Gordon is a true gentleman and of course a technical genius. I have had several interactions with him and he was very kind and helpful going out of his way to help me. Thank you, Gordon.
I have been thinking a bit about this video, and I now understand why the ACE isn't ideal for digitizing video tapes. Essentially it was created for a completely different use case, and that was frame synchronisation. Back in the old days if you wanted to have wipes or fades you needed a video mixer. Professional ones are (conceptually) rather simple devices as all your sources are completely in sync. You could essentially have an "audio-mixer" capable of 5MHz to fade between your sources. Now with video cameras there were some better prosumer cameras which were able to genlock into an existing signal, but if you dealt with consumer VTRs getting those to sync was virtually impossible. So what you needed was a frame synchronizer to sync whatever signal you got to the rest of your studio. Some consumer video mixers included those, but that was a rather expensive component in them. Since the frame store was expensive, they often used it to also do "framestore effects" like still or mosaic. And that is the main usecase for those ACEs, that's why they do things like trying to swap fields. The whole idea is to change video that is off sync to its reference sync. So if you have a VTR running to fast or to slow, it'll drop or add fields so it gets constant 50fields per second. If I'd design such a device with the possibilities of today, I'd go a different route. I would first try to find the horizontal and vertical sync pulses as well as the drop-out/discontinuity caused by the head switching. I'd then smooth over the horizontal sync pulses a bit so that noise on them is less important, if an impulse is missing I'd add it. Then I'd determine in which line the vertical sync pulse is and number the horizontal sync pulses based on this. If the vertical sync-pulse is lost by a drop-out I'd interpolate it from previous and next frames/fields. If that fails I'd fall back to declaring the middle of the space between the head switching to be the middle of the frame/field. Once I know where each line starts and ends, it's trivial to "unstretch" it and generate an image out of it. I'd detect the sync pulses like this: Horizontal sync: Cross correlation with a reference pulse. (essentially averaging over it's length) Vertical sync: Cross correlation with a reference pulse. (probably via actual cross correlation as the pulse is rather complex, alternatively averaging over several lines) Head switching: Sudden jump in the period of the horizontal sync pulses.
An extremely interesting conversation! I’ve been using my ACE for many years and even though I’ve been retired since 2000, I’ve hung on to it for when a friend or relative asks me to do a conversion from an ancient tape! I’ve now reached the point where I just have to clear out some equipment and I’ve just listed my ACE on eBay, so it will be interesting to see just how much it goes for now!
WOW, pity I am so late to this video. I have a Mark 2 which I purchased new in Australia via an import. I will NEVER sell mine, it is simply amazing. Nice to know I could get firmware upgrades of needed and repairs. Great video and what a lovely clever man, well done mate !
Great interview and what a nice chap. Reminds me of Tony Murray (Murraypro) who makes the ppm monitoring boxes and test equipment (and probably a similar age). I guess I’m this era of raspberry pi’s etc, there are more opportunities for one man band operations, although there is less of market, though having said that, the archive market has created a new demand.
Just a brief interest, in audio and video. Butj ust to say thanks for going through all the effort to get it on video, times move fast you've got it on video for many years to come. Interesting to watch thanks
What a loverly man, he didn't seem to mind all the questions and i'm sure you learned a lot :-D At 70+ i would not expect him to have much interest in something he created a long time go. But he seemed to enjoy the conversation :-D Enjoy and appreciate the magic he created :-D A very clever man indeed :-D
A delightful and fascinating conversation! I'm curious how this purpose could be achieved with technology we have today. What would a 2022 prosumer TBC look like?
Well you can buy such a thing. It's essentially an "all in one" TV-IC connected to a LVDS to HDMI converter IC. Those are available as black metal enclosed SCART+HDMI to HDMI converters.
@@alphabeets Those things come from China under a variety of names, but search for SCART->HDMI converters. They should all be based on that basic idea. The disadvantage is of course that they scale the image.
What a fascinating interview, and what a nice chap! I enjoyed every second of that. He left BT Labs two years before I started my career there as a Graduate :-)
If a new design to be sought it would be by today's standards, Only digital out needed (USB4 preferred), no need to physical knobs and buttons, 3D comb filtering for composite sources, better DNR for chroma and luma, Full field and line correction at 12 bit, All color formats possible, 24 bit audio.
Well you are probably looking for a capture device. The preferred route I'd go is to digitize the signal that comes off the heads and do the rest in software. There is little use fighting against early 1980s image processing in VCRs.
