OSSC Pro

The place for all discussion on gaming hardware
Post Reply
User avatar
Gunstar
Posts: 655
Joined: Wed Apr 09, 2014 10:29 am
Location: UK

Re: OSSC Pro

Post by Gunstar »

charlizardon wrote: Fri May 22, 2026 1:21 pm
Gunstar wrote: Fri May 22, 2026 10:18 am
marqs wrote: Thu May 21, 2026 8:28 pm It's not possible to feed CVBS or Y/C directly into Pro or Extra AV out card. I'll need to get in touch with original designer and request update to both Legacy and Extra AV cards
That would be great if such a card could be made. Thank you, marqs.
I think he meant to get in touch with the original designer of the .stl case files. Two different answers.
Ahhh lol
BuckoA51 wrote: Fri May 22, 2026 2:02 pm You could always just buy two OSSC Pros :mrgreen:
Finally the OSSC Elite 4k! :mrgreen:
User avatar
VEGETA
Posts: 483
Joined: Mon May 31, 2021 10:40 am

Re: OSSC Pro

Post by VEGETA »

Finally the OSSC Elite 4k! :mrgreen:
for a moment I thought the ossc pro will support 4k60. :D


_______

I wanted to ask about composite and s-video input card, what ADC will it use? I saw morph 4k card uses TW9912 as well as RT5X but not sure about RT4K, the tinks say they have time-base correction (4k one at least) but this particular chip does not do it.

I previously spoke about ISL51002 being obsolete a time ago, and alternatives like adv7842 is available which i believe has cvbs and s-video internally as well as full 444 processing. Another interesting chip is adv7802 which only differs that it does not have hdmi input.

So maybe ossc pro next revision can use one of these all in one chips as well as an hdmi 2.0 transmitter so that it allows 4k60 since its fpga is capable of that (similar to tink4k ce). I believe if it uses altera video scaling ip it should be very easy. Just a suggestion to this great platform.

best regards,
Vegeta
User avatar
Konsolkongen
Posts: 2398
Joined: Fri May 16, 2008 8:28 pm
Location: Denmark

Re: OSSC Pro

Post by Konsolkongen »

VEGETA wrote: Sun May 24, 2026 9:59 am I saw morph 4k card uses TW9912 as well as RT5X but not sure about RT4K, the tinks say they have time-base correction (4k one at least) but this particular chip does not do it.
Both the RT4K and Morph use a dual ADC setup using TVP7002 and TW9912 as far as I know.
User avatar
VEGETA
Posts: 483
Joined: Mon May 31, 2021 10:40 am

Re: OSSC Pro

Post by VEGETA »

Konsolkongen wrote: Sun May 24, 2026 10:57 am
VEGETA wrote: Sun May 24, 2026 9:59 am I saw morph 4k card uses TW9912 as well as RT5X but not sure about RT4K, the tinks say they have time-base correction (4k one at least) but this particular chip does not do it.
Both the RT4K and Morph use a dual ADC setup using TVP7002 and TW9912 as far as I know.
Yes, the RT4K uses TVP7002 as seen in some pictures but not sure if it uses TW9912 as this one does not mention time-base correction but rather stuff about weak sync signals handling. Not sure about RT5X though.

Also, according to what I know, it claims 444 handling from start to finish but TW9912 subsamples below 444 for sure.

____

AD7802 seems the all in one solution (and ADV7842 if HDMI 1.4 is needed), which truly has TBC and many other stuff. If dual chip solution, then AD9984A seems more modern, there is also THC7984-17 but this one is not well known or poor documentation and examples. Seems that AD9984A is the best when it comes to just RGB/YPbPr handling without CVBS and YC.

Price-wise, AD7802 is not cheap at all but from other distributors can be ok.

I believe "OSSC Pro Pro" version is worth it for 4k60, but that is not my decision. :wink:
User avatar
Zacabeb
Posts: 39
Joined: Sun Apr 21, 2024 8:53 pm

Re: OSSC Pro

Post by Zacabeb »

Whatever it would use, it would need robust decoding of non-interlaced and non-standard sources, as well as an override to disable the comb filtering for correct decoding of artifact colors. Some of the integrated solutions seem to give very little control over how color decoding is done.

The ADV7280 currently used in the Legacy AV In card is not only lacking a proper TBC, but it also has a hardware/microcode bug making its notch filters apply at the wrong frequencies (6% above the target center frequency.) So that one would be ruled out for several reasons in addition to possible discontinuation.

