Comparing capture cards handling lousy tapes?
8 Attachment(s)
Warning: long post ahead
I wanted to make this post for a while, hoping it could better illustrate the use of a TBC or TBCish device can have. It took some more time than I hoped as the VCR I was using decided to develop some mechanical issue and wrinkle the tape halfway through, so I ended up having to restart with another one. I'm not particularly satisfied with the captures, but hopefully they can be of some help at least. I would also encourage others to post some comparisons if they are able. The tape I used is some off-air recording of the movie "Reisen til julestjernen" (don't think it has an english title). It's telecined PAL (24 fps (presumably) movie sped up to 25fps), so not much combing. The telecine process doesn't seem to have been great and the video is also quite noisy. Virtualsettings were mostly the same as in the guide, I'll write them down later. Clips are HEVC/h265 to keep file size down (and with 4:2:2 subsampling to preserve chroma). Can provide lossless or single frame pngs if there is interest. Caveats: This is just one tape, and there are way worse tapes out there, though this one is at least dodgy enough to give the capture cards some trouble. It doesn't demonstrate horizontal jitter all that well, but enough to show some difference at least. I didn't do a lot of level tweaking so the levels are not going to be equal on all tapes and there could at worst be some clipping in some clips. I used a Samsung SV-261X, which is a very basic 2-head VHS deck from the early 2000s. I wanted to use something basic, and I needed something that wasn't too heavy so I ended up with this. As it's a plain VHS deck, the capture is done using composite video, which also gives some idea of how capable the comb filters in the devices are, though it's not really a very good test of them. Didn't have any portable working S-VHS deck and the tape is a full-size VHS tape, so couldn't use a camera. The tape seemed to get a bit worn after a few plays, and the start of the tape is a bit crinkled so the video isn't going to be entirely the same on all the recordings, especially in the beginning, which can throws off the comparison a little bit. Also note that dealing with bad signals well does not imply good image quality (which this comparison isn't focused on), and vice versa. A longer capture would better illustrate audio sync problems, as that usually adds up over time. --- Samples were captured with Virtualdub, except stuff captured using the blackmagic card where amarectv was used as I have had some issues with Vdub on the specific computer I used it on after an update. Used VC500 to capture from pass-through/TBCs using S-Video besides the previously mentioned. Linux samples were captured with ffmpeg. Capture cards: Note that USB dongles bearing the same name and casing can have hardware inside.
As you can see, the raw signal from the VCR gives the capture cards trouble to varying degrees. Will add more posts with TBCs and DVRs tomorrow, which may be the more interesting bit. Also got two older pci cards that I haven't made samples for yet. |
Mp4 lossy encoding makes these tests dubious, IMO. But if this is a test of lossy capture and encoding using different capture software, the tests are still inconclusive. The mpo4 encoder adeds another layer to consider. Still, congrats on getting this work done. Something to think about.
The BlackMagic cap looks somewhat denuded and then oversharpened, relatively plastick-y, like DV and not like film source. The VC500 looks more like the original film source, which I would consider relatively more "accurate". I've had experience with 4 different copies of the VC500 using VirtualDub on different Windows systems, for just over 4 years. They all had Conexant chips. I didn't see any dot crawl problems (because I was always using s-video?). I agree, the EZCap does look rather odd, sort of nervous and "buzzy". Color is a bit gauche. Terratec looks really noisy, especially in shadows. Reminds me of cheap late-model SONY VCRs. So far, all else being equal, I'd go for the VC500. What is meant by "raw"' mp4 ? |
Interesting topic Hodgey.
Captures with an ATi All in Wonder, Canopus NX, Canopus ADVC 300 & Pinnacle Movie Box,TV card with Philips 7134 and Booktree 878 would more complete the testing. But maybe we should think about what we expect from a capture card (jitter correction, full resolution 720x576/480, full range 0-255, correctly colours, adjustable AGC,real proc amp,chroma/noise reduction and ... and don't forget the audio part and is the video/audio in sync. What would be important for capture cards? With the Blackmagic capture you have demonstrate how the Blackmagic Card capture on all analog inputs. Without a stable signal the card is useless. That's why a tbc/dvd recorder in the capture chain is necessary (for all of your tested hardware). But one thing you could see too, that the Blackmagic has some light jitter correction. |
So the plan with this thread was really more to illustrate some different TBC and TBCis devices, the raw capture card examples were mostly to show how that differs from capturing directly from the VCR. So starting the thread with them may have been a bit misleading.