Yeah, I was thinking the same thing. Since the knobs board has basically converted the signals to digital, I could use an Arduino to interface with a PC to provide digital controls. That would be fairly easy to do. Someone mentioned the RF digitization of the doomsday duplicator device. I think the implementation of a TBC in code seems like the ultimate. I’m wondering if more in depth gathering of Gordon’s knowledge here would be helpful to such efforts? Or possibly even Gordon considering making his software open source at some point? This was a great video and something I never really expected to see! Thanks!
@@instantwow If someone is patient enough there are capture devices that show up online with built in line and frame TBCs, I personally have the Brighteye75 and S&W TBS800 they are better than any conventional TBC that converts back to analog. The output is SD SDI 4:2:2 10bit, so a SDI to USB adapter is needed.
@@Capturing-Memories Well from a processing time point of view, there's no reason why it shouldn't run in roughly real-time on modest hardware. TBCs aren't that complex, particularly if you have the RF off the tape. What is complicated is taking a composite signal and splitting chroma and luma.
Great Video. Was tempted to get his last one, but looks like someone else got there first. Maybe time to get upgraded formware for my old one. Talking of TBCs, I am transferring an old VHS-C tape for someone and one section has a video unstable picture. I have an old ACE, and a Datavideo TBC-1000 but the only "TBC" that solved the issue completely for this particular tape is the one built into my JVC HR-S9850. The ACE helped somewhat and the Datavideo made no difference.
Something I've always thought is that if a timebase corrector converts the signal to digital before correcting it, why does it need to convert it back to analogue to be captured for digital? Is it because it wasn't designed for a capture workflow? If anyone wanted to make a modern version of a TBC, I would guess the best thing to do would be to leave the first digital conversion as is, and capture that.
Hi - you're briefly covering video delay caused by the proces of going through a TBC. Do you use an audio delay to compensate for this when digitising? The one or two field delay caused by the unit discussed in this video most people won't notice, but still, it's a "flaw" in the proces of digitising, as I see it. Especially if you're doing professionel work. I have a couple of FOR-A F310 TBC's I use for digitizing from U-Matic (with DUB in). I Haven't noticed any problems, but it seems logically that some delay must ocur so I've bought some digital audio delays to compensate for this. What's your experience regarding this problem? None of my clients have ever complaint, but recently I was asked to edit some stuff I had digitized for a client, and I ended up having to move the audio two frames from the video for all clips, to make it visually perfectly lip sync . This was before I bought the audio delay boxes... haven't had the time to test digitizing with both TBC and audio delay yet, but I would like to hear your thoughts about this challange of digitizing. TIA, a huge fan of your video channel
I don't add an audio delay because you really can't detect one frame. But audio delay could be inserted after the video file is captured if it was found to be necessary by the end customer.
Hi, thank you for what you are doing, it is very interesting and informative! Please help me with advice, how and where can I buy ACE timebase correctors? I am from Russia and it is just unreal to find this device here.
They occasionally appear on eBay, especially UK eBay. Be careful to get the version 2 which has the timebase corrector and converter, the early version is not so useful. One was on UK eBay a few weeks ago (auction number 266774818906).
Actually it's odd that we are still talking about TBCs in the context of digitisation. After all if your capture device would have essentially the same kind of input your TBC has, there wouldn't be any problem with it.
Well really what would be nice was if analogue to digital converters had a built in TBC front-end, then we could miss out a D/A and A/D step. I believe a few professional TBCs have an SDI output.
I'm fascinated now, why did vhs and beta drop the colour information by one line per generation? Was the first line then monochrome? The colour subcarrier must still be present on the first line of picture 🤔
Gordon is a true gentleman and of course a technical genius. I have had several interactions with him and he was very kind and helpful going out of his way to help me. Thank you, Gordon.
I have been thinking a bit about this video, and I now understand why the ACE isn't ideal for digitizing video tapes. Essentially it was created for a completely different use case, and that was frame synchronisation. Back in the old days if you wanted to have wipes or fades you needed a video mixer. Professional ones are (conceptually) rather simple devices as all your sources are completely in sync. You could essentially have an "audio-mixer" capable of 5MHz to fade between your sources. Now with video cameras there were some better prosumer cameras which were able to genlock into an existing signal, but if you dealt with consumer VTRs getting those to sync was virtually impossible. So what you needed was a frame synchronizer to sync whatever signal you got to the rest of your studio. Some consumer video mixers included those, but that was a rather expensive component in them. Since the frame store was expensive, they often used it to also do "framestore effects" like still or mosaic. And that is the main usecase for those ACEs, that's why they do things like trying to swap fields. The whole idea is to change video that is off sync to its reference sync. So if you have a VTR running to fast or to slow, it'll drop or add fields so it gets constant 50fields per second.
