OSSC Pro

[orange808]

If this alternative firmware is really feasible and the resources a real limit, I wonder if it will be interesting to chain two ossc pro. One only doing deinterlacing, the other doing the rest.
I'm sure there will be people interested by that and ok to pay the price, while still keeping the price of only one ossc "reasonable" for oher pepole.

If possible, can you share the specific feature(s) of the current annouced OSSC Pro specs that: makes you want to choose the OSSC Pro over the Retrotink5x?

I worry that my own list isn't long enough. I admit the Retrotink5x isn't available or finished, but it appears to be tough competition.

If the Retrotink5x delivers robust features for less, I'm concerned the OSSC Pro will not find a large audience.

If the Retrotink5x already embodies what the low end market considers to be "reasonable", where does that leave the OSSC Pro?

[Joelepain]

To be honest, I don't know anything about what the Retrotink5x promises to offer.
I was more reflecting about marqs and Fudoh suggestion about the deinterlacing, which seems to be quite a sensitive subject (and if the retrotink doesnt use a beefier FPGA, won't really offer more than the OSSC).

[orange808]

To be honest, I don't know anything about what the Retrotink5x promises to offer.

But, you will know when you choose which machine to buy.

I'm not trying to upset anybody, but I am doubtful the current motion adaptive deinterlacing will outperform the Retrotink5x. The beefier FPGA may not be enough of an improvement to differentiate itself. Both machines appear to be using similar approaches and the Retrotink5x seems to have enough power to handle it.

Yes, a custom firmware would be really good, but how many people will buy a machine for only 480i?

[fernan1234]

This forum will make you believe that most of the market for this kind of device is really clamoring for cutting-edge deinterlacing (or at least deinterlacing as good as that found in some old dedicated video processors that came out during the SD TV era's twilight). But I'm pretty sure most people buying the RT5Xs and OSSC Pros of the world care mainly about scaling to the higher resolution flat panels of today with the highest compatibility possible. A smaller but significant number will also care about the best scaling and compatibility for video capture and streaming purposes.

What I feel sure about is that the RT5X will stick to the philosophy of versatility in input options and virtual plug-and-play usage, whereas the OSSC Pro will be open to new implementations and features throughout much of its lifespan, and that can include great deinterlacing options whether in the stock FW or a separate one.

Earlier today I was on the A-series BVM sync issues topic talking about how a lot of those problems would be solved with a video processor that can normalize the refresh rate of certain sources like arcade boards, Neo Geo MVS, and their corresponding MiSTer cores, without needing scaling for the CRT. I'd bet the OSSC Pro will be able to do that, with the added convenience of having an HDMI input for stuff MiSTer cores in direct video mode. What else in the market will let us take care of this kind of "edge use case" conveniently, if at all?

[orange808]

This forum will make you believe that most of the market for this kind of device is really clamoring for cutting-edge deinterlacing (or at least deinterlacing as good as that found in some old dedicated video processors that came out during the SD TV era's twilight).

Actually, we're seeking something better. Lots of us already own those old processors.

This is bigger than just deinterlacing, though. I'm wondering if the OSSC Pro will find a market. The Retrotink5x is already planning to almost match the announced OSSC Pro.

Video machines are a hobby for me and I'm not sure if the OSSC Pro would be my choice right now. That probably doesn't matter, but there it is.

But I'm pretty sure most people buying the RT5Xs and OSSC Pros of the world care mainly about scaling to the higher resolution flat panels of today with the highest compatibility possible. A smaller but significant number will also care about the best scaling and compatibility for video capture and streaming purposes.

So, 1440p will be enough to carry the OSSC Pro? Maybe.

For compatibility, the Retrotink5x is going to offer a nice standard and compatible 1080p. Sounds like it will do a decent job with sampling as well.

What I feel sure about is that the RT5X will stick to the philosophy of versatility in input options and virtual plug-and-play usage, whereas the OSSC Pro will be open to new implementations and features throughout much of its lifespan, and that can include great deinterlacing options whether in the stock FW or a separate one.

