OSSC Pro

[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?

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

I don't assume anybody knows anything anymore, else why would they ask a question. Sorry I tried to help with what I thought was a basic question.

[NormalFish]

I don't assume anybody knows anything anymore, else why would they ask a question. Sorry I tried to help with what I thought was a basic question.

Your original response wasn't helpful, it was smug shitposting.

[Blacksheep]

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

What are you hoping to gain from a faster Cyclone 5?
Can you de-solder and solder 484-pins FineBGA?
What are you going to do with the removed ~$130 chip?

[marqs]

If there's a number of people willing to pay $100-200 extra for a beefier FPGA, we could consider making a handful of boards with 5CEFA7F23C6N which has significantly more logic resources and ~15% extra performance. Recompiling existing fw on that would be trivial, but maintaining a more featured fw (with sobel edge interpolation etc.) alongside the standard one would require more work.

[anexanhume]

If there's a number of people willing to pay $100-200 extra for a beefier FPGA, we could consider making a handful of boards with 5CEFA7F23C6N which has significantly more logic resources and ~15% extra performance. Recompiling existing fw on that would be trivial, but maintaining a more featured fw (with sobel edge interpolation etc.) alongside the standard one would require more work.

An OSSC Pro Pro of Numerous Things

[ldeveraux]

I don't assume anybody knows anything anymore, else why would they ask a question. Sorry I tried to help with what I thought was a basic question.

Your original response wasn't helpful, it was smug shitposting.

And then I followed up with something that clarified my initial response. Taken out of context it may sound like that to you, I can't control what feeling my words emote when you read them. Try reading it differently next time. That's smug shitposting. This is a terribly unfun cycle, thanks for maintaining its momentum.

[fernan1234]

If there's a number of people willing to pay $100-200 extra for a beefier FPGA, we could consider making a handful of boards with 5CEFA7F23C6N which has significantly more logic resources and ~15% extra performance. Recompiling existing fw on that would be trivial, but maintaining a more featured fw (with sobel edge interpolation etc.) alongside the standard one would require more work.

What kind of numbers do you think would be needed to make this special run viable? Perhaps there can be a sign up list where the most avid video enthusiasts can enter their info, to see if enough will sign up to meet the needed numbers.

After having a clearer idea of what concrete benefits would or might eventually follow from such an upgrade, I'd bet that more than a few would sign up.

[Elrinth]

If I'd have to lash out even more money for a better hardware FPGA or peripherals to get the best result, then so be it

Here's what I hope for the OSSC Pro:
* High compatibility input (rgb-video variants, component, vga, s-video, composhite)
* High compatibility output (close to hdmi specifications)
* Sync stability of an analogue tv or as close as possible.
* Low latency
* Sharp pixels in output (good, easy to use settings in resolutions)
* Vibrant and true colors (current OSSC already does this IMO)
* Presets for all kinds of common consoles/computers to get the best possible 1080p or 4K image for the best common GAME hdtv LG CX (We can make these ourselves ofcourse, but it would've been awesome to have it all built-in working outta the box):
- Super GrafX (Component or RGB)
- PC-FX (S-Video)
- MSX (RGB)
- PS1 (RGB)
- PS2 (Component)
- Saturn (RGB)
- FMTowns (VGAport RGBH)
- X68000 (VGAport RGBH)
- PC9801 (VGAport RGBH)
- Virtual Boy (RGB or VGAport)
- N64 (RGB)
- SNES (RGB)
- NES (RGB)
- Analogue Mega SG + 32X & DAC (RGB or Component)

My current problems with the normal OSSC:
* Sync stability (On my analog tv, it's a stable image but on the OSSC I'm losing sync on some systems making it impossible if I'd like to record footage or stream video footage to the internet of these systems)
Well impossible is the wrong word, but losing sync makes it not very fun at all to look at.
* Either pretty small image less sharp or large clipped/culled image for some japanese computers such as X68000, PC98, FMTowns etc.

[bahamutfan64]

Yeah, the big question is the handling of systems that spit out different resolutions based on the game. Juggling different profiles for 320x240 vs 352x224 etc is no fun.

