Help me choose a cheap PCIE or USB analog SD capture device for VirtualDub

[crespo80]

Hello everyone, I'm from Italy

I have a Philips VCR in good working conditions and my only target is to capture 50 VHS-C tapes at the maximum possible quality (capture with lagarith compression, then deinterlace with QTGMC in avisynth and compress with x264 in MeGUI).

I want a cheap solution, compatible with VirtualDub (both audio and video) and Windows 8 64bit (my PC specs on my profile).
I cannot find any cheap PCI-E capture card (I don't have PCI slots), is there any good in any of these 30/50€ USB devices, are they compatible with Virtualdub?

- Terratec Grabby USB
- Avermedia DVD Ezmaker USB
- Hauppauge USB LIVE2

There's the Terratec AV300MX for about 50€ but how about Virtualdub?

Is it worth it to double my budget to 80/90€ and look at these?

- Terratec AV 350 MX which is Virtualdub guaranteed on their site.
- Pinnacle Dazzle DVC 100/101/103/107 (64bit OS driver) but I don't know the differences and the compatibility with VirtualDub.

I can find a used Pinnacle Dazzle DVC101 for aroud 40/50€, is it a good choice?

Thanks in advance for your help

[Brad]

You may be limiting the field by using Windows 8 instead of Win7; I don't know personally how the compatibility works.

In terms of the analog-to-digital converter chips:

• Terratec Grabby uses SAA711x, most likely Philips / NXP SAA7113 commonly used in video capture devices for over a decade.
• Hauppauge USB-Live2 uses Conexant CX23102.
• There are several AVerMedia DVD EZMaker devices: USB Plus, USB Gold (C038), 7 (C039).
• Terratec Grabster AV 300 uses Silan SC8113 (same as $8 Easycap devices).
• Terratec Grabster AV 350 MX uses TI TVP5150 (same as ATI 600 USB).

There is a lot of hatred towards Dazzle devices, and some confusion as to the innards. Can't be bothered looking into it right now.

Is importing an option, or is the Italian postal system as bad as I hear?

[hech54]

The Hauppauge does capture AVI with DScaler. I got it to work ONCE with VDub. As a matter of fact the only reason I tried DScaler is because I could no longer get the Hauppauge to work with VDub for some reason. Still haven't figured out why. Probably user error.

[crespo80]

Thank you both for the precious infos.
So I'll avoid the Dazzle series (I also had driver problems with Pinnacle) and the higher priced Terratec G1/G3/AV300/AV350 which aren't worth the extra money over the cheap Grabby.
So the only two USB capture devices worth their price are the Hauppage USB-LIVE2 and the Terratec Grabby, don't know which should provide better compatibility with VirtualDub.

Importing from outside EU can be a problem because the italian customs is slow as hell, the shipping cost is too high for such low-value devices and I lose any warranty, but if there's a really valid option I can think about it


Searching for PCI-E cards, I found an Hauppauge WinTV HVR-1250 PCI-Express for 40€ shipped http://www.ebay.it/itm/120967943994?ssPageName=STRK:MEWAX:IT&_trksid=p3984.m1438.l2649
The drivers are WHQL, updated in 2013 and 64bit compatible.
It should be another Conexant CX23885 based card, it's a NTSC version but I assume it's only for the TV signal, I should be able to use the composite/S-Video input with my PAL recorder, is that right?

I also found an Hauppauge WinTV HVR-3300 PCI-Express new for 45€ shipped
http://www.amazon.de/Hauppauge-WinTV-HVR-3300-Tuner-Karte-Bulk/dp/B003XU0Q2Q/ref=cm_cr...pr_product_top
The drivers should be the same of the HVR-1250 because on hauppauge.it it says they're suited for all HVR-12x0-14x0-17x0-33x0-44x0-55x0, though they are outdated and I should use the ones for the HVR-1250 on hauppauge.com, is that right?

Should I also consider the inexpensive Genuine EzCap I can get for 30€ shipped from the official ebay store?http://www.ebay.it/itm/EZCAP-TV-116-EzCAPTURE-DC60-Full-retail-1yr-warranty-/121187927...75d0512&_uhb=1



There are also these other cards, but I don't know if they're worth the extra money
LifeView LV8H (CX23885) 60€
Hauppauge impactVCB-E (CX23885) 73€
Avermedia avertv HD (AD9985A) 82€

[Brad]

If you mean the AVerMedia AVerTV HD DVR (C027), don't buy it for analog capture. The ADC is actually Trident TM6200. It adds bad sharpening that can't be disabled, and basically requires that you use a TBC all the time for workable VHS capture.