Which specific new features are you expecting? Tate mode sounds nice.

Earlier today I was on the A-series BVM sync issues topic talking a
bout how a lot of those problems would be solved with a video processor that can normalize the refresh rate of certain sources like arcade boards, Neo Geo MVS, and their corresponding MiSTer cores, without needing scaling for the CRT. I'd bet the OSSC Pro will be able to do that, with the added convenience of having an HDMI input for stuff MiSTer cores in direct video mode. What else in the market will let us take care of this kind of "edge use case" conveniently, if at all?

The MiSTer cores I use already offer ways to standardize the output--although all my CRTs seem to play nice with analog output without concerns. What you're suggesting is frame rate conversion. That can't be done effectively without latency. So, the MiSTer is probably already doing the best you can expect--or it will in the near future.

I didn't know the Neo Geo MVS didn't work on pro monitors. That's odd.

[fernan1234]

The MiSTer cores I use already offer ways to standardize the output--although all my CRTs seem to play nice with analog output without concerns. What you're suggesting is frame rate conversion. That can't be done effectively without latency. So, the MiSTer is probably already doing the best you can expect--or it will in the near future.

I didn't know the Neo Geo MVS didn't work on pro monitors. That's odd.

Not talking about pro monitors in general, but really only a few models/series that came out as the last few pro CRT monitors manufactured, the A-series Sony BVMs as well as some late Ikegami and JVC monitors that are highly intolerant of off-spec refresh rates like those found in arcade boards, MVS, and of course the MiSTer cores that operate in the same refresh rates.

And yes, what I suggested would most likely come at the cost of some latency, but for many of us it would be an acceptable compromise in the case of these particular monitors. We're already also going to accept some latency for the benefit of features that require frame buffering.

The bigger point is that only the OSSC Pro is giving us hope for solutions to many of this kind of niche problems.

[marqs]

This is bigger than just deinterlacing, though. I'm wondering if the OSSC Pro will find a market. The Retrotink5x is already planning to almost match the announced OSSC Pro.

Video machines are a hobby for me and I'm not sure if the OSSC Pro would be my choice right now. That probably doesn't matter, but there it is.

Many people forget that OSSC Pro is also a hobby project, done with limited time and budget. Upgrading the now near-final HW when something possibly similar is coming out isn't very tempting - it would be another 6-12 month delay and money drowned on prototypes etc.

[NormalFish]

Many people forget that OSSC Pro is also a hobby project, done with limited time and budget. Upgrading the now near-final HW when something possibly similar is coming out isn't very tempting - it would be another 6-12 month delay and money drowned on prototypes etc.

Not to mention any competition that does emerge isn't really an issue. Doubly so for an open source design. Different distribution methods might make one more economical for one region and another for another region, or one device might work better for a current pre-existing set up than another. It gives the community more options. Hardly an issue that needs solving. Let alone by multiplying the cost of the device.

[Fudoh]

@fernan1234

This forum will make you believe that most of the market for this kind of device is really clamoring for cutting-edge deinterlacing (or at least deinterlacing as good as that found in some old dedicated video processors that came out during the SD TV era's twilight).

point is that the OSSC has been a pretty good recommendation since it's release, but for me it always came with that asterisk that said "unless you need really good 480i handling, then look into the FM" and it would just be such a delight to finally get an "ultimate" machine to just end all that discussion. The reason why I mentioned (or asked for) the possibility of doing an alternative FW branch at some point is that it hurts a little see a POSSIBLY superior deinterlacing algorithm readily available (through Intel's SDK), while the OSSC Pro settles for a down scaled version of it because of the plethora of other features. To do it like this makes sense of course, but possibly creating a different FW down the road can't hurt either. **

I agree that the RT5X is a strong contender, especially for those who want an upscaler to be as much plug and play as possible and for those it might easily the better choice over the OSSC, but I would also argue that for those the FM has always been a better solution than the OSSC to start with. I see the RT5X much more as a spiritual successor to the Micomsoft line of processors and especially to a be drop-in Framemeister replacement, while the OSSC Pro plays on a completely different level in terms of options and possibilities.