[DirkSwizzler]

If there's a number of people willing to pay $100-200 extra for a beefier FPGA, we could consider making a handful of boards with 5CEFA7F23C6N which has significantly more logic resources and ~15% extra performance. Recompiling existing fw on that would be trivial, but maintaining a more featured fw (with sobel edge interpolation etc.) alongside the standard one would require more work.

I'm interested in a beefy board or 2. I fully understand what you're saying w.r.t. a split firmware being less likely to actually materialize reliably (or at all). I just choose to believe that it will materialize in the long run.

[TrantaLocked_]

Will the OSSC Pro be able to do graphical transform on the picture like the DSC 301 HD or other upscalers? And what kind of lag is possible with decent upscaling, possibly near zero I'd imagine?

[Harrumph]

Will the OSSC Pro be able to do graphical transform on the picture like the DSC 301 HD or other upscalers? And what kind of lag is possible with decent upscaling, possibly near zero I'd imagine?

Specs are in the first post...:
viewtopic.php?p=1395651#p1395651

So the answer is obviously yes.
But graphical transform/rotation cannot be done with near zero lag.

[Gara]

If there's a number of people willing to pay $100-200 extra for a beefier FPGA, we could consider making a handful of boards with 5CEFA7F23C6N which has significantly more logic resources and ~15% extra performance. Recompiling existing fw on that would be trivial, but maintaining a more featured fw (with sobel edge interpolation etc.) alongside the standard one would require more work.

I would go for that. We're all in for the long haul. That could open some very interesting doors

[Danexmurder]

If there's a number of people willing to pay $100-200 extra for a beefier FPGA, we could consider making a handful of boards with 5CEFA7F23C6N which has significantly more logic resources and ~15% extra performance. Recompiling existing fw on that would be trivial, but maintaining a more featured fw (with sobel edge interpolation etc.) alongside the standard one would require more work.

Would using a beefier FPGA like this give the pro the potential to upscale to 4K? If so I'd happily shell out the extra dough. Deinterlacing isn't a big concern for me. 4K scaling is the number one feature I've been waiting for for years.

[fernan1234]

Would using a beefier FPGA like this give the pro the potential to upscale to 4K? If so I'd happily shell out the extra dough. Deinterlacing isn't a big concern for me. 4K scaling is the number one feature I've been waiting for for years.

4K is not happening. An FPGA capable of 4K upscaling is in a totally different price range. If money is not an issue, which would be a requirement anyway for a theoretical 4K OSSC, just buy an Extron DSC HD-HD 4K PLUS A, or maybe a Kramer VP-551X, and pair it with the Pro.

[anexanhume]

Would using a beefier FPGA like this give the pro the potential to upscale to 4K? If so I'd happily shell out the extra dough. Deinterlacing isn't a big concern for me. 4K scaling is the number one feature I've been waiting for for years.

4K is not happening. An FPGA capable of 4K upscaling is in a totally different price range. If money is not an issue, which would be a requirement anyway for a theoretical 4K OSSC, just buy an Extron DSC HD-HD 4K PLUS A, or maybe a Kramer VP-551X, and pair it with the Pro.

Seems to me the path forward would be to find a TV SoC that has acceptable scaling and can be sourced in small quantities. Alternatively, do proof of concept on an FPGA, then take that to an ASIC conversion design house and fabricate through a crowd-funding campaign.

[marqs]

What kind of numbers do you think would be needed to make this special run viable? Perhaps there can be a sign up list where the most avid video enthusiasts can enter their info, to see if enough will sign up to meet the needed numbers.

It'd still be the same PCB, just one different part. It's thus not so much about a minimum number, but rather the amount of premium those preferring a higher-grade fpga would need to pay. Obviously it'd get a bit cheaper the more people would sign up for that, but at the same time standard model would get more expensive as economies of scale would diminish on the basic fpga.

Will the OSSC Pro be able to do graphical transform on the picture like the DSC 301 HD or other upscalers? And what kind of lag is possible with decent upscaling, possibly near zero I'd imagine?

