SingMai SM03 review? (Sing Mai SM-03)
 

Has anybody tested the Singmai SM03 and can they speak to its quality? I'm curious how the TBC/frame sync feature worked on this device.

I initially tried to revive an older thread on this topic but was prompted to start a new thread. I know a few users had expressed their intent to buy and provide a review about 6 months ago.

I don't know much about SDI. The SM03 outputs 10bit SDI, but the Magewell PCIe/USB capture cards that bundle with the SM03 capture YUV2 4:2:2 at 8bit (I think). Do the bundled Magewell cards hobble the SM03's output? I know 10bit is overkill for VHS anyway but I'm still curious. I have an Aja IO HD (great product) and I think I could pass the SM03 SDI output into the IO HD to capture 10bit uncompressed (and eliminate the need for a Magewell bundled card).

Any info would be appreciated--I'm thinking of biting the bullet and trying it out.

https://www.singmai.com/Modules/sm03.html

From day one, I've had my reservations and doubts about this unit.

Too many people see "TBC!" and want to buy one, most of whom don't see or understand details and fine print. But I do, and what I see gives me pause, squint my eyes, tilt my head.

I'm not some all-knowing sage, but over and over again, I'm proven right. (And the few times I doubted myself, did it anyway, it was a mistake, lost time/money. I was correct all along, but wrongly gave into peer pressure anyway. That feeling sucks, being pissed at yourself.)

Lots of devices make promises, far fewer deliver. I'd bet money this is the latter.

The status quo remains fully unchanged. If you want a good TBC, get a good TBC. Not some "also has" devices that really has nothing, or at least nothing as you think it would/should/does.

Remember, the Singmai device is really just an SDI converter, and those have downsides and limitations. And if you want an SDI workflow (which isn't really suggested for consumer source like VHS, is closed-loop, etc), then there are several options already.

It's not a TBC, and based on the info in the user manual, does not contain the necessary chipset needed for true TBC whatsoever. A TBC is not a single chip on a board!

Sometimes "biting the bullet" is simply playing Russian roulette. Put down the weapon, think it through.

Thanks for the reply Smurf I appreciate it. I already have a TBC-1000 and it works great. My sense/first instinct is that a new product like the SM03 would offer at least SOME kind of advantages over my current kit that predates me in some cases. I’m hoping anybody who has tested the SM03 can speak to that. I mean, even if I keep the tbc-1000 in the chain and feed its output into the SM03–it could still convert that frame synched signal to SDI and (potentially) produce an image that is higher quality than other recommended capture gear?

Two more questions for you:
options for SDI—are there any you recommend for consumer VHS/Hi8? I’m more interested in removing the TBC-1000 from the workflow. Are there any SDI devices that meet that need?

Chipset—most interesting to me bc I have no idea here. What is the necessary chipset for TBC? And i assume the chipset in the TBC-1000 is an ASIC? The SM03 is FPGA—so wouldn’t a TBC function be programmed on it? This is what the manual says:

Nothing in the SM03 appears to be new or unique, if not outright inferior.
What do you imagine would be different?

It's just an SDI converter. Those already exist.

No. VHS has signal and quality issues that exceed the design of these professional units, which were made for somewhat more stable BetacamSP and comparable. If the tapes are excellent it may be fine. If the tape has typical issues, not so much.

Not really. The TBC-1000 (and other DataVideo, Cypress, few others) was specifically design for frame sync of crappy consumer sources. SDI gear is not really tuned to expect crappy quality input. (You must understand that "crappy quality" is not necessarily visual, but includes invisible signal and non-visible visual errors. It's not just what you see on a TV.)

An FPGA is indeed a chip ... a single chip in a set of chips, aka chipset. But I know for fact that you can't just slap a random FPGA on a board, and declare it a finished product. It's incredibly more complicated than that. Nor does any current SD-enabled off-the-shelf chip have defaults that works well (or at all) ... which seems to be the case, especially due to the option to reprogram it.

Certain features, like NR, are already on-chip, and (as far as I can tell, based on what I know about TBCs, and these chips) it's nothing Singmai created, but rather just enabled (or didn't disable). This is why I'm not giddy like some folks have been. More impressive would have been actual chroma NR (cNR), not what exists there.

That brings up another point. There's nothing "modern" to be had, when the cNR attempt just mushes together, and makes "mouse trails" (blended chroma). This is a rough on-chip option going back 25+ years, and should just be avoided for actual removal methods now in software (and for at least a decade now).