The most obvious differences right from the most basic specs are that the 5X doesn't have a VGA or HDMI input, while the OSSC has both and there comes a lot of possibilities by utilizing them.


(** just wanting to clarify that we still only talk video deinterlacing. Cadence based film deinterlacing is a whole different topic and I would never expect any processor aimed at video gaming to venture into that territories.)

[fernan1234]

The most obvious differences right from the most basic specs are that the 5X doesn't have a VGA or HDMI input, while the OSSC has both and there comes a lot of possibilities by utilizing them.

These two connectors are absolutely of tremendous value. As someone who doesn't use SCART for RGB, having the "VGA" connector is a great convenience. The OSSC Classic has one too but with the limitation of lacking filter options for it, but the Pro will address that. More beneficial in general, the HD15 input makes it easier to integrate or chain the OSSC Pro with many other devices that use it or BNC for connections.

But IMO the HDMI input is much more massive in its potential. It really opens up the OSSC Pro to go way beyond retro gaming. Modern game and video can greatly benefit in many more ways than those that have been brought up here already.

I do want to clarify that my previous comment about deinterlacing should not be taken as a dismissal of the value in implementing the best possible options on this device. Even if it is not at the top of the list of people buying this device after release, it would definitely be a cause for celebration for all video enthusiasts if the OSSC Pro can allow people to stop chasing after all FMs and other video processors out there for this one purpose. Pretty sure that the Pro's unique features and possibilities will speak for themselves once it hits the market and faces competition.

[nmalinoski]

The most obvious differences right from the most basic specs are that the 5X doesn't have a VGA or HDMI input, while the OSSC has both and there comes a lot of possibilities by utilizing them.

These two connectors are absolutely of tremendous value. As someone who doesn't use SCART for RGB, having the "VGA" connector is a great convenience. The OSSC Classic has one too but with the limitation of lacking filter options for it, but the Pro will address that. More beneficial in general, the HD15 input makes it easier to integrate or chain the OSSC Pro with many other devices that use it or BNC for connections.

But IMO the HDMI input is much more massive in its potential. It really opens up the OSSC Pro to go way beyond retro gaming. Modern game and video can greatly benefit in many more ways than those that have been brought up here already.

Building on this, one of the biggest differentiators for me is that the OSSC Pro has discrete inputs, whereas the RT5X, like the RT2X products before it, appears to only have a single analogue stereo input.

I quite like that I can have both SCART and YPbPr component sources hooked up to my OSSC at the same time, and that I can cycle between them with the automatic input switching. This will carry over to the OSSC Pro, with the addition of HDMI and TOSLINK inputs. I'm particularly excited about the TOSLINK input, because the component switcher in my YPbPr pipeline also switches TOSLINK, and I've got a PS2, Xbox, and Xbox 360 (Xenon) that can all take advantage of that digital audio.

On the other hand, if I were to use an RT5X, sure, I'd be able to hook up composite and S-Video sources without extra equipment, but then I'd either have to manually disconnect and reconnect each format-specific pipeline or add a bunch of extra equipment to consolidate composite/S-Video/YPbPr/RGBS to one format/connection (Or buy multiple RT5X units, which I'd not be keen on doing).

[Guspaz]

The original OSSC had zero or one audio inputs, depending on the revision, and that was never really a problem. You don't need fancy hardware or multiple devices (be it an OSSC or RT5X) to handle that scenario. Just use a cheap stereo audio switch, or use y connectors and only turn on one device type at a time. Most audio switching would be handled automatically by your component or SCART switch. It's possible that the RT5X has internal audio switching for SCART versus CVBS/SVID/YPbPr, I have no idea.

[Gunstar]

If this is something that can be implemented without too much work, I'd be very VERY interested in an alternative firmware down the road just geared towards 480i. I'm pretty sure you'd be reaching a whole new customer base by making something like this available.