You might want to elaborate on which kind of transformations you mean, but basic zoom, pan, scaling, +-90deg rotate are on the cards. Anything more sophisticated makes you question whether it makes sense to do those on FPGA because it's very costly to do things on them that do not need to be custom-designed. Some might even argue that for scaling & deinterlacing it'd be better to use a cheap FPGA combined with a video processor ASIC (somewhat alike Framemeister) so that FPGA only handles simple things like scanlines. While that might make sense from price perspective, you lose a lot of the flexibility at that point and the project in itself would become much less interesting too.

Would using a beefier FPGA like this give the pro the potential to upscale to 4K? If so I'd happily shell out the extra dough. Deinterlacing isn't a big concern for me. 4K scaling is the number one feature I've been waiting for for years.

4K is not happening. An FPGA capable of 4K upscaling is in a totally different price range. If money is not an issue, which would be a requirement anyway for a theoretical 4K OSSC, just buy an Extron DSC HD-HD 4K PLUS A, or maybe a Kramer VP-551X, and pair it with the Pro.

It should be possible to do 4K on Cyclone 10 GX, but it'd be quite crippled since you'd most likely need to run the processing pipeline 4 pixels in parallel since it wouldn't run 2pix/clk @300MHz or 1pix/clk @600MHz (more pixels in parallel means more resources burned). A non-compromize 4K solution would require Arria 10 or similar chips which can cost 4-digit figures.

Alternatively, do proof of concept on an FPGA, then take that to an ASIC conversion design house and fabricate through a crowd-funding campaign.

ASIC design and fabrication (incl. verification, testing etc.) costs millions, especially if we're speaking about more modern nodes. It's not really a viable option for niche products.

[anexanhume]

Alternatively, do proof of concept on an FPGA, then take that to an ASIC conversion design house and fabricate through a crowd-funding campaign.

ASIC design and fabrication (incl. verification, testing etc.) costs millions, especially if we're speaking about more modern nodes. It's not really a viable option for niche products.

The kind I’m talking about is in the tens of thousands to low hundreds, but you’re restricted to older nodes with less stringent design rules, such as 250nm/180nm nodes. It’s also possible to use open source tools on those nodes, particularly if you’re just doing digital designs (which I presume we’re talking about a purely digital 1080P->2160P scaler). That assumes a lot more risk, which is why I proposed the FPGA -> ASIC conversion route.

[Einzelherz]

I don't assume anybody knows anything anymore, else why would they ask a question. Sorry I tried to help with what I thought was a basic question.

Your original response wasn't helpful, it was smug shitposting.

And then I followed up with something that clarified my initial response. Taken out of context it may sound like that to you, I can't control what feeling my words emote when you read them. Try reading it differently next time. That's smug shitposting. This is a terribly unfun cycle, thanks for maintaining its momentum.

Just popping in to say that 90% of your posts are confrontational and you should probably do some self reflecting, my dude.

[Josh128]

Cant we all get along guys? Life is too fucking short. Marqs is the man and is doing shit most of us here can only do in our dreams, so lets try to stay on topic and be respectful to each other.

[TrantaLocked_]

I'm thinking now with 4K TVs being the new norm, and the long development cycle of each new OSSC, maybe it would be right to have a more expensive 4K scaling version. Imagine five years down the line, a Pro 4K version would still be killing it on market share in the space. I don't think it's make or break but if it's possible to do, I'd bet at the very least content creators who make money on their content would be willing to pay $500-$600 for that device. Maybe in a couple years the right chip will be more affordable and a revision could be done then.

[Blacksheep]

I'm thinking now with 4K TVs being the new norm, and the long development cycle of each new OSSC, maybe it would be right to have a more expensive 4K scaling version. Imagine five years down the line, a Pro 4K version would still be killing it on market share in the space. I don't think it's make or break but if it's possible to do, I'd bet at the very least content creators who make money on their content would be willing to pay $500-$600 for that device. Maybe in a couple years the right chip will be more affordable and a revision could be done then.

First page:

Will this support higher resolutions/line doubling than 1080p/5x for 240p? With 4k TVs being standard it'd be really awesome to see if the hardware supports it. What about HDMI's optional QMS for really fast resolution switching (240p/480i switching).

*EDIT* I know nothing of this stuff but some Googling found https://www.renesas.com/us/en/www/doc/d ... l51002.pdf which lists max resolutions up to 1080p@60hz. This mean we're limited to 1080p still?

