I went for the EH55 for now.
I see on eBay they sell for cheap the ATI 750 PCIe cards which should have 3D comb filters. Has anyone tried them?
Try StreamFab Downloader and download from Netflix, Amazon, Youtube! Or Try DVDFab and copy Blu-rays! or rip iTunes movies!
+ Reply to Thread
Results 31 to 60 of 71
Thread
-
-
-
The ATI 750 PCIe cards are reported to have problems with the automatic gain control which can cause a "flashing" effect in their captures. (Some frames are blown-out (too bright) but subsequently, the brightness drops to a normal level after a few frames or seconds.)
Ignore list: hello_hello, tried, TechLord, Snoopy329 -
food for thought
AGC works a lot like Horiz and Vert sync recovery (or decode) from a field or frame
if analog (like a CRT TV or Monitor) it samples the black in the vertical blanking interval at the start of the VBI and the white in the same area to set the 'extremes' for that field or frame
copy protection work by inserting rapidly changing VBI information across these lines of information, an analog AGC would just "muddy" the extremes but still be viewable, a digital AGC would rapidly change from one field or frame to the other making noticable flicker.. with the goal to make it unwatchable
most TBCs will simply 'wipe' or reset these to eliminate the artificial varying VBI information.. and in the process eliminate the 'effect' and the 'macrovision' signal many digital capture chips use to decide whether to disable capture
so passing through a TBC that does 'blanking' will effectively (turn 'off') AGC control for any capture chips that are trying to use it.. or re-format the video signal such that its no longer a problem
again.. food for thought -
Right now I asked streamverse if they have the the Yuan SC350N1-L 3D which has 3D comb filtering (should be using the TW9919 chip).
I noticed the PD652 USB dongle they have (the one cloned by I-o Data) has a discontinued 3D noise reduction version, but I-o Data keeps selling them as "GV USB2 HQ". I wonder how that would affect Laserdisc captures.
When I'll receive the EH55 I'll post results here. -
EETimes 2007 has a nice article on ADV7180, ADV7184 and ADV7188 and line-locked time base correction (tbc).
Processing nonstandard TV signals in a digital world
Its a bit detailed, but in a Pop-Sci kind of way, so its accessible.
The gist of it is that those are 'short-term' horizontal jitter and vertical roll correction devices, they do their job, and can eliminate 'Top Curl' but because they are not partnered with a long term buffer they aren't a universal solution.
For their day they were excellent, but much better solutions came along later.
If you use one of these its best to partner it with a frame synchronization device that arbitrates with a substitution method, or duplication method to 'make up' for instances where no frame is recoverable to maintain video signal continuity with audio signal samples. Gen-locking is in general "not enough" since it actually has to use the full frame buffer to deliberately "do something" which people sit on various sides of the fence in hating or agreeing to live with. For some.. frame drops and 'aborting' the capture is the better path.. for others 'continuing' with inserted duplicate frames called 'freeze frames' is preferred.. and in some devices like the Matrox MXO2 synthesizing 'hatch frames' by blending most recent frame with a next newest frame is the better option. Some devices will also offer 'black or blue screen frame inserts'. One choice doesn't seem to satisfy everyone.
The ADV7180 was very popular in a few of the Pinnacle Dazzle DVC devices.
The ADV7180 was also popular in the BlackMagic Intensity Pro and Intensity Pro 4K PCI express cards.. so they are 'miniTBC' or 'ADLLT' based line-locked timebase 'short-term' correctors.. but have no facility for handling actual corrupt or missing fields or frames.
Later versions of the ADV7403, and ADV7842 'optionally' offered "full frame" sync buffers for various reasons.. it was an evolving buffet of ways of dealing with degrading source material.. but cost a lot more, memory chips in those days were far from inexpensive. To control costs they often left them out.. as with Blackmagic. Those that did not, like Ensemble Designs or Snell & Wilcox had to sell their devices at over $1000 usd MSRP.. much more than $1000 and they have held their value on eBay to this day.. and are scarce.
I've been having subtle conversations with people in private messages about the choices available today.. but there isn't a lot of good news to share.
Grass Valley made one of the last full frame devices.. with a custom IVT chip to try and massage the artifacts from freeze frames and hatch frames.. but it went off the market in 2017 and it was over $2000 and had to be imported from Japan.