It's a nifty SDI converter, few extra features, but not at all what some are thinking it is.

Most of this stuff was built into early BV10 units 15 years ago now.

You can probably expect subpar line correction (vs. in-VCR line TBCs, or ES10/15 type), mushy NR, non-fluid proc amp, potentially weaker frame sync. Of all those, I'd expect frame to be the only coin toss, 50/50 odds of being any degree of quality, but the rest are unlikely.

If you still want to get it, fine. But lower your expectations. It's not the second coming of the wheel, or of sliced bread. It's mostly just a re-engineered (or reverse engineered) product that has existed for decades. And we already know how that went in the early 2010s, giving us "black AVT-8710" type flawed units.

The Youtube example videos were wholly unimpressive, uninspiring.

I always shake my head when I read text like "video source such as VCR" (pages 11, 14). It makes you wonder how much they truly understand video, or formats.

I could go on, but that's enough. As already stated, I see things most people would miss, and it's not good. Be wiser, more discerning, don't be easily fooled.

I have had several e-mail exchanges with Daniel the head of SingMai and he seem to know a lot about analog signal timing and VHS, While I haven't tried the SM03 personally due to its price and I already have similar devices from the late 2000's based on the SDI architecture, I gave him a suggestion about making it affordable for the masses but he never gave me a definitive answer if that will ever happen.

As I said I used similar devices and I posted samples over at VH comparing conventional capture to the S&W TBS800 and the difference is night and day, The first thing that stands out about such devices is the stability of the frame in terms of line timing, It is the most advanced line timing I've ever seen, even my JVC VCR line TBC could not perform that well, I wouldn't doubt a device made in the last two years such as the SM03 would perform more than the TBS800 given the amount of power the current processors have compared to 15-20 years ago.

I do agree with LS on the price, But if you factor in how much analog TBC's cost nowadays that should be a great deal, Even the SDI devices from 10 years ago go for over a grand.

If you do wish to proceed with a purchase please keep us posted with samples.

Don't confuse these types of chips with x86 or ARM, or even ASIC or GPU. It's just not anywhere near the same leaps in performance. More advanced than yesterday, yes. But most are about size, heat, and energy use. With something like SD video signal processing, the excess speed is pure overhead, idle overkill. Speed wise, these things are measured in low triple-digit megahertz.

Yes, but the standard TBC has resale value, while a boutique SDI card will be so overly niche that resale is unlikely to impossible. There are some other video items that I never discuss, mostly because it's hard to locate, expensive when found, and something you'll stick it your closet when done. Rare is rare, valuable is valuable, and while those often intersect, it's not always so. When it comes to video capture/conversion tools, for use by non-pros, I do think "big picture".

Yep. :congrats:

Or loan it to me for a deep test, PM me.

Daniel from SingMai just sent me an e-mail with a link to a new video capture he's done with the SM03, The VCR is probably low end and despite being B&W and heavily compressed by YouTube it looks promising, I asked him for more tests.
https://youtu.be/C7sxn0FJOwk

Interesting. Thanks for the update!

You would have thought that somebody who has built a brand new TBC and is trying to flog it would be putting in more effort to promote it, not just pop a B&W on YT. I'd be bombarding people here with proper samples. Perhaps they're afraid of the scrutiny...

True. "The VCR is a JVC HR-S9600 with the TBC turned off. ". :)

Reply to questions about the SM03
 

The SM03 is not just an SDI converter. SDI has a very precise jitter specification, so to convert any video other than near broadcast specification, to SDI, you have to retime it. So that is one function of the SM03, and it does this using a frame synchroniser. It isn't, in the true sense of the word, a TBC, as this will have velocity compensation, and we don't have access to those signals (some VCRs had a TBC output but VHS machines do not have this). But it accepts jittery video inputs and retimes them to provide a stable output (i.e. no tearing at the top of the video).

More than this, the SM03 provides broadcast quality decoding of NTSC and PAL using a 3D comb filter. Even in YC mode the comb filter operates to help remove artifacts from the image.

The SM03 provides luma and chroma noise reduction. This helps remove the tape noise from VHS tape sources and it does this without visible trails or contouring.

Lastly it provides chroma edge enhancement so improving vertical transitions, and it does this without overshoots.