+1 It would be amazing to get top tier 480i deinterlacing on the pro with that Intel tech Fudoh mentioned

[bobrocks95]

An HDMI input remains very appealing for my GCVideo, but the automatic adjustment of the Retrotink sounds very nice for me. I love the adjustments the OSSC, but generally I just leave it on generic x5 and don't touch anything else.

[6t8k]

Not being open source, I didn't expect Intel VIP as an option. As I understood it, the idea behind it is to have a relatively quick way to fit the OSSC Pro with weave/motion adaptive deinterlacing capabilities, so that the device can ship in a timely manner without having gaps in these places of the projected feature set. Isn't it a separate question whether the amount of logic area used, casting doubt on the economic feasibility, is due to the inherent complexity of 'good' deinterlacing (be it 'Framemeister-good', or 'sobel edge interpolation-good' - we haven't seen the latter in action here yet) or if it's specifically Intel's implementation that uses more resources than necessary? If it's the former then it would be a shame, even if the issue could be alleviated by providing multiple flavors of firmware. If it's the latter, perhaps a custom implementation could still provide good deinterlacing for video games while using up less resources than Intel's implementation.

[strayan]

Will the OSSO Pro be able to display (on screen) the exact modeline it’s receiving on its HDMI input?

[nmalinoski]

Will the OSSO Pro be able to display (on screen) the exact modeline it’s receiving on its HDMI input?

Assuming it either has or will get an OSD at some point, if you've got it in a line-buffered mode, I imagine it'll be the same as the OSSC--will only work if the connected display supports the resulting video signal--however, in any of the framebuffered modes, it should be able to always display the incoming video mode.

[strayan]

Whoops, I meant will it be able to display details about the signal it’s receiving i.e. CVT, CVT RB, DMT, GTF, CEA compliant etc and refresh rate, vsync, hsync, front porch, back porch etc.

[marqs]

Not being open source, I didn't expect Intel VIP as an option. As I understood it, the idea behind it is to have a relatively quick way to fit the OSSC Pro with weave/motion adaptive deinterlacing capabilities, so that the device can ship in a timely manner without having gaps in these places of the projected feature set. Isn't it a separate question whether the amount of logic area used, casting doubt on the economic feasibility, is due to the inherent complexity of 'good' deinterlacing (be it 'Framemeister-good', or 'sobel edge interpolation-good' - we haven't seen the latter in action here yet) or if it's specifically Intel's implementation that uses more resources than necessary? If it's the former then it would be a shame, even if the issue could be alleviated by providing multiple flavors of firmware. If it's the latter, perhaps a custom implementation could still provide good deinterlacing for video games while using up less resources than Intel's implementation.

The primary reason for trying out VIP is to ensure the FPGA performance and resources are sufficient for scaler mode features as it's too late to upgrade HW once it's in production. It also helped verifying there are no FPGA/memory related PCB issues. Secondary reason is evaluation of the IP itself, with the consideration it could be actually used to implement the features, at least until custom-designed blocks are available (which might be a long time). So far it's been relatively simple to integrate VIP deinterlacer+scaler+framebuffer chain and it seems to offer enough flexibility to be suitable for gaming content too. Deinterlacer is particulary heavy of resource usage and perfromance requirements (1080i60 was not trivial to get running), but I don't think any motion adaptive implementation would be particulary light. At least here are 2 options, and the basic one is already relatively good.

Whoops, I meant will it be able to display details about the signal it’s receiving i.e. CVT, CVT RB, DMT, GTF, CEA compliant etc and refresh rate, vsync, hsync, front porch, back porch etc.

It currently displays basic timings on information screen as in this example video (0:08->). With analog sources V.total obviously is the only true signal parameter while rest are from preset, but with digital sources all are from the signal itself.

[PearlJammzz]

An HDMI input remains very appealing for my GCVideo, but the automatic adjustment of the Retrotink sounds very nice for me. I love the adjustments the OSSC, but generally I just leave it on generic x5 and don't touch anything else.