For component processing, while the THC7984-17 does have a proven track record as it's been used in various Sony TVs in the past, it has an analog H-PLL and may also see discontinuation by THine.
User avatar
VEGETA
Posts: 483
Joined: Mon May 31, 2021 10:40 am

Re: OSSC Pro

Post by VEGETA »

Whatever it would use, it would need robust decoding of non-interlaced and non-standard sources, as well as an override to disable the comb filtering for correct decoding of artifact colors. Some of the integrated solutions seem to give very little control over how color decoding is done.
Are TVP7002 and ISL5002 capable of doing so? how about the suggested AD9984A and ADV7802?
or are you talking about composite?
The ADV7280 currently used in the Legacy AV In card is not only lacking a proper TBC, but it also has a hardware/microcode bug making its notch filters apply at the wrong frequencies (6% above the target center frequency.) So that one would be ruled out for several reasons in addition to possible discontinuation.
As you mentioned, it is obsolete, thus no one should use it. I didn't check thoroughly but seems all these small SD digitizers got discontinued by AD leaving you with the choice of using ADV7802 or ADV7842. There are others but those 2 are the best and most available.
For component processing, while the THC7984-17 does have a proven track record as it's been used in various Sony TVs in the past, it has an analog H-PLL and may also see discontinuation by THine.
I never saw THC7984-17 being used in retro gaming community though.
looks like AD9984A is the direct alternative to it, not sure about your specific PLL needs though. Still, even the tink4k and morph used the old TVP7002 despite being old and 0 support for it. Price could be the reason but TVP7002 is about 9-14$ on Digikey while AD9984A is just 13$ on Rochester which is a reliable supplier of AD chips.

___

Another question I would ask is the scaler IP used in such great devices like RT4K, Morph, and OSSC Pro. OSSC Pro and RT4K being altera FPGA based could mean using the Altera scaler and deinterlacer IP but this is unknown. I read some pages claims custom made but this is gonna be lots of work.
User avatar
marqs
Posts: 1139
Joined: Sat Dec 15, 2012 12:11 pm
Location: Finland

Re: OSSC Pro

Post by marqs »

VEGETA wrote: Sun May 24, 2026 12:41 pm
Whatever it would use, it would need robust decoding of non-interlaced and non-standard sources, as well as an override to disable the comb filtering for correct decoding of artifact colors. Some of the integrated solutions seem to give very little control over how color decoding is done.
Are TVP7002 and ISL5002 capable of doing so? how about the suggested AD9984A and ADV7802?
or are you talking about composite?
They both (especially TVP7002) have issues decoding non-standard sources (e.g. interlace via RGBHV), but optionally provide minimally processed sync outputs which can be used for mode detection etc. on FPGA. You only need their PLL to keep line lock, otherwise you want to ignore most automation the chips provide as it's not robust.
The ADV7280 currently used in the Legacy AV In card is not only lacking a proper TBC, but it also has a hardware/microcode bug making its notch filters apply at the wrong frequencies (6% above the target center frequency.) So that one would be ruled out for several reasons in addition to possible discontinuation.
As you mentioned, it is obsolete, thus no one should use it. I didn't check thoroughly but seems all these small SD digitizers got discontinued by AD leaving you with the choice of using ADV7802 or ADV7842. There are others but those 2 are the best and most available.
All of these chips are more or less obsolete, availability & price just gets worse all the time. I don't see much benefit having on-chip TBC as we're not doing a feedback system with tape player and FPGA can do the usual triple-buffering if needed.
Another question I would ask is the scaler IP used in such great devices like RT4K, Morph, and OSSC Pro. OSSC Pro and RT4K being altera FPGA based could mean using the Altera scaler and deinterlacer IP but this is unknown. I read some pages claims custom made but this is gonna be lots of work.
The video IP does it job but it's hard to recommend it for any project which has more developers at hand. There are hints in its documentation that it originates from Zipcores to which Altera perhaps outsourced development decades ago after which it has been transferred to Intel and now it's obsolete in the sense that nobody cares about the bugs and support is laughable.
User avatar
VEGETA
Posts: 483
Joined: Mon May 31, 2021 10:40 am

Re: OSSC Pro

Post by VEGETA »

All of these chips are more or less obsolete, availability & price just gets worse all the time. I don't see much benefit having on-chip TBC as we're not doing a feedback system with tape player and FPGA can do the usual triple-buffering if needed.
well, they are still active and available except for adv7842 being hdmi but if you have your hdmi license then no problem, can sources from even reliable sources like LCSC. ad7802 is also active, but 7800 is obsolete.

their benefit exceeds the TBC since they offer 444 decoding for all signals in one chip and so on. Its support is still active.
The video IP does it job but it's hard to recommend it for any project which has more developers at hand. There are hints in its documentation that it originates from Zipcores to which Altera perhaps outsourced development decades ago after which it has been transferred to Intel and now it's obsolete in the sense that nobody cares about the bugs and support is laughable.
so you confirm ossc pro use intel altera video scaler ip? what about your custom modes and so on? Also, does tink4k and 5x also use it?

i see now they migrated to a new one called video and vision processing suite (VVP), it is 5000$ on mouser (5k in fpga world seems cheap) but if you read its description, it has everything in it from scratch to end. is it the one you used or the previous one labelled scaller II and deinterlacer II etc...? since they are different, can't find the old ones.

what would you recommend? the old or new one or what exactly?