As noted it's not a great comparison of the image quality of the cards, there is an older thread somewhere that has a much better comparison of image quality with test cards etc. Though, I don't think it has any of the newer Analog Devices based cards like the blackmagic or magewell. I don't have any ATI capture cards (or any of the other ones bogilein mentions other than a BT878, conexant version) at the moment, so I don't feel I can make a good comparison of that as of now. Also one would want to use S-Video rather than composite for that. Faithful image reproduction is oviously the main thing to worry about when capturing with a TBC or similar in the chain that gives out a stable signal, as stabilizing the signal is the job of the TBC. The samples were all captured to lossless, either huffyuv or utvideo, and subsequently encoded to hevc with ffmpeg. "Raw" was simply how I named the samples that were captured directly from the VCR to distinguish the samples from the ones that were captured with a pass-through device. And yeah being lossily compressed is not ideal, for comparing image quality it would probably be better with some lossless test images. I wanted to have samples of some length for demonstrating instability, so that was a trade-off. I can of course post shorter lossless samples or images if anyone wants. As for the Y/C stuff on the VC500, it doesn't act like that with S-Video, it's just with composite, on windows, that there is some weirdness. As shown, the comb filter does it's job when using it on linux. Maybe it's different with an older driver (was using the newest one from the website). |
3 Attachment(s)
TBCs
I believe all these would qualify as proper TBCs as per LS' definition. All captured with a Diamond VC500 using the S-Video output on the TBC. These all gave a stable output signal,
|
6 Attachment(s)
DVD-Recorders
These all gave a stable output with no dropped/inserted frames. All except the philips seem to output a remade macrovision signal when playing a macrovision protected tape, which some cards don't like. The VC500 doesn't mind it though. All except JVC and Sony 750 captured with VC500, the two others with the blackmagic intensity. Noise reduction was turned off when possible. I have not noticed any difference between the different inputs and outputs as with the US ES10, but I haven't tested exhaustively.
|
Can you upload the best capture you can do with the tape and a capture with an Panasonic DMR to compare and see the differnece. I would like to see what the Panasonic DMR do with the snow (clipping?) in the video.
|
1 Attachment(s)
Definitely long thread :eek:, replying as I read...
Thank you for attaching all clips to the forum. :congrats: For testing, the quality of the capture doesn't necessarily have to matter. I have some pretty lousy tapes for test uses, and the contact doesn't matter. Errors do, and seeing how the hardware responds to it. However, I do try to avoid retail sources, anything that isn't format native. Those can add unforeseen variables. Blackmagic, known issues, no surprise. Drivers can affect performance, yes. I've seen this with ATI for many years. There are a lot of reported weirdness with VC500, and I've seen some unsatisfactory color levels myself. This may explain it. I've seen too many decent VC500 captures to dismiss the card outright. eMPIA is not a capture chip. It's merely the bridge used by most USB capture cards. So you have to look deeper to see what that Magix cards actually is. And in all likelihood, especially given the craptastic performance, just another Easycap clone. EasyCap aka EZcap really screws with the image quality in every way, from luma to chroma to gamma to IRE. Comparing a solid ATI capture to an EZcap capture is like sometimes comparing VHS to HD. The issues are many, but very obviously overall damaged. @sanlyn: I don't have an issue with quality MP4 encoding. Not every sample needs to be lossless. @Bogelein: I actually have ATI AIW, USB 600, Canopus ADVC, Pinnacle, and many more. But I lack time. It's good to see hodgey make some clips to illustrate known issue, rather than us just talk about it. I am sitting on lots of research, and hope to release it in the coming year, on a new section of this site currently in development. Some are actually detailed sample clips comparing hardware. Yes, accuracy/transparency to the best version of the video signal is the goal. Removing all errors, yet don't add more. It's not an easy thing to attain when using cheap VCRs, cheap capture cards, bad TBCs, and other not-ideal devices. Everybody here needs to remember that PAL vs. NTSC makes differences in hardware and captures. External frame sync TBCs are really not made to