All of the above design is proprietary and the design is run on a single FPGA. It does not use a chip set or any off-the-shelf decoder from ADI or Philips or whoever. All the design is unique to SingMai.

The SM03 was not originally designed for tape archiving or playback of analogue laserdiscs (that product is the now the SM10). Because of this we are working with two companies who are providing us with feedback so we can optimise the performance. Anyone who has already bought an SM03 gets a programmer so any improvements can be programmed into their SM03.

I have been working in video for over 40 years and our customers include companies such as Lockheed, NASA and General Dynamics. I was principal design engineer at Snell and Wilcox and also worked designing video decoders at Philips and LSI Logic.

If anyone wants to suggest video clips we can run through the SM03 I would be happy to do so. Also, if anyone buys the SM03 and feels it does not perform as they think it should we offer a full refund (less postage). Also, if anyone has any questions again I am happy to answer them.

Well there you have it, This should clear some ambiguity.
Keep us updated with more tests on youtube.

Thanks for the info, @SingMai. If I had the dosh I'd buy one. C'Mon Lordsmurf, go for it!

As I wrote, as SingMai confirmed, not for videotapes.

If and when the happens, SingMai is welcome to contact me, send a product sample for extensive testing and review.

You must understand my perspective here. For at least 15 years now, I've been told that various random devices were "also TBCs" or "contained TBCs", or that a product "will do __ in the next revision". The first was never accurate, and the latter never happened. If all those TBC conversations to date were a baseball game, it'd be a no hitter. Hence extreme skepticism now, unwilling to waste more time and funds on promises that don't deliver.

But again, if and when. :)

Replies to more comments.
 

The SM03 is specifically designed for the archiving of video tapes and the archiving or viewing of other analogue sources.

The SM03 comes with a 3 year guarantee. One of our products is installed in the International Space Station so they are reliable. The components we use are from premier franchised suppliers and are industrial temperature specification. Each unit is put through a 24 hour soak test before we despatch.

We self-certify our equipment to the CE mark (EU equivalent to FCC). All of our equipment is tested using industry standard test equipment from Rhode and Schwarz, Tektronix or Phabrix.

All our products are supplied with schematics, but we also offer a repair service (which to date has never been required). We know of one of products that has been operating continuously for 8 years now without any issues.

I'm afraid we can't afford to send any more free units out. Two units are with two UK companies who are giving us feedback on its performance. Both companies archive thousands of tapes a day and have a variety of tape sources (including non-VHS).

Longevity is about construction, such as how it handles heat dissipation, or even heat generation to begin with, not warranties. Also components, such as ceramic caps, not cheap electrolytic caps.

Longevity is measured in decades, not years. Far too much gear has already failed us.

It's a pity that it cannot be tested to vet the claims being made.

@SingMai
I think I see the misunderstanding here. First off, the price of the device is not really a concern compared to other similar devices of the era that were selling for over $1000 15 years ago, if you account for inflation that's almost double, So let's put price talk aside for now, Also quality is no doubt there compared to Chinese devices.

What Lordsmurf and other members are skeptical about is the performance of the device with a variety of consumer tapes including, VCR recorded tapes from TV and home videos recorded with budget camcorders, When they hear companies are doing samples it jumps to mind that those are professional archiving businesses, Those deal with mould and other tape problems, They don't deal with issues due to hardware quality or tracking errors due to a miss aligned camcorder or VCR, they don't deal with improper levels due to user error, they don't deal with Funai and RCA VCR's period.

While most of the hardcore hobbyists don't see 500 euros expensive, they just don't see the need for such a device because they have an arsenal of them, I personally at some point had 4 BE75, 2 S&W TBS800 and a bunch of other consumer devices, But the masses who are potential buyers for your device need to see experiments and samples from within forums like this, They simply don't trust companies doing testing for them, That's not to say that if you provide enough convincing samples online highlighting the issues that the average Joe faces with his mom and dad VCR, I'm pretty sure there is a number of them want such a device knowing that consumer TBC's are like $2000 now if you can find one.

Another issue, and I've already discussed it with you in previous e-mails is that people don't know what to do with an SDI port, If you tell them you need another device to connect to the computer they will see another expense and often they get confused about which one is the capture device the one next to the VCR or the one next to the computer, If you can find a way to put a USB or thunderbolt port or a port for an external storage device like a SSD that can save the uncompressed and uncropped 10/8bit 4:2:2 AVI that would be the device of the century.