If I'd design such a device with the possibilities of today, I'd go a different route. I would first try to find the horizontal and vertical sync pulses as well as the drop-out/discontinuity caused by the head switching. I'd then smooth over the horizontal sync pulses a bit so that noise on them is less important, if an impulse is missing I'd add it. Then I'd determine in which line the vertical sync pulse is and number the horizontal sync pulses based on this. If the vertical sync-pulse is lost by a drop-out I'd interpolate it from previous and next frames/fields. If that fails I'd fall back to declaring the middle of the space between the head switching to be the middle of the frame/field. Once I know where each line starts and ends, it's trivial to "unstretch" it and generate an image out of it.
I'd detect the sync pulses like this:
Horizontal sync: Cross correlation with a reference pulse. (essentially averaging over it's length)
Vertical sync: Cross correlation with a reference pulse. (probably via actual cross correlation as the pulse is rather complex, alternatively averaging over several lines)
Head switching: Sudden jump in the period of the horizontal sync pulses.
An extremely interesting conversation! I’ve been using my ACE for many years and even though I’ve been retired since 2000, I’ve hung on to it for when a friend or relative asks me to do a conversion from an ancient tape! I’ve now reached the point where I just have to clear out some equipment and I’ve just listed my ACE on eBay, so it will be interesting to see just how much it goes for now!
WOW, pity I am so late to this video. I have a Mark 2 which I purchased new in Australia via an import. I will NEVER sell mine, it is simply amazing. Nice to know I could get firmware upgrades of needed and repairs. Great video and what a lovely clever man, well done mate !
Great interview and what a nice chap. Reminds me of Tony Murray (Murraypro) who makes the ppm monitoring boxes and test equipment (and probably a similar age). I guess I’m this era of raspberry pi’s etc, there are more opportunities for one man band operations, although there is less of market, though having said that, the archive market has created a new demand.
Thank you very much for setting up and creating this fantastic interview, I am lucky enough to own a Mark 2 (TBC) version of these wonderful units.
Nice interview. I was hoping to see some of these nice features in use.
Just a brief interest, in audio and video. Butj ust to say thanks for going through all the effort to get it on video, times move fast you've got it on video for many years to come. Interesting to watch thanks
What a loverly man, he didn't seem to mind all the questions and i'm sure you learned a lot :-D
At 70+ i would not expect him to have much interest in something he created a long time go.
But he seemed to enjoy the conversation :-D
Enjoy and appreciate the magic he created :-D
A very clever man indeed :-D
Brilliant interview. I would love to hear more about his early years working for Decca. I bet it's fascinating.
That was an amazing interview! Lots of interesting in-depth information, and not a second too long or short. Thanks for posting!
A delightful and fascinating conversation! I'm curious how this purpose could be achieved with technology we have today. What would a 2022 prosumer TBC look like?
Well you can buy such a thing. It's essentially an "all in one" TV-IC connected to a LVDS to HDMI converter IC. Those are available as black metal enclosed SCART+HDMI to HDMI converters.
@@wrtlpfmpfI am very interested in this. Can you please give us a specific product to search for? One that has a TBC, of course.
@@alphabeets Those things come from China under a variety of names, but search for SCART->HDMI converters. They should all be based on that basic idea. The disadvantage is of course that they scale the image.
What an interesting conversation about a great and often under appreciated area of audio/visual technology. Many thanks.
What a fascinating interview, and what a nice chap! I enjoyed every second of that. He left BT Labs two years before I started my career there as a Graduate :-)
Brilliant interview
For U-Matic the chroma noise reduction that shifts the colour down half a line was an optional feature of some TBCs.
Picture's Always Lovely.
If a new design to be sought it would be by today's standards, Only digital out needed (USB4 preferred), no need to physical knobs and buttons, 3D comb filtering for composite sources, better DNR for chroma and luma, Full field and line correction at 12 bit, All color formats possible, 24 bit audio.
Well you are probably looking for a capture device. The preferred route I'd go is to digitize the signal that comes off the heads and do the rest in software. There is little use fighting against early 1980s image processing in VCRs.