I don't know of any good "game capture" devices that have full tbc and frame buffers.. they seem to stop at line-locked tbc with no buffers.
HDMI seems to have swamped efforts to support NTSC / SDTV capture.. the 7842 has the features.. but no one seems to have built it out.. except for the eval card.. and there is a $10,000 tax in the form of the HDMI adopters requirement to be allowed to purchase one.
anyhoo.. for what its worth
someone is still thinking about this problem
just no good answers right nowLast edited by jwillis84; 26th Mar 2022 at 18:09.
-
also, for what its worth
The Panasonic LSI chips that did "all in one"
MN673744, MN673744HL
MN673747, MN673747HL
Found in much later DVD recorders.. not the early DMR-ES series.. and I don't think the EH50 series.. those for pass-thru might be the closest thing to "accessible" today.. when they are found
I know in Japan, a series called the Power Stabilizers (funky Japanese translation in to 'Anglish'.. by Google translate) dealt specifically with this problem.. but they are few and far between.. and never sold on the open market in North America.
The early versions didn't handle wide-screen signaling and anamorphically "squished" wide screen material.. but by the second generation they understood WSS and by the third generation they even had full digital proc-amp controls in the same device with CVBS, S-Video and Component In/Out.
I opened one of these to have a look.. and its pretty much just "LSI chip" and "SDRAM" and S-Video Input and S-Video Output connector.. a breath taking beautiful lesson in simplicity.
If they still made the chips.. they could have sold a boat load of these devices in the US from 2009 thru 2019
Unfortunately Panasonic closed their chip fab in 2017 due to competition.. so its no more.. and these all in one chips were discontinued long before that.Last edited by jwillis84; 26th Mar 2022 at 18:03.
-
Techwell.. which looks really good in my opinion.. and works with just about everything software wise.. if you can get past the Japanese instructions for installing the device drivers.
Made two chips;
TW9903
TW9910
The 9903 has a proprietary 2H Comb filter
The 9910 has a proprietary 4H Comb filter
The 2H versus 4H appears to refer to the size of the Line Buffers used to hold previous and forward field lines for comparison and color separation.
For obvious reasons the 4H (Techwell 9910) version is preferred.
In my current experience.. only the I-O Data GV-USB2 .. imported from Japan has the 9910
So you kind of need to be careful of the Techwell 9903 since it was more commonly used by several companies.
A Techwell is not a Techwell is not a Techwell.. unless you pay attention to which one your getting.
I will avoid pointing out specific vendors or models.. but some seemed to capitalize on "It has a Techwell chip" late in the game.
The GV USB2 is usually sold for around $70 usd, knock offs that look like it tend to go for $50 or less.. so be careful.. if you explore that route.
----
The only "Larger, Bigger, Better" Comb Filter I've seen was in the Late ADV7441 which had a 5 Line, Superadaptive 2D Comb Filter.
I am not sure the terminology between 2D and 3D is being used in the same way by different manufacturers.
In the early days 2D seemed to refer to field by field comparisons.
Later the 3D boasting seemed to imply some motion compensation was being used to prevent artifacts from smearing color in high velocity motion scenes.
And finally the 3D monicker seemed to be retired.. or re-used to refer to the 120 Hz .. no glasses required thing with early LCD TVs.. so they went back to just calling it 2D Comb Filtering.. but with more lines for estimation.
In Japan a "D" refers to their version of early HDTV, or Component frames.. I think basically progressive 720p (or 480i) over Component.
So going back to using 2D might be a reference to using both fields of a Single 'D' frame for Comb filtering.. only with more 'horizontal lines' held in buffer space for filtering.
.. at least I've talked myself into believing this was what came of all this..
I do know for certain.. early 2D Comb Filtering .. is as bad as Analog Comb Filtering.. and later 2D Comb filtering is strictly Digital and "much, much" betterLast edited by jwillis84; 26th Mar 2022 at 17:54.
-
-
Carefull the TBS versus the TBC models.. they were different.
I'm no expert on Snell & Wilcox.. I only have one model, the slim CVR with tbc and frame sync.. but many people run afoul of the full size (width? rack mount) TBS100 thinking it does the same things as the TBC100 .. they are very different. I just avoid that line.
I "think" the CVR line were in the 1/3 width desktop units that could be adapted to the rack mounts with large elven "ears" that reach from here to eternity.