I did ask you before for a free sample to test and was hoping that I can pass it along to LS, Jwillis and other senior members here and at videohelp before it finds its way back to you or one of them buys it, but for now I'm very busy with some family paperwork, After I'm done I will order one and do an extensive testing with consumer made medias such as VHS, V8, Beta, PCM VHS and PCB Beta.

Meanwhile feel free to send me over the cloud short lossless samples of tests to put them on my YouTube channel, I will do the proper de-interlacing and encoding so it wouldn't lose too much quality after YouTube processing.

I have added another video clip for the SM03: https://youtu.be/YNzOLW0kydI

Thank you for the feedback. I will add some information about SDI and capturing the video to the user manual - I will post a link when I have done that. I will also post more clips.

Thanks for that; that You Tube bitrate is very very (too) low. The display Ratio is also 5:4; I think should be 4:3. I'd suggest uploading to YT at at least a 1440x1080 at 4:3.

Could you post up a Lagarith or HUFFYUV lossless file (max 99mb) on the forum.

I noticed that there is more blanking space around the frame compared to other capture devices, What native resolution the ADC samples at and what is the output resolution if different than the sampling resolution? Ideally the device should not be doing any resizing at all.
In future tests try some LP home recorded tapes, SP commercial tapes are more robust than home made ones.

Thank you for the suggestion - I'll do the test this weekend.

Or just an S-Video connector for output so I can easily integrate it into my existing capture chain.

Also, there seems to be some overlap in capability between the SM-03, SM-10, and SM-14, so it would be good to have a chart that shows the differences.

No, having S-Video out defeats the purpose of the devise's existence entirely, The SM03 is a capture device that converts analog to digital, If you go back to analog it incurs an unnecessary lossy step and requires another capture device instead of just a SDI to USB interface, Not to mention that they have to fit in another DAC which increases the cost.

Answers to further questions.
 

Hello,

In response to comments above:

The SM03 webpage now has a table showing the differences between the SM03, SM10 and SM14: https://www.singmai.com/Modules/sm03.html

I have added more information in Chapter 2 of the user manual on SDI and how to display and capture the video: https://www.singmai.com/Documents/SM...r%20Manual.pdf

I am adding NTSC443 and PAL60 standards to those supported by the SM03.

And just a comment, the Magewell modules can record uncompressed video if required.

I welcome any other comments or questions.

Hold on, are you saying that other external TBCs are pure analog devices from end to end, they don't have a DAC in them?

Other than very ancient ones for early open reel tape formats no. If there is analog output there will be a DAC of some sort that turns the digital signal back to analog.

Haven't said anything about conventional TBC's, I was talking about the SM03 being a capture device with integrated TBC functionality after the ADC, But just to answer your question, yes conventional TBC's do have both ADC and DAC for video only. The SM03 does not need a DAC, it's function is to digitize both video ad audio, do the timing and other digital processing and send the digital signal to computer in its lossless form, My point was why do you want to convert the signal back to analog, for what purpose?

I'm being picky, but that's not strictly correct. It only outputs SDI. You need another gadget to "convert" the SDI signal into computerspeak ie USB ingest.

Singmai provides that converter gadget (PCIe or USB) with the SM03 (for an additional cost), noting of course your previous comment that it would be great to have that integrated into the SM-03 box.

Detailers, proc amps, etc. And not everybody wants to be locked in with married devices. Not all capture cards play nice with all tapes, even with TBCs in the workflow. Same for all TBCs, but you want a certain capture card. As mentioned many times, SDI is closed loop. If you want SDI, fine, but the lack of DAC is not a very strong argument. The issues with DAC is really more about the quality (or lack thereof), not mere existence.

Nice to see NTSC4.43 and PAL60 but how about native SECAM and MESECAM? Oh by the way the community here seem to dislike any advice to use OBS, Don't know what has changed over the years but it use to have problems capturing SD rec.601 based materials.

It's not just that, either.

OBS connects to capture cards at the screen recording layer, sandwiches between actual device connection, and overlay/display. Computer display has multiple layers. At the non-device layer, it's subjected to errors/corruption and alterations. Some NLEs used to operate this was as well, contributing to why NLEs all sucked at capture (and still do, for other reasons, mostly still resources overhead).

So OBS isn't capturing with the hardware, it's screen recording, but at a less-molested way that many understand screen recording. In this way, OBS is a more advanced screen recorder than most.