I would imagine that the CX23885 is a step down from the CX23102 since it is from the previous generation, but can't say for sure. I bought a CX23883 card yesterday but need to get it running and test it out.

[PuzZLeR]

I have a Philips VCR in good working conditions and my only target is to capture 50 VHS-C tapes at the maximum possible quality (capture with lagarith compression, then deinterlace with QTGMC in avisynth and compress with x264 in MeGUI).

Lagarith, or HuffYUV, or UT is good for lossless. Don't deinterlace for archiving, only if the playback medium requires it. Regardless, you should always archive the interlaced orignal, ideally in lossless, but if the big HDD space available today isn't suitable for you, you can encode it to smaller files (for archiving), such as DV, or high bitrate MPEG-2.

You can do whatever you want in encoding for your playback needs (delivery), but keep it in a good, and true-to-original-interlaced format for archiving. Never confuse acquisition/archiving with delivery. They are two separate entities and storage requirements.

Keep note of the italics.

I want a cheap solution, compatible with VirtualDub (both audio and video) and Windows 8 64bit

For VHS captures, all you really need is a good SD capture stick. Forget gutting the PC for the slots. Just use a good proven USB device, such as the Hauppauge USB-Live2, the authentic ezcap.tv, or, if you wanna eBay shop, you can look into the ubiquitous ATI 600 USB. Anything higher end is overkill for VHS.

All are compatible with VirtualDub for lossless, and more. However, as for Windows 8, I have no clue. That operating system scares me, and the direction MS is going with it (for today's ADD generation) is scary, to be honest.

I can find a used Pinnacle Dazzle DVC101 for aroud 40/50€, is it a good choice?

No.

There is a lot of hatred towards Dazzle devices, and some confusion as to the innards. Can't be bothered looking into it right now.

One of my favorite sayings: The Dazzle products are so bad, you can blame a divorce on them.

So I'll avoid the Dazzle series.

Your marriage is safe.

Should I also consider the inexpensive Genuine EzCap I can get for 30€ shipped from the official ebay store?

Didn't check the link, but if it's from


http://www.ezcap.tv/


then it's a good product. Will work for you.

There are also these other cards, but I don't know if they're worth the extra money

For VHS capture, anything beyond is overkill. Don't bother. Save your money.

[crespo80]

Ok, so I decided to go for the genuine Ezcap.Tv because it only has the basic features I need and the right price, I have 1 year warranty, I can easily find support from thousands of users, it is compatible with Windows 8 64bit, and you convinced me that anything beyond that is overkill for VHS capture (and probably an hassle driver-wise).
It really seems a clone of the Terratec Grabby (or maybe it's the opposite ), it has the same Empia EM2860 USB chip and Philips SAA7113 ADC.
I ordered it from the official channel and should receive it in a few days, thank you all for the help, it was really helpful!

As for the archive/delivery matter, I think I'll save a barely compressed (probably Mjpeg) true-to-original interlaced version, and then compress it from there for delivery (x264 and QTGMC deinterlaced)

[hech54]

I can easily find support from thousands of users

Much of the discussion you see in the internet are frustrated people who bought the fake "EasyCap". You should not have major problems with an original.
90% of the stuff sold in between the original from YEARS ago to the EzCapTV version today were absolute garbage.

[crespo80]

I see they closed their forum so that only confirmed purchasers of an authentic product can have access to it, I should receive my invitation inside the box.The only thing that puzzles me is the driver: they say the ezcap is compatible up to Windows 8 64bit, but the driver on their site is from 2009. Should I install it or maybe a standard driver is included in Windows itself?

[PuzZLeR]

Hi again.

Hope the ezcap.tv works for you, and I'm glad you avoided all that trash that was fake. Many of them sell for, like $5, but for the headaches and bad results you get in the end they'd have to pay YOU to use the crummy thing. The extra money you spent on the authentic version, I assure you, is worth it.

I currently have no experience with Windows 8, so hopefully that's not an issue for you. But I can offer a few more tips.

If you can reduce the sharpness level of the ezcap.tv, that would help. The default level, in my opinion, seems a little bit high. Then again, some like it that way.