reduce jitter. However, in the case of DataVideo's classic TBCs, it can merely be a byproduct at times, on some tape sources. The inverse situations is actually more true, meaning that the DataVideo prevents more jitters/vibrations from happening. That one is usually far easier to see with the TBC removed. I've never liked Sony TBCs. Black AVT-8710 is no more, new stock seems exhausted everywhere. It's a shame those are in use out there, the owners not aware of the issues, or simply accept it as "that must be how it works". LSI only made the record/playback chipset. You have find more specs on the DMN-86xx online. I forget which one is used by the JVC, but I know the DR-M10/30 and DR-M100/300 are different (with 100/300 being newer). I've not read the specs sheets in many years, but I've attached on to this thread. I know I have more docs archived. JVC made some of the other DNR, "Super" something. Philips recorders were always terrible, even models using the LSI. Pioneer made both really good and really bad recorders. I believe it was bad 1st, then good, then more bad last, then lines ended. Sony was never great, and had several chipsets over the years, some of which (I believe) were indeed shared with Pioneer models as well. Again, it's been 10+ years since I heavily researched DVD recorders, made much of it public (at VH and this site), going from memory. Toshiba also had not-great, then nice, then more not-great (as the last batch was just rebadged Funai). The 1st generation had IRE issues, while the 2nd had IRE fixed with step levels options. I had actually tried to buy a Toshiba at least 5 times over the years, but never could get one. Most of my research came from agreed-upon sources and workflows, with many samples discs, from VH member gshelley (a retired broadcast engineer). We collaborated quite a bit in the mid-2000s, on various video hardware that may have otherwise been forgotten (Vidicraft, SignVideo, Elite, etc). @Bogilein, most Panasonic have pretty rotten recording quality, lots of mosquito noise, but a few had that useful passthrough. Had it not been for passthrough, nobody would have ever talked about Panasonics. Those would have been doomed to the ash heap like non-LSI LG, Sony, Cyberhome, and many more. Unwanted, value-less. |
Thanks for the reply.
I'm hoping the examples at least show the advantage of using a TBC or TBCis device are, and I guess also why most of the ezcap cards are not all that great. Maybe we can get some more detailed comparisons of the gear that is the most usable later. It may be especially useful now that the good Datavideo and Cypress-based TBCs are getting rarer and rarer. I use the Pioneer and the newer Sony RDR-HX750 quite a bit these days, as they do a decent job as Line-TBCs and help fix tearing issues. The Toshiba also works for this if the video levels are adjusted for, but mine has some issue with one of the audio channels, so I'm not using it at the moment. They do as noted have their limitations so I guess one could see them as a maybe not quite as strong alternative to the ES10/15. These Pioneer/Sony dvrs are are way easier to find for sale used in my country than panasonics and toshibas, don't know if it's the same elsewhere. The other thing I find useful with them as an European is that they are NTSC/PAL switchable and can handle PAL-60/NTSC-4.43, which gives me a lot more playback options for NTSC tapes without having to have a lot of native NTSC VCRs around. The JVC works as an A/D converter and okayish as a frame sync, though it's not very usable as Line-TBC. The older Sony DVR likes to gray screen when the signal is too bad, and I also noticed that it seems to process in 4:2:0 chroma internally based on a test capture, so it's not all that usable. The Philips has the blurred image/lack of high frequencies, haven't checked whether it's noticeable on a VHS source, but it is on S-VHS or Hi8 at least, so it's not usable for that. It might otherwise have had some use as a frame sync. Quote:
On the magix one there is only a single IC, it seems they combined everything into one, including the capture part. I guess they may have bundled one of the crappier A/D converters used on the earlier designs but I could be wrong. |
1 Attachment(s)
Quote:
The tape is rather noisy in the first place. Guessing it's rf noise either from reception or the VCR that was used to record. It does look a little cleaner on our Philips VR1100 SVHS deck (smart picture: distinct(edit), tbc/dnr on, 3R off): Attachment 9203 |
1 Attachment(s)
Thanks Hodgey for the new clip with the VR1100 (JVC 9600 clone I think). Looks not so bad but it get worse when the man jump from the wagon. The capture card was the Diamond 500?