That datasheet is just for the video ADC, but yes, the HDMI transmitter chip is also limited to around 165MHz. Proper 4K processing is very expensive to implement on FPGA today, but the model is designed so that it'd be straightforward to make a 4K edition one day when the hardware (and required design tool) costs drop.

[anexanhume]

Has anyone evaluated the potential use of Efinix FPGAs for a low cost custom scaler? I realize the tools are a huge part of the battle, but they have comparable size in their Trion T120 FPGA to the Cyclone V for a quarter of the cost.

[ZellSF]

I don't think it's make or break but if it's possible to do, I'd bet at the very least content creators who make money on their content would be willing to pay $500-$600 for that device.

I don't see what content creators would want 4K for.

The difference between 1080p and 4K for 240p content is pretty small as is, with content creators you're also throwing video compression into that reducing the difference even more.

That's assuming the worst case scenario that this content creator doesn't have a capture setup that accepts 1440p, and doesn't have a setup that allows whatever scaling he wants to be done on the capture side.

If you want people to spend money on irrelevant things, go after bored people looking for entertainment, not people doing their jobs. The latter group will buy expensive stuff, sure, but not uncritically.

[TrantaLocked_]

I don't see what content creators would want 4K for.

The difference between 1080p and 4K for 240p content is pretty small as is, with content creators you're also throwing video compression into that reducing the difference even more.

That's assuming the worst case scenario that this content creator doesn't have a capture setup that accepts 1440p, and doesn't have a setup that allows whatever scaling he wants to be done on the capture side.

If you want people to spend money on irrelevant things, go after bored people looking for entertainment, not people doing their jobs. The latter group will buy expensive stuff, sure, but not uncritically.

It's just another way to have more control and/or peace of mind. It means if I'm streaming I can send a direct native 4k signal to my monitor without making my monitor do possibly inferior or laggy scaling, or have less lag or processing resources used on the computer side for upscaling in real time. I'm not saying the solution people already have are necessarily bad, but I'm sure there would be cases where having the OSSC upscale to 4K would benefit some people who create content. The main reason being that scaling on the OSSC Pro will be done in a proper way that isn't bullshit, that will often be better than the bullshit found in a lot of consumer electronics including monitors and capture cards.

[Unseen]

It's just another way

...to spread FUD:

do possibly inferior or laggy scaling

Weasel Words

or have less lag or processing resources used on the computer side for upscaling in real time

Baseless assertion. Nearest-Neighbour scaling is the cheapest scaling method in existence - if you computer cannot handle it in real time, it cannot handle 4K content at all.

I'm not saying the solution people already have are necessarily bad

No, but you are heavily implying it by using this wording.

but I'm sure there would be cases where having the OSSC upscale to 4K would benefit some people who create content.

Impressively vague, but in the end a very low-hanging-fruit way of constructing an argument. You take a very small subset of the target audience and imply that there might be unspecified benefits for at least one person within that subset. While this argument could easily become true in a mathematical sense, "maybe it will be better for one person" is not a useful product design guideline.

The main reason being that scaling on the OSSC Pro will be done in a proper way that isn't bullshit, that will often be better than the bullshit found in a lot of consumer electronics including monitors and capture cards.

A better solution to the capture card problem is to not let the capture card scale at all and do it in software instead.

[TrantaLocked_]

...to spread FUD:

The only one spreading FUD is the one saying that FUD is being spread. Imagine taking a harmless conversation and then accusing someone of spreading FUD.

Weasel Words

So describing real situations that actually exist is weasel words.

Baseless assertion. Nearest-Neighbour scaling is the cheapest scaling method in existence - if you computer cannot handle it in real time, it cannot handle 4K content at all.

There are other methods of scaling and it isn't just about your computer being able to handle it. Doing all of the scaling at once in a single part of the chain results in the best image quality with the least artifacts, especially if the type of scaling is mis-matched between the OSSC and computer. Also, me suggesting that simplifying a setup that may not necessarily require simplification does not mean that I am spreading FUD or wrong about the value I know people find in improving their equipment.