And "i think" the "R" meant "Aspect Ratio Correction" feature.. decoding all of the first three letters.. gave me a massive headache.. i've given up for now.
For sure when the BBC or the employees who left to form Snell & Wilcox they were making the large rack mount equipment with 'feature" cards.. and kept rev'ving the first three letters to indicate how the boxes were initially delivered with features.
After the re-org into a smaller company / acquisition as just "Snell" they started putting out the CVR line.. which had a master board.. and mostly ADV chips.
By the time Grass Valley acquired them.. GV ended the Standard Def stuff and went full bore on HDMI stuff and never looked back.
.. now that I think about it.. when GV acquired them.. was about the time GV started Importing their G-series from Japan
So I guess those filled that niche.. nose bleed prices.. but I guess they got the job done.. admittedly probably better than the Snell models.. but that's just my opinion
I kind of like the Doctor Who-"ish" little Blue Boxes
ha !! .. G-series "nose bleed" and a "face pounding" man those G-Series were expensive in old B&H Photo ads
the GA-series that followed (Born to the Purple) were even "more" expensive.. I saw a Youtube Interview in German (with english subtitles) of a German interviewer of a Japanese man from GV asking what the difference in the 'Purple' were.. he said.. nothing.. just the custom curved heatsink and purple color... golly gee.. $1000 more for cosmetics.
... you know.. I can 'almost' understand why Blackmagic avoided putting a frame buffer in their products.. the controversy of what to do with "Lost Frames" was so rampant.. it was like trying to make Pizza for everyone.. there was no way to win.. once a Frame can't be reconstructed.. you had to make a design decision to either abort.. or put something else in its place.. there was no way to win.. they were going to freakout someone.
Unless they made a far more comprehensive.. and complex product that let you chose what to do.. at the $2000.. like AJA FS1, FS2 kind of does.. for a price.
Putting all those complex decisions into a simple $500 product from Blackmagic.. would be next to impossibleLast edited by jwillis84; 26th Mar 2022 at 19:48.
-
I just noticed a coencidence.
The Hauppauge Colossus Version 1 PVR capture card, used an ADV7441, which is the same video decoder used by the last ADVC-G1A only discontinued in 2017 put out by Grass Valley.
It had the most advanced 2D, 5 Line Comb filter available, and the ADLLT miniTBC.
As a PCI express card capturing CVBS or S-Video there would only be "one" A to D conversion step instead of "two" A to D conversion steps when using a stand alone time base corrector.
A stand alone tbc effectively has to perform one A to D on its Inputs, then D to A on its Outputs.. and the a second A to D when entering the capture device.
Using a Capture card, avoids this second step.
What the capture software does with a lost signal, field or frame.. is pretty much entirely up to the capture card software.. so potentially it could be "all options" it depends upon the software.. where as a stand alone frame sychronizer either has to make those decisions by design or default
Its only speculation on my part but the Colossus I looks like a pretty good candidate for "best" Comb filter available.. at a low cost.. for Laser disc capture
NOTE: there is a well published "optional" registry setting for activating the miniTBC feature.. which for some reason was not included in the original installation
A Laser disc is not a VCR, so it won't have the same problems as a VCR, and frame sychronizer / buffer problems should not be an issue.. even miniTBC features should be redundant.. so ultimately the most important thing should be its Comb filter -
That Kudos Plus HD TBS800, looks very interesting ADV7802
But I'm not sure if the 3D Comb filter is right or wrong for Laserdisc capture
I'm really inexperienced with Comb filters.. but it looks like 3D is only applicable to HD or RGB signals and only for Progressive frames?
If so.. that really goes a long ways to explaining to me why the 7802 also still has the alternative 2D Comb filter for Interlaced signals.
3D looks for Temporal estimation where no spatial information is available.. its already been "Crushed".. so its the best you can do with what you have left.
2D looks for Spatial estimation where there is nothing wrong with the Temporal information.. you still have the two fields.. so its better to take advantage of the superior signal information
What I don't know.. is how the Laserdisc stores the information.. does it store it interlaced as an RF signal.. then collapse or "Upscale" it on the way out of the player to "fit" in RGB outputs?
Or, does it store the signal as a progressive set of frames.. and there is nothing left of the original broadcast signal.. it was Upscaled before stuffing it on the disc?