Does your Phillips have an S-Video option from your video-out? That would be a better choice over RCA/composite. If you have a VCR that does this, you can get an adapter to play your VHS-C tapes. S-video is usually a better choice (for the video capture).

As well, the ezcap.tv is not the greatest at audio capture (in fact, most others aren't either). But that's not a problem - I recommend you capture your audio, separately, through your sound card, such as with an RCA/composite-to-3.5mm cable from the audio-out of your VCR. Your onboard sound card should be sufficient, even if your tapes have SQPB audio. You will get optimal results with the "line in" at 48KHz, 16bit, stereo, uncompressed PCM wave.

VirtualDub allows you to use your sound card, and these settings.

If you need a new sound card, you can get one dirt cheap, such as an Asus (especially now that you've freed an internal slot ). The more expensive ones only really offer more KHz and bits, but this too will be overkill for VHS audio.

If you have about 50 VHS-C tapes at full time (was 180min, or even 240min, the max they came up with?), without edits or omissions, you're looking at about 5TB (give or take) of storage with Lagarith. This is minor today, and only a couple of external drives (plus backups), with physical space less than a handful of tapes. You may not even need to bother with MJPEG.

This is important to note if the content is precious to you, and especially more so if you plan to dispose of the tapes.

In other words, go ahead and use x264, and deinterlace, and it's a good idea if it meets your playback means. But do keep the lossless Source separately, and keep it interlaced (since de-interlacing, as you know, is destructive to quality, luma/chroma info, temporal resolution, etc.)

[hech54]

I see they closed their forum so that only confirmed purchasers of an authentic product can have access to it, I should receive my invitation inside the box.The only thing that puzzles me is the driver: they say the ezcap is compatible up to Windows 8 64bit, but the driver on their site is from 2009. Should I install it or maybe a standard driver is included in Windows itself?

You should be concerning yourself with things you can actually do right now. Make sure the heads are clean on the VCR.....etc etc etc.

[crespo80]

@Puzzler
Thanks for the tips, the VCR is a standard VHS recorder, so only composite output.
As for the sound, I'll make some tests and see if the quality is so bas as you say. My primary concern is about synchronization, so I'd prefer using the Ezcap inputs, but I also have an Asus Xonar sound card (plus the Realtek integrated in the mainboard).

Regarding the lenght of my tapes, they are standard 30/45 min VHS-C tapes, and I've never seen any above 60 min
For economic reasons, I want to use only two 1TB hard disks I already own to store the captures (each containing all the captures, for backup purposes), so I'm also considering to capture in Xvid all-I format (target quantizer=2.0, interlaced): I made some tests in the past and it outperforms MJpeg (better compression with less artifacts) while being very fast and it's almost the same quality as the lossless formats. I can nitpick differences only watchin the encodes frame-by-frame in VirtualDub enlarged 200%, and it uses 1/4 of the space!


@hech54
The VHS Philips recorder purchased some years ago but it bas barely used, and kept with care, it's like new.
The heads are likely immaculate, but I hope inactivity did make no harm to the internal mechanics!

[PuzZLeR]

For a giggle, not too long ago, I did experiment with Xvid capture (at higher bitrates/lower quantizer, interlaced) - it was surprisingly quite good! You do notice a bit more blockiness than with lossless, as expected, but yeah, you'd need to compare multiple instances of VirtualDub at 200% frame-by-frame (which is what I too use alot in my comparisons). Then again, you can always look into an Avisynth deblocker for your x264 encodes later if the blockiness is an issue.

I'm a purist using lossless, and would find Xvid a bit more difficult to edit with some features above cuts/joins, and still quite inferior with non-stills, but you will definitely get more value/bit with Xvid in quality, and your project should fit easily under 1TB-2TB.

I too have an Asus Xonar. I leave it up to you, but if you don't like the sound coming from your eazycap.tv, you at least have an alternative using your "line in" port. Perfect sync every time for me.

[hech54]

@hech54
The VHS Philips recorder purchased some years ago but it bas barely used, and kept with care, it's like new.
The heads are likely immaculate, but I hope inactivity did make no harm to the internal mechanics!

My point is....don't confuse yourself with issues about drivers and stuff until you get it.
Too many forum users LOVE to impress complete strangers with their knowledge of stuff that that does not concern you at all.
One step at a time.