No, but you are heavily implying it by using this wording.

No, I'm not. You're the one who believes this and it isn't my problem.

Impressively vague, but in the end a very low-hanging-fruit way of constructing an argument. You take a very small subset of the target audience and imply that there might be unspecified benefits for at least one person within that subset. While this argument could easily become true in a mathematical sense, "maybe it will be better for one person" is not a useful product design guideline.

My comment was to be conservative as not to suggest that literally everyone making content would want to actually buy a 4K OSSC Pro. Mainly that if someone, especially new to the space setting up a new chain of devices, were choosing new equipment he would be way more likely to choose something that can do all of the scaling at one point early in the chain if he had the choice. The large choice in scaling types that would be implemented flawlessly by the OSSC Pro and perfectly match both his monitor and intended output resolution to Twitch and YouTube would be quite enticing in simplifying what s/he needs to have setup. Less hiccups, often better quality or lower lag especially to the monitor, simplified setup, peace of mind. Those reasons do not mean a setup that upscales later in the chain is bad or unusable or even close to it; I am not talking about how the existing configurations are needing to be saved by the OSSC Pro. Which is why I said not "make or break."

A better solution to the capture card problem is to not let the capture card scale at all and do it in software instead.

Everyone has their own solutions to their problems, don't they. It's like having more options gives more people with specific beliefs about their how they want their chain to work is a good thing.

[ZellSF]

You haven't really said how it's relevant to content creation though. It's not likely going to be relevant to your viewers if your image (that they can't see) is imperceptibly better, or that you have slightly more lag that you probably can't perceive either. The latter might matter if you're playing on a professional level and you haven't bought a low latency display, but that's an unlikely scenario.

I want 4K too, but saying there's any sizeable market in content creators doesn't strike me as realistic.

and perfectly match both his monitor and intended output resolution to Twitch and YouTube would be quite enticing in simplifying what s/he needs to have setup

Twitch recommends you do not stream in 4K.

[TrantaLocked_]

You haven't really said how it's relevant to content creation though. It's not likely going to be relevant to your viewers if your image (that they can't see) is imperceptibly better, or that you have slightly more lag that you probably can't perceive either. The latter might matter if you're playing on a professional level and you haven't bought a low latency display, but that's an unlikely scenario.

I want 4K too, but saying there's any sizeable market in content creators doesn't strike me as realistic.

and perfectly match both his monitor and intended output resolution to Twitch and YouTube would be quite enticing in simplifying what s/he needs to have setup

Twitch recommends you do not stream in 4K.

Streaming won't be stuck at 1080p forever, once AV1 kicks in 4K streaming on Twitch will become more of a thing and that will be within a couple years. But even for YouTube content, what I said about streamlining the process still applies despite the ability to upscale in software. Content creators also have more of a reason to spend ridiculous amounts of money on their equipment, whether for content quality or tax write offs, so like a theoretical Pro with 4K upscaling I actually think content creators would be more likely to pay $600 for that than a regular gamer. And there are people who if they were choosing between converters for retro gaming (including PS2/GC/Xbox), the "4K upscale" label could be beneficial for the marketing. I did mention about making a 4K version later when the chips for it get cheap enough as marqs said. That is 99% probably what is going to happen and that's fine because the Pro will still be in a class of its own anyway. I'm just saying if it's possible to do it now that I could see that being good for the Pro's popularity and reach long term even if purely as a marketing strategy, and even if it sells less units now, it will pick up steam as time goes on and as the price goes down. In no way am I intending to play down what the Pro already is, which is a beast of a project.

[ZellSF]

You still haven't mentioned any practical applications of 4K for content creators I can see. It will be a marketing advantage to everyone, sure, but for content creators especially I don't think you've highlighted anything worth caring about.

Streaming won't be stuck at 1080p forever, once AV1 kicks in 4K streaming on Twitch will become more of a thing and that will be within a couple years.

Twitch doesn't even recommend streaming in 1440p yet which is 1080p*1.33. 4K is 1080p*2. Video codec efficiency upgrades do help. They're not going to make up that difference. They also recommend max bitrate of 6000kbps for 1080p, which isn't actually adequate for 1080p to begin with.