I could even see a choice where they stored it as a series of Progressive film scanned frames..
My simpleton mind would say.. don't over think it.. its stored as an interlaced format.. and its best to take whatever interlaced output is available.. and then put that through a 2D Comb filter to get the best picture quality possible
I just don't know enough about Laserdisc.. I think there were two standards.. one capacitive and one based on pits and divits more like a CDROM or DVD.. but I'm pretty sure the signals were not digital.. but closer to the raw RF signal.. very different from VHS hetrodyned RF decoded to IF and stored on tape.
I know a lot of people jump to the conclusion that Upscale is always "better".. but you can't really tell.. without testing, the least processed by the older electronics in the player is usually best.. since later developments in filters and newer fresher components will introduce less noise when reconstructing the signal. -
Laserdisc is optical, CED is capacitive read mechanically by a stylus like a vinyl record.
LD contains composite signal recorded field by field using PWM (pulse width modulation) analog signal, with digital sound on some titles, You can actually see the fields on the disc with the naked eye.
For materials scanned from a film strip the frames are essentially progressive with a pulldown scheme, so it won't matter much using 2D or 3D comb filters, Honestly I don't know how much softness a 3D comb filter creates, I would like to see some samples, But as far as I know 3D comb filters are much better in overal quality compared to 2D ones. -
Worth reading on Laserdisc and Comb filtering:
http://notonbluray.com/blog/comb-tb-tests/
And a good introduction about Comb Filter principles and terminology here:
http://www.cockam.com/vidcomb.htmLast edited by Sharc; 27th Mar 2022 at 07:38. Reason: 2nd link added
-
Re that link - the successors of the NEC 3D Y/C ICs that are noted there as being in the HLD-X0 and CLD-99 (but not the X9 which seems to have a panasonic chip?) is what is found in pioneer and later Sony dvd-recorders that feature 3D Y/C, and presumably also in TVs, AV Recievers etc based around the same chipsets.
3D comb filtering is used for interlaced material too - analog composite SD video doesn't really have a concept of "progressive frames", every field is stored sequentially in the signal. Progressive scan video is going to still stored interlaced by taking every other line of a progressive frame and creating two fields. For PAL it's simply transmitted like that (2:2 pulldown) while for NTSC some fields would be repeated to make 24 fps film somewhat fit into 59.94 fields per second (3:2 pulldown).
(There is also the "240p/288p" progressive scan variant often used by game consoles and on-screen display stuff which is simply modifying the signal slightly so each field is put in the same place on the monitor but that's more of a hack.)
Laserdisc (and some broadcast standards I think) could include metadata in the vertical blanking interval to "inform" that the material was originally progressive, whether that's taken advantage of by the Y/C filters I don't know though.Last edited by oln; 27th Mar 2022 at 07:29.
-
Also take a look here, e.g. section Comb Filter Terminology:
http://www.cockam.com/vidcomb.htmLast edited by Sharc; 27th Mar 2022 at 09:31.
-
The S&W pitza box size devices did not do well for VHS if used with crappy VCR's that lack line TBC but the half rack ones that have line TBC such as the TBS800 and CRV600 did an excelent job, Also keep in mind that those are analog to digital converters via SDI they are not meant to be used like a consumer TBC with an analog output, There is no need to go back to analog with those devices.
-
also a nice article on the ADV7802 versus the ADV7180 and 2D / 3D Comb Filtering
Optimizing standard-definition video on high-definition displays
a bit over the top technical detail wise, but the summary is short
"Professional-quality video decoders, such as the ADV7802, deliver better than 62-dB SNR (signal-to-noise ratio) using a 12-bit ADC. It is important to note that the differential-phase and -gain figures for performance-driven applications can exceed 0.45° and 0.45%, respectively. Cost-sensitive applications may use a video decoder with nine-bit ADCs, such as the Analog Devices ADV7180."