@@wrtlpfmpf If you are talking about vhs decode project it is going too slow.
Yeah, I was thinking the same thing. Since the knobs board has basically converted the signals to digital, I could use an Arduino to interface with a PC to provide digital controls. That would be fairly easy to do.
Someone mentioned the RF digitization of the doomsday duplicator device. I think the implementation of a TBC in code seems like the ultimate. I’m wondering if more in depth gathering of Gordon’s knowledge here would be helpful to such efforts? Or possibly even Gordon considering making his software open source at some point?
This was a great video and something I never really expected to see! Thanks!
@@instantwow If someone is patient enough there are capture devices that show up online with built in line and frame TBCs, I personally have the Brighteye75 and S&W TBS800 they are better than any conventional TBC that converts back to analog. The output is SD SDI 4:2:2 10bit, so a SDI to USB adapter is needed.
@@Capturing-Memories Well from a processing time point of view, there's no reason why it shouldn't run in roughly real-time on modest hardware. TBCs aren't that complex, particularly if you have the RF off the tape.
What is complicated is taking a composite signal and splitting chroma and luma.
Very interesting interview 👍
You know it's good stuff when you see large chips in the device. Use your better microphone next time.
I did use the good microphone but alas there was a connection fault which resulted in poor sound. Just had to live with it.
Great Video. Was tempted to get his last one, but looks like someone else got there first. Maybe time to get upgraded formware for my old one.
Talking of TBCs, I am transferring an old VHS-C tape for someone and one section has a video unstable picture. I have an old ACE, and a Datavideo TBC-1000 but the only "TBC" that solved the issue completely for this particular tape is the one built into my JVC HR-S9850. The ACE helped somewhat and the Datavideo made no difference.
I thought it would be sold pretty quick.
He didn’t mention who took over Baird, but it was BRC/Thorn.
Something I've always thought is that if a timebase corrector converts the signal to digital before correcting it, why does it need to convert it back to analogue to be captured for digital? Is it because it wasn't designed for a capture workflow? If anyone wanted to make a modern version of a TBC, I would guess the best thing to do would be to leave the first digital conversion as is, and capture that.
I mention exactly that point in the video I think. If someone could create a new TBC with SDI and Firewire outputs, that would certainly sell.
Hi - you're briefly covering video delay caused by the proces of going through a TBC. Do you use an audio delay to compensate for this when digitising?
The one or two field delay caused by the unit discussed in this video most people won't notice, but still, it's a "flaw" in the proces of digitising, as I see it. Especially if you're doing professionel work.
I have a couple of FOR-A F310 TBC's I use for digitizing from U-Matic (with DUB in). I Haven't noticed any problems, but it seems logically that some delay must ocur so I've bought some digital audio delays to compensate for this.
What's your experience regarding this problem?
None of my clients have ever complaint, but recently I was asked to edit some stuff I had digitized for a client, and I ended up having to move the audio two frames from the video for all clips, to make it visually perfectly lip sync . This was before I bought the audio delay boxes... haven't had the time to test digitizing with both TBC and audio delay yet, but I would like to hear your thoughts about this challange of digitizing.
TIA, a huge fan of your video channel
I don't add an audio delay because you really can't detect one frame. But audio delay could be inserted after the video file is captured if it was found to be necessary by the end customer.
Would it be possible to build one of these diy? Using the schematics?
Hi, thank you for what you are doing, it is very interesting and informative! Please help me with advice, how and where can I buy ACE timebase correctors? I am from Russia and it is just unreal to find this device here.
They occasionally appear on eBay, especially UK eBay. Be careful to get the version 2 which has the timebase corrector and converter, the early version is not so useful. One was on UK eBay a few weeks ago (auction number 266774818906).
Actually it's odd that we are still talking about TBCs in the context of digitisation. After all if your capture device would have essentially the same kind of input your TBC has, there wouldn't be any problem with it.
Well really what would be nice was if analogue to digital converters had a built in TBC front-end, then we could miss out a D/A and A/D step. I believe a few professional TBCs have an SDI output.
I'm fascinated now, why did vhs and beta drop the colour information by one line per generation? Was the first line then monochrome? The colour subcarrier must still be present on the first line of picture 🤔
I have to confess, I don't know the answer to that. It was an engineering compromise but I don't know why in detail.
If these Ace T.B.C. available anywhere,if you can share link
The one we discussed in the video has now been sold. It's best just to keep an eye out on eBay and similar.