I can't explain all the extreme technicals too well here, but it's essentially DirectX, which you'll note is a requirement of OBS.

OBS is crap for analog capture. It's fine for everything else. Not unusual, as it's an "also has" feature, and those are never good with anything in the capturing world, hardware or software. A poor "feature".

This is the conversation with Lord Smurf from the Singmai Scammer post. However, since this has nothing to do with scam, I'm copying the posts here with my response.




Yes, I was referring to the user Nidi with the two devices.

The blue SM03 device appears to be without a Y/C input. It would then have to be a device from 2021 to mid-2022. The SM03 V1 devices with Y/C were available from mid-2022 to early 2023. The Singmai V03 V2 devices were available from early April 2024 to late April 2024.

According to the Singmai Facebook page from May 6, 2024, the first devices were delivered to companies.

After the Media Production Show in May, Dan was apparently in such poor health that he could no longer deliver or assemble devices. There may be a few more V1 devices than V2 versions. Privately owned by users involved in video capture, I think you can count the devices on two hands.
I have packed all the pictures of the two devices and user manuals I have into a zip file and attached it here.

That would be me, just discovered this forum and saw my pseudonym.

Sad to hear Daniel has pased. I had e-mails back and forth with him for years.
captured some scenes and commented on his work.

I was mainly interested in good 3D Y/C comb filters to capture some clips from my LD collection.

tried several 3D Y/C models to see if it would better the already very good NEC 3D Y/C comb filter in Pioneer's Flagship MUSE HLD-X0 player.

3D Y/C comb filter tests
 

here's the link to my comb filter tests that I made a while ago

https://forum.lddb.com/viewtopic.php?f=50&t=11095

pictures ands videos of menu system
 

link to pictures and videos of the 2 units

https://forum.lddb.com/viewtopic.php?f=50&t=9602&start=100

Ah yes, the old "what is the best YC comb filter" question - IE, what separates composite TO S-Video the best with the fewest artifacts (typically rainbowing, dot crawl, and/or sharpness loss).

Most people I've talked to say that the Harris/Leitch DPS-X75 with A3D composite input card does the best overall, but I don't think I've seen an actual comparison to the SM03 - I did check out the "MEGA.nz link" posted on the laserdisc database and I don't think I saw anything captured with the SM03 unfortunately.

I'm not totally sold on the X75 being the best having not done the testing yet, but I do have one for testing. I'd be using a TPG21 with the Snell and Wilcox SW2 pattern on it as the source which is an uncompressed version of what's on the different video essentials products. I do think just using the video essentials DVD with the SW2 pattern is probably good enough to show the sort of problems we would be looking for.

There also is some question as to how well these different TBCs handle line errors and for the most part the first digital device in a chain will "bake in" any errors, so using something like an X75 *AFTER* say a passthrough DVD recorder, I don't think you'd get any benefit over just using the S-Video output of the DVD recorder.

The X75 isn't particularly super rare and I think that is why several people have tried it, but there are lots of other interesting devices such as more modern For.A (thinking FA-9000/9100/9600), Snell and Wilcox (CVR and TBS series). I've even seen some Snell and Wilcox composite video decoders that actually have laserdisc specific modes in the menus - those are typically in the MDD series, though they do make a variety of modular composite decoder cards also.

I may put up a separate post with just the SW2 pattern from the video essentials DVD as a burnable ISO and ask people to post their results of burning the disc and then playing it via composite into their chain to show the results that can be publicly posted and compared. Variances in DVD or bluray players shouldn't matter for the sort of errors we are looking for as far as the performance of comb filters. If you wanted to get fancier, you'd then record that pattern from DVD (via S-Video) to VHS, then try to capture that tape to see how the same chains handle timebase errors via composite output from the VCR playback also. While I also haven't experimented with patterns recorded to VHS, you might be able to purposely induce more timebase errors by playing the tape back on a different VHS player than the one that it was recorded on (ideally different brand altogether). The key here is making the recording with an SVHS recorder with S-Video in. The playback part can be with a basic composite output only VCR or at least with any internal TBC features turned off.

Yes, there are no Singmai captures, as I don't have a way to capture via a PC.
I can however use a camera and record off my screen.

created a new directory called Singmai, there's a recording off my screen

You can use a SDI recorder. Do you have the latest version of the SM device after the chip shortage or before?

version
 

I do have the old blue one and the new one with the LCD display.