3D adds processing time and you need to store and "delay" the audio to prevent lip sync issues, further driving its cost up, but as long as your funding "large memory buffers" to support 3D Combing, when your not using all that memory (or can't because the scene has too much motion) you might as well use all that buffer space for 3D Noise Reduction.. but they are mutually exclusive.. you can do one or the other but not both.. unless you splurge for even more memory costs to separate the two functions into different circuits
what I got out of it.. was 3D Combing.. is (or was) so expensive.. it was far outside the range of normal consumer video capture devices.. and only became sort of accessible to consumers when Tuner based MPEG hardware chips started including it on a limited basis
the "Technology" of 3D Combing also seemed susceptible to "complete failure" at detecting motion that would cause artifacts and required human supervision to oversee and selectively "switch it off" when it wasn't working well with a particular program
offering 2D Combing was less "effort intensive".. "adaptive 2D Combing" could more easily detect motion between two or four consecutive lines in a single field and switch itself off and go back to notch filtering to raise the signal to noise ratio of Luma and Chroma
really eye-opening.. i didn't understand much of the math until reading several articles
its far from "simple" to use a 3D Comb filter.. its a lot of workLast edited by jwillis84; 27th Mar 2022 at 16:44.
-
Lol.. yeah.. I can't imagine someone using a Snell & Wilcox to go to all the trouble of a 3D Comb, only to convert it back into an Analog signal and throw it back on a noisy video cable to another box for re-capture. Once through the A to D converter is enough.. let alone the bumpy ride through yet another analog jumper cable.
S&W are true "Professional" grade capture devices.. and SDI is broadcast grade.. not a lot of consumers were buying up SDI capture cards pre-Y2K brand new.
Still a consumer grade Hauppauge Colossus I or II only takes the video signal through the "tumble dry" A to D conversion one time.. and keeps it there. If you can't obtain.. or afford.. a Professional 3D Comb SDI capture workflow.. a 5 line, 2D Adaptive to PCI express workflow is certainly nothing to sneeze at. The only thing missing then is a frame sync... which its debatable if "abort" when a missing frame is detected is "desirable" or your capture software should do something like automatically "insert" a duplicate frame.. or something else. -
yeah.. I'm pretty stupid.. I had a lot of preformed 'guesses' on how things worked until I dug into the math
2D or 3D its all "static" image processing
the 2D is comparing nearby lines in the same field (or frame) to construct a mathematical "filter" to separate Luma from Chroma information and amplify it from the noise floor at the same time, its a "better" filter than the simple "notch" filter..
but either 2D or Notch can fail.. and lead to the "dot crawl" problem.. when its working there is no "dot crawl" when its not.. it creates an artifact that no amount of filtering can remove later downstream
the 3D is comparing two or more frames stored in memory "first" to see if there is temporal motion differences between the frames.. which means objects are moving.. and indicates that this method will fail.. so it aborts and sends the whole stack back through 2D.. if 2D fails it sends it back through Notch.. so its a waterfall of "methods" for separating Color information from Brightness information
each "method" 3D > 2D > Notch can fail.. the worst at failing is 3D.. but it also has the potential for doing the most good.. with perfectly "static" and no motion scenes.. reality is.. some scenes are going to have motion in them.. and it will fail.. so it has to fall down the waterfall to see what method can be used.
i would worry that all this shifting between methods might lead to some kind of "visible" artifact "shimmer" within the moving objects.. which is apparently the problem.. some people can get motion sickness or headaches.. others not so much
3D is really hard to pull off well.. it works great with Museum walk throughs.. not so much with the 'Fast and the Furious' -
There are modern 3D comb filters that are specifically designed for composite sources like Laserdisc such as the one I mentioned in post #6 from Singmai, It remains to be seen on how it performs, I don't have composite sources to capture to make the investment worth otherwise I would have aqcuired one and posted some results here.
-
I ended up trying my old Dazzle device with the ADV7180 chip (the one with hardware encoding). It apparently has a noise removal filter by default, but the audio has a lot of background noise. It also only seems to work with Pinnacle software. If there’s any other device with that chip that outputs uncompressed video let me know.
I don’t have a Colussus but I have the original HD PVR from Hauppauge. I know both use hardware H264 encoding.
As for Techwell, I have another device that uses the TW9910 chip (the other Yuan clone) and besides the proc amp having different default settings, I noticed that it handles SECAM VHS signals way differently (and with way less noise).
I’ll wait for the EH55 first, then I’ll look into more options. I’m not gonna buy very expensive hardware just to be disappointed after. -
if i were you I would wait on buying anything else and spend time learning more about Comb filters
-
@dellsam34 I'm curious about the SM14 you mentioned in post 6. I have been using an EVAL 7842 board, but it's finicky sometimes. And the built-in comb filter on the HLD-X0 is decent at best. The SM14 is about $320 US so a little high just to test.
