fps is not valid for PAL DVD, BluRay, or AVCHD. But many players can handle 50fps standard definition video encoded as mkv or mp4, and newer players can likely handle SD MPEG at 50fps.

I have a few old tapes that just laugh at my attempts to restore them in MPEG or h264 without some skipping or judder. Happy to say that my OPPO BD player handles them like a champ.

[Brad]

As i still have some problem doing a correct adjustment, do you know of a iso file to burn a dvd or files with video patterns that I can use to better try to adjust the video proc. amp.

The test patterns I mentioned earlier can help you verify the neutral setting of the sharpness control, but for the other adjustments you should just be setting Brightness and Contrast based on the levels you're getting from each tape.

something that should be obvious, especially to more experienced members here.

None of them have interjected to agree with either of us. Hopefully we won't lead FLP437 astray.

[LMotlow]

As i still have some problem doing a correct adjustment, do you know of a iso file to burn a dvd or files with video patterns that I can use to better try to adjust the video proc. amp.

The test patterns I mentioned earlier can help you verify the neutral setting of the sharpness control, but for the other adjustments you should just be setting Brightness and Contrast based on the levels you're getting from each tape.

Gotta remember that home made analog tape, and lots of retail issues as well, are hyperactive about changing levels and color balance with every shot. The best you can do during capture is find a happy medium and correct after capture. The levels in the parade clip look OK to be (they're at RGB 16-235). There's some mild bright blowout, but that's analog tape for you. Best to set capture levels by cropping off black borders in VDub, set levels, then don't forget to disable the cropping. Takes some fiddling, but you'll get the hang of it.

something that should be obvious, especially to more experienced members here.

None of them have interjected to agree with either of us. Hopefully we won't lead FLP437 astray.

Okay, point about snarkness taken, so let's try some images -- by popular demand, more or less.

In the capture usb3hdcap_1.avi I see a ,ot of line twitter and aliasing, especially on movement and especially on fine angular lines. When something moves the edge bristle is really noisy IMO. I think it's on the tapes. Deinterlace, decomb, anti-alias, whatnot, nothing does anything for it. There's both aliasing and broken lines.

Below is the original frame 617 of the USB3 clip, as an example. Focus on the bright archway column in the lower left:


Below is a 2x blowup of a portion of that frame. You can see the angular lines are broken, and excessive combing. The breaks are different in every field for several frames.


You'd think deinterlacing might smooth that out, but it doesn't. Below are EVEN (left) and ODD (right) fields as deinterlaced with Bob() and 2X blowup:


The same EVEN and ODD frames with yadif, 2X blowup:


The same two deinterlaced with QTGMC (slow). Now, we might be getting somewhere with this:


Below, after interlace with QTGMC and re-interlaced. Looks like we're almost back where we started.


This is one of the more obvious examples, but there are many others large and small. Maybe there's a way to fix this, but I gave up on it except for keeping the video progressive.

A few other problems, too, like the odd artifact below. Notice how red turns bright green every time a red uniform enters a certain area along the right border. I suspect that's in the original, too.

original frame 357:


frame 1157 (deinterlaced):


Attached is higher bitrate mp4 at progressive 50p with most of the noisy stuff gone. AGC in the camera changes the color balance, especially in the reds, as the camera zooms back at about frame 350 (~frame 700 in the progressive clip). You can see luma take a quick dip when the bright archway enters the picture right around frame 550 or so (frame 1100 in the progressive clip). Guess we'll have to live with those changes.