[Brad]

I can nitpick differences only watchin the encodes frame-by-frame in VirtualDub enlarged 200%

Fair enough, but do keep in mind that they will be enlarged 225% for playback on a 1080p screen (and more once we get into 2160p!).

[crespo80]

One step at a time.

Got it

Fair enough, but do keep in mind that they will be enlarged 225% for playback on a 1080p screen (and more once we get into 2160p!).

I haven't thought about that maybe I'll go with Target Quantizer 1 instead of 2 (it's double the size but still half that of Lagarith/UT) that's almost identical to the original, with even every single pixel of noise intact (what a luck )

[jagabo]

almost identical to the original, with even every single pixel of noise intact (what a luck )

And as a bonus you get lots of DCT ringing artifacts and 4:2:0 chroma subsampling.

[crespo80]

I captured some screens in virtualdub at 400% magnification.
I can't spot ANY difference between the original and the one compressed with xvid Target Quantizer 1 (which is half the size of lagarith).TQ 2 seems to apply some light denoising but it's still quite perfect (considering it's half the size of TQ 1)


http://www.ilculto.it/cesare/images/test_original.png
http://www.ilculto.it/cesare/images/test_xvid_motionsearch-off_tq1.png
http://www.ilculto.it/cesare/images/test_xvid_motionsearch-off_tq2.png

[PuzZLeR]

I know I said Xvid is surprisingly good for capturing, and it still is among the best in quality/bitrate, and I was trying not to be insensitive about your budget restriction, but let me clarify about me being a lossless purist as well.

Maybe frame-by-frame reveals little difference for many scenes but, keep in mind, compression artifacts, such as those from a highly compressed medium like Xvid, will reveal a few more artifacts during motion (echoing, blurring, blocking, and more).

Those pictures look like stills. What about for fast moving scenes? Being that this is VHS-C, used typically for home video shoots, I would bet that there are many camera panning movements as well. This can be devastating with Xvid, or even MPEG. DV is better suited for that, however lossless is still superior.

I personally do use some high-compression formats for caps, but only for B or C level stuff I don't want to think deeply about - such to settle an indecisiveness about keeping some dorky TV recording I have a passive interest for - definitely not for precious irreplaceable content (like I'm assuming yours must be).

You have to be a purist for such content.

I understand about money, but try and rethink it. If money's tight, look at it as an investment, especially when HDD space today has never been more conceivable at price/GB. I say this, because Xvid today may be a decision you will regret ten years from now if you chuck your tapes.

[crespo80]

I can't follow you

When you disable motion search, xvid is like Mjpeg (but with better compression IMHO), there's no difference if a scene is slow or fast moving, because every frame is compressed indipendently from the others, just like a sequence of jpeg images, all frames are Key frames.
And if you use a target quantizer of 1, it's encoding every frame always at the maximum possible quality!
The pictures I posted look like stills because they ARE stills

Years ago I captured some VHS-C tapes (with lot of noise and shaky camera movements) and I used xvid with these settings for some of them (for test purposes, my old CPU usage was too high to use it reliably back then): absolutely perfect results, and good for editing too, because it's all I frames, no B nor P ones!

[PuzZLeR]

Ahh, I see your point now, missed when you mentioned all-I frames on Xvid.

That's why in my earlier post, I said DV is better suited for camera panning movements (and editting) - due to its intra-frame design.

But I never tested Xvid capture with only I-frames. I do notice many artifacts, even with high bitrates, but would be curious if many will still exist with key frames enforced every 1. Will be doing some interesting testing now in the next day or so.

[crespo80]

Just a few tips if you want to test it (obviously use only Target Quantizer encoding method): from my tests I found xvid can only use integer quantizers, even if you can choose decimals: if you choose TargetQuantizer=1.1, xvid will encode 9 frames with quantizer=1 and 1 frame with quantizer=2, then other 9 frames with quantizer=1 and 1 frame with quantizer=2 and so on... if you choose TargetQuantizer=1.5, it will encode 1 frame with quant=1 and 1 frame with quant=2 and so on...

With TargetQuantizer=1 I cannot see any difference whatsoever from the original, with TQ=2 it seems there's only some very light denoise applied to the image, but still no artifacts. Only from TQ=4 I can start to see some compression artifacts.