The SM14 outputs in SDI. I'm green at SDI. My cap card is an Osprey 827e which does accept HDMI/DVI, and analog BNC. Do you know if I could capture from the SM14 with this card? -
I don't know about your card but SD SDI is basically lossless AVI 4:2:2 8/10bit 720x480(576 PAL-SECAM) with embeded audio 16/24bit 48KHz on a sigle coaxial wire, no decoding needed, straight transfer to hard drive using a PCIe or USB 3.x adapter, Many PCIe SDI cards can be had for as little as $10 used, USB 3.x devices are little pricy, they don't work on USB 2.0, not enough data bandwidth. SingMai supposed to be coming out with a Y-C and composite box, the SM14 is composite only, I asked them about USB 3.x implementation on the device but there is no plans for such port according to them.
-
I saw some of your comments over at LDDB about the SM03. Looks like at the time of that thread they didn't have the SM14 developed yet. Glad they made something to work with composite sources (laserdiscs). If I buy one I'll post some results using the Snell & Wilcox test pattern from Video Essentials laserdisc.
I used to have a BlackMagic Intensify Shuttle which is usb 3.0. -
I saw some of your comments over at LDDB about the SM03. Looks like at the time of that thread they didn't have the SM14 developed yet. Glad they made something to work with composite sources (laserdiscs). If I buy one I'll post some results using the Snell & Wilcox test pattern from Video Essentials laserdisc.
I used to have a BlackMagic Intensify Shuttle which is usb 3.0. -
it appears
most people looking to capture Laserdisc composite
tended to buy a high-end 'Scaler' intended for preparing the signal for viewing on an RGB Projector or other large screen format
they also looked into high-end AV Amplifiers, which also had a 3D Comb filter
the commonality seemed to come down to ADV7800/01/02 or above which were the first ADV products with a 3D Comb
or the TI TVP5160 which also had a 3D Comb
or the Conexant products that had a 3D Comb
Since its easier to use S-Video from an S-VHS VCR to avoid the need for 3D Comb, and instead focus on 3D Noise Reduction (3DNR) 3D Comb was never an issue for VHS capture artists
i really think this question is better answered in some of the messages published in a different forum long ago
it appears 'many' USB capture systems like the ADVC mini (which used the Micronas AVF4x 3D Comb) were available.. it was mostly a matter of the year in which the product came out and if it used one of the chips that had 3D Comb
given the nature of how 3D Comb worked.. it might explain why the ATI 650 and later had flicker issues.. that was what worried me about how switching from one method of Composite to Y/C conversion worked.. if done crudely.. it would exactly artifact by "muddying" the water depending on the motion content from one frame or field to another.. that would produce something sort of like an AGC problem.. but wouldn't be able to be corrected by a tbc
for churches and schools, 3D Comb would be good for Upscaling static "Powerpoint" presentations.. or low motion video.. but once it turned action filled.. it would quickly become less desirable
this leaves me wondering if those old ATI 650 solutions would work much better if you only used their S-Video Inputs.. or could turn off the 3D Comb for composite or OTA
A lot of the ATI 650 Combo devices were recalled from places like Best Buy for "tuner" related issues.. but could easily have been because they were attempting to use 3D Comb on broadcast signals (which is kind of ridiculous.. if you think about it) -
Jwillis I think you are over thinking this.
Similar Threads
-
Laserdisc S-video or composite capture?
By digicube in forum CapturingReplies: 26Last Post: 12th Jan 2021, 18:29 -
Looking for a VCR with a good comb filter
By TheRealOC in forum Newbie / General discussionsReplies: 0Last Post: 16th Dec 2020, 05:30 -
Multi-format VCR player with good comb filter
By TheRealOC in forum Newbie / General discussionsReplies: 16Last Post: 8th Dec 2020, 00:03 -
How do I hook up a HX9 Muse Laserdisc player to my capture card best?
By wakalimus in forum Newbie / General discussionsReplies: 14Last Post: 13th Dec 2018, 03:17 -
What it he best analog capture card for laserdisc AND MUSE AND Hi8?
By wakalimus in forum CapturingReplies: 6Last Post: 12th Dec 2018, 05:55