Is it useful to use a DVD Recorder after the Philips with TBC on setting? I thought it would be better to use a real tbc like the Davideo-1000 after a vcr with tbc. It's interesting how the TV One 1-TBC and the Davideo-1000 TBC perform because I haven't one. Another feature I haven't known is that the TV One use a the philips 7114 chipset. This one you can find in various graphic cards with video in & video out (for example: Leadtek Winfast A180 TDH GF4 MX440 MyViVo). Quote:
I don't use the DVD-Recorders to record, I only use them in passthrough mode. But my experience is that the panasonics dmrs clip the white level. I have testet this with 4 of them the ES10,ES15,EH65 & EH495. If you know what you have looking for you'll notice this. I have made this comparision on a german video board. Watch the picture and look at the white font on grey backgroud "WTA-Tour". You can't read it and there is no way in postproduction to make it visible again. Attachment 9206 Another intersting point is that Hodgey use different hardware as I use and none of them are have been popular in Germany. For example I have AIW 7200&9000, TV-cards with Philips7134, Booktree 878, Canopus ADVC300, Canopus NX, Pinnacle USB 500, Viewcast Osprey 240e, Blackmagic Intensity Shuttle and a cheap USB Stick with Empia 28284 chipset and 13 different DVD-Recorders. I found it very interesting how other hardware work (I couldn' buy all) and I always hope to learn something new about video capturing. I'm looking forward for your new section. Do you have any samples of bad tapes to make a comparision from capture cards,TBC/DVD Recorders? I compare them with a tape where the picture rolling from up to down, which have bad jitter/flagging (2nd,3rd generation copy and longplay tapes), tape where the picture jumps from the right to the left, poor tape with drop outs, the tennis match from the picture for the white levels and a dvd with test pattern. Any suggestions what content I should use, as the ones I do. |
Quote:
I actually have a PAL unit, but no way to plug it in right now. The PSU doesn't state 110-240, but many do not, yet accept 110. The bigger issue is the plug shape is odd, the end going into the recorder, and I've not yet tracked down one with North American socket prongs on the other end. Worst case is it won't power on with 110. |
The VR1100 clip was captured from the HDMI out on the JVC DVD recorder. In this case it's more to get stable audio sync and avoid dropping frames. TBC on on the VCR stops any later correction of horizontal jitter. The TBC-3000 would deal with the jump a little better yeah, but on most tapes the JVC is sufficient. If the TBC on the Philips has trouble dealing with the tape I would use the sony or pioneer instead, or a different VCR. I use the VC500 with the other recorders though, it's nice as it doesn't mind macrovision.
Quote:
I also got an ES10 coming, so I'll do a comparison once that arrives provided it works. |
2 Attachment(s)
So my TBC-3000 has the chips sanded, but looking at some images of the TBC-100, and TBC-1000, could it be right that the datavideo TBCs use a Broktree BT878 / Conexant 878 A/D chip? Most images seem to show a conexant chip, but I've seen a few with the Bt logo as well, which would suggest the 878 since it was later made by conexant. I think they had some kind of programmable firmware, so it would make sense.
http://www.digitalfaq.com/forum/vide...8710-vs-3.html |
Quote:
But Cypress did that one quite a few of their TBCs. Did you mean the AVT-8710? Quote:
The TBC-1000 is the TBC-100 married to a VP-299. The TBC-3000 uses a single VP301 board, and the TBC-4000 uses 2x stacked 301s. I've never seen the guts of the 7000, but it likely has a single 301 with a separate genlock board. I can take images of all those, but not now, busy, PM me a reminder in Jan. There's not any BT/CX anywhere that I remember. Same for Cypress. |
No, I mean the TBC-3000 Don't have a pic of mine at the moment, but you can see the sanded chips on this persons TBC-3000. The one I got off you looks the same inside with the markings on those chips removed.
You can spot the conexant logo on the largest chip in the TBC-1000 in the thread I linked to, and on one of the attached images, and a Bt logo on the other attached image. Granted, it could be a D/A converter too I guess. Just curious about what hardware they use. |
Ah, that's one of the 90s 1st gen jumper boards.
I kept the 3rd gen DIP boards for myself, as the 3rd gen looks best (nitpicking only) for NTSC. As with other items, perhaps there was variance of components over time. :hmm: I'll need to open all of mine and look again. I actually have some photos already, but on my Mac, which is still disconnected for surgery/upgrades. |
So I finally got my ES10, the postal service managed to screw up and return it for some unknown reason a few hours before I went to pick it up, so it was send to a return center and back again in the middle of Christmas traffic.
Thankfully it seems to be fully functional, so I'll do a comparison capture this weekend. Based on a quick test on a different VCR it seems to act as expected: more stable than the Pioneer/Sony/Toshiba dvrs - i.e didn't freeze the image when the signal dropped out, and also seemed to have some noise reduction or posterization going on. I also noticed it does not output a signal other than the first few seconds of switching inputs if it's set to a channel there is nothing connected to, or if the VCR connected to it was off. None of the other DVRs did this, they always output frames. Don't know how this affects it when capturing, but I can see why that may be one reason to use DVK or other TBC after it in the chain. Lastly, it had a input system setting that could be set to NTSC. I didn't test it yet so I don't know whether it will swap to proper ntsc, or if it will do some PAL-60 or NTSC 4.43 stuff with that set. |
Site design, images and content © 2002-2024 The Digital FAQ, www.digitalFAQ.com
Forum Software by vBulletin · Copyright © 2024 Jelsoft Enterprises Ltd.