I am genuinely interested in this as I would like to make my own in the future but i am bad at programming, but really love fpgas and will dedicate time to master them. since you did it all, then you are the right person to ask.

always thanks for everything.
User avatar
marqs
Posts: 1139
Joined: Sat Dec 15, 2012 12:11 pm
Location: Finland

Re: OSSC Pro

Post by marqs »

VEGETA wrote: Sun May 24, 2026 7:39 pm
All of these chips are more or less obsolete, availability & price just gets worse all the time. I don't see much benefit having on-chip TBC as we're not doing a feedback system with tape player and FPGA can do the usual triple-buffering if needed.
well, they are still active and available except for adv7842 being hdmi but if you have your hdmi license then no problem, can sources from even reliable sources like LCSC. ad7802 is also active, but 7800 is obsolete.

their benefit exceeds the TBC since they offer 444 decoding for all signals in one chip and so on. Its support is still active.
My intention was to try ADV7842 but never had time for it. Looks promising on paper, but clearly its focus is on HDMI receiver and not the legacy inputs. If it was superior for retro stuff with good availability & price, I guess it would have been already utilized by others.
so you confirm ossc pro use intel altera video scaler ip? what about your custom modes and so on? Also, does tink4k and 5x also use it?
It uses the Intel VIP suite (not the new/repackaged VVP) as stated earlier and in README. I'm not the right person to answer for Retrotinks.
User avatar
VEGETA
Posts: 483
Joined: Mon May 31, 2021 10:40 am

Re: OSSC Pro

Post by VEGETA »

My intention was to try ADV7842 but never had time for it. Looks promising on paper, but clearly its focus is on HDMI receiver and not the legacy inputs. If it was superior for retro stuff with good availability & price, I guess it would have been already utilized by others.
well, it is kinda hard to get the evaluation board because laser disc community starting purchasing it as an end-user product at some point + being HDMI and licensing.
It uses the Intel VIP suite (not the new/repackaged VVP) as stated earlier and in README. I'm not the right person to answer for Retrotinks.
so you feed the 24-bit or 30-bit video data coming from the ADC to the FPGA into the scaler/deinterlacer ip blocks and then do the control? where can I learn about this? i will get a dev board soon.

How much did it cost if you can answer that? I read somewhere that it is not paid (the old one).

you and the tink say custom polyphase therefore i wanted to ask, but in your experience is it hard to build something like that or better from scratch or better just use the vendor ip?
User avatar
marqs
Posts: 1139
Joined: Sat Dec 15, 2012 12:11 pm
Location: Finland

Re: OSSC Pro

Post by marqs »

VEGETA wrote: Sun May 24, 2026 10:51 pm so you feed the 24-bit or 30-bit video data coming from the ADC to the FPGA into the scaler/deinterlacer ip blocks and then do the control? where can I learn about this? i will get a dev board soon.
In simple terms, yes. There is still quite a bit of custom logic before and after VIP pipeline. You should look into the documentation provided by Intel to learn how the IP blocks are meant to be used.
How much did it cost if you can answer that? I read somewhere that it is not paid (the old one).
I recall it was somewhere around the sum you mentioned. Not awfully lot within ASIC/FPGA IP world, but not peanuts either considering low quantities and margin of the end product.
you and the tink say custom polyphase therefore i wanted to ask, but in your experience is it hard to build something like that or better from scratch or better just use the vendor ip?
The polyphase scaler is probably the most "standard" block of the suite and its result should look exactly same as a custom implementation. Only the optional edge-adaptive mode can make it more specialized.
User avatar
VEGETA
Posts: 483
Joined: Mon May 31, 2021 10:40 am

Re: OSSC Pro

Post by VEGETA »

In simple terms, yes. There is still quite a bit of custom logic before and after VIP pipeline. You should look into the documentation provided by Intel to learn how the IP blocks are meant to be used.
what type of other logic is necessary? I believe the fpga will interface with a microcontroller via spi or i2c so that the MCU sends commands to control the ips inside?

I plan to get one of the modern new gen Altera dev boards which should be capable of doing all, I am excited for this but need to settle some stuff before I start. where do you recommend I start in FPGA world? I suck at regular programming though.
I recall it was somewhere around the sum you mentioned. Not awfully lot within ASIC/FPGA IP world, but not peanuts either considering low quantities and margin of the end product.
Unfortunately. FPGA world only likes triple digits. 5000$ seems "ok" price and I even saw zipcores 4k video ip listed at 9000$! HDMI 2.1 IP from Altera is 15K (without actual HDMI license).
The question is: did you had the ability to freely develop and test with this IP and its blocks before to purchasing it? because I feel this is a must.

I read somewhere that even HDMI 2.1 IP is freely usable and synthesize-able but once you disconnect your official dev board programming cable it turns off after some time. what is your experience on this when you used their IPs?

The polyphase scaler is probably the most "standard" block of the suite and its result should look exactly same as a custom implementation. Only the optional edge-adaptive mode can make it more specialized.
so you did not had to write it up from start but rather used it immediately.
by custom or edge adaptive meaning you had to write custom coefficients in case traditional ones did not do it nicely?
My understanding is that it first de-interlaces then scales the lines... but doing motion adaptive deinterlacing, how many frames or mS of latency does it add? If you want it you should at least wait for one field to gather, then once the next field arrives pixel by pixel you can output the correct result.
Post Reply