I haven't seen any mention of this but I reckon the OSSC Pro could probably also get this feature. I'll be curious to see how the RT5X looks compared to an OSSC with optimal timings and generic. I reckon the RT5X will look way closer to generic than optimal. Generic is what I steer most people towards just cause it makes the device way more usable. I shouldn't have to switch between 2-3 profiles depending on which display output the game is in at the moment on my Saturn for example. It's a PITA. Even just being able to set 'Saturn' and then have a profile that has all the dialed in settings for every resolution that console has would be sufficient but I am not sure if there is a way for the OSSC to tell exactly what it's getting every time.

Regardless, I wouldn't count the feature out on the OSSC and I'd also wait to see what the RT5X looks like and performs like when it comes out before I tie my cart to that horse.

[BuckoA51]

I haven't seen any mention of this but I reckon the OSSC Pro could probably also get this feature.

The RetroTINK 5x is based on the old Cyclone 4 FPGA, same as OSSC Classic. Anything a Cyclone 4 can do a Cylcone 5 certainly can.

[marqs]

Anything a Cyclone 4 can do a Cylcone 5 certainly can.

It's actually not that simple, even if Cyclone V are 28nm chips while Cyclone IV are 65nm. Both families come with different sizes and speed grades - largest and fastest Cyclone IV is superior to smallest and slowest Cyclone V, but naturally it costs more as well. The model used in OSSC Pro is on the mid-range of Cyclone V family - any larger or faster variant would be significantly more expensive.

[vol.2]

any larger or faster variant would be significantly more expensive.

Would it be possible to "upgrade" the Cyclone 5 to a faster variant after the fact (as an end-user)?

[ldeveraux]

any larger or faster variant would be significantly more expensive.

Would it be possible to "upgrade" the Cyclone 5 to a faster variant after the fact (as an end-user)?

It's an FPGA...

[vol.2]

any larger or faster variant would be significantly more expensive.

Would it be possible to "upgrade" the Cyclone 5 to a faster variant after the fact (as an end-user)?

It's an FPGA...

Yeah, I don't know anything about the Cyclone stuff at all. I don't think it being an FPGA is a limitation in and of itself though. I have some limited experience with Xilinx FPGAs, and used upgraded parts in the Spartan range that were pin compatible and only required a (minor) firmware rewrite and recompile.

Sorry if it was a stupid question though. I guess there's something obvious I'm missing.

[ldeveraux]

Yeah, I don't know anything about the Cyclone stuff at all. I don't think it being an FPGA is a limitation in and of itself though. I have some limited experience with Xilinx FPGAs, and used upgraded parts in the Spartan range that were pin compatible and only required a (minor) firmware rewrite and recompile.

Sorry if it was a stupid question though. I guess there's something obvious I'm missing.

It's not a stupid question, it's just that everything on an FPGA board is soldered down. You can't pull it out and replace the chip like a socketed board.

[maxtherabbit]

Yeah, I don't know anything about the Cyclone stuff at all. I don't think it being an FPGA is a limitation in and of itself though. I have some limited experience with Xilinx FPGAs, and used upgraded parts in the Spartan range that were pin compatible and only required a (minor) firmware rewrite and recompile.

Sorry if it was a stupid question though. I guess there's something obvious I'm missing.

It's not a stupid question, it's just that everything on an FPGA board is soldered down. You can't pull it out and replace the chip like a socketed board.

no shit sherlock, he's well capable of doing the soldering

[fernan1234]

The bigger issue is probably that it may not always be very straightforward to port the firmware to a chip with higher specs even if within the same family.

[ldeveraux]

no shit sherlock, he's well capable of doing the soldering

I'm supposed to know that some guy one the internet is a master solderer?

[maxtherabbit]

no shit sherlock, he's well capable of doing the soldering

I'm supposed to know that some guy one the internet is a master solderer?

not really, but you could have at least given him the benefit of the doubt that he was asking about the logical ramifications of changing FPGAs and had already accounted for the physical before even asking