My settings: QuantizationType=MPEG (the default H.263 is optimized for low bitrates and softens the image), AdaptiveQuantization=OFF, InterlacedEncoding=Flagged, TargetQuantizer=1.00, MotionSearchPrecision=0 (this disables the motion search and allows all-I encoding)

[jagabo]

I never tested Xvid capture with only I-frames. I do notice many artifacts, even with high bitrates, but would be curious if many will still exist with key frames enforced every 1.

You will still get DCT ringing artifacts (all DCT based codecs will produce this) and 4:2:0 chroma blur (if your source isn't 4:2:0 already).

RGB 4:4:4 source:


RGB 4:4:4 video encoded with Xvid all I frames:


Contrast stretched to bring out the DCT ringing:


Be sure to view the images at full size with 1:1 pixel mapping.

Of course, the DCT ringing isn't as obvious with real world video capture because you don't have sharp edges and flat shading like in the above images. And all the noise in analog caps helps hide artifacts. But you should know what you're getting into and what to look for.

[PuzZLeR]

Hi Jagabo.

I should expect the typical DCT ringing either way (and much more so with encoding over analog captures) which is why I'm still sticking with lossless for precious content. But a viable storage compromise for B or C level content is always neat.

I'm curious how an all-keyframe enforced Xvid will perform in motion, particularly where camera pans and fast moving scenes would yield an eyesore of blocks and blurs otherwise.

(At work now, but will test it soon enough.)

[crespo80]

Interesting test image Jagabo, I tried to compress it in Jpeg (the best known DCT-based image compression algorithm) at maximum quality 12 and it's perfectly identical to the original, no ringing artifacts.




I then created a 1 second long video with avisynth from the PNG image and then compressed it in virtualdub with lagarith in RGB, YUY2 and YV12, and then compressed with xvid TQ=1 YV12 (the only color space available).



LAGARITH RGB



http://www.ilculto.it/cesare/images/encoding_tests/rgb444.rgb.lagarith.avi



LAGARITH YUY2

http://www.ilculto.it/cesare/images/encoding_tests/rgb444.yuy2.lagarith.avi



LAGARITH YV12

http://www.ilculto.it/cesare/images/encoding_tests/rgb444.yv12.lagarith.avi



XVID YV12



http://www.ilculto.it/cesare/images/encoding_tests/rgb444.yv12.xvid.avi



In the xvid video there are no ringing artifacts, but there are chroma subsampling compression artifacts, just like the lagarith YV12 version.
All in all, since my capture card already works in YUY2 colorspace and my signal is composite, the YV12 compression of xvid doesn't make any harm to the video, and the Target Quantizer set to 1 causes no ringing artifacts


PS
it's interesting to note that, in this specific case where all the frames of the video are identical, the lagarith version is waaaaay smaller than the xvid, because it uses motion analysis algorithms, while xvid does't because I disabled motion search

[jagabo]

In the xvid video there are no ringing artifacts

Yes there are. If you look closely you can see them. Increase the contrast and they'll really show up. Your Xvid image with contrast stretch:

it's interesting to note that the lagarith video is waaaaay smaller than the xvid, because it uses motion analysis algorithms

No. Lagarith has no motion analysis. But it recognizes when a frame is identical to the last and encodes it as a "null frame" which takes almost no bitrate. Also, flat, computer generated graphics compress much better with entropy encoders than with DCT encoders.

[PuzZLeR]

Yes, a closeup, even using a simple Windows Picture and Fax Viewer, reveals the artifacts.

This is all fine and dandy, and I expect such artifacts, but what about motion scenes and camera pans of real video? Will Xvid, when using only I-frames, perform significantly better (without those increased blocks and blurs)?

This is what I'd like to test when I get home.

[crespo80]

jagabo you're right, by doing contrast strecthing on the images I can now see the artifacts even on the quality-12 jpeg image

But I think I'll continue using xvid anyway, because my eyes can't see the difference if I don't manipulate the output and look very closely, and QTGMC doesn't care too much and will always output the same results (again, only if I look at them in virtualdub at 400% zoom and not in photoshop at 800% zoom with contrast stretching).
At least, now I am more aware of the limits of DCT-based compression algorithms, thank you for your good explanation!


PS
I didn't know how the lagarith algorithm worked, I just guessed based on my limited experience

[jagabo]

By the way, for those using slower computers, Divx is twice as fast as Xvid when both are set at their fastest settings. Compression and artifacting are similar.

[crespo80]

I tried Divx as well, but I couldn't find an all-I setting, do you know how to achieve that? (if it is possible with the non-PRO version)