RetroTink 4K All But Officially Announced ....

[Josh128]

Looks like Mike has the good parts figured out and its only a matter of building the supporting video hardware around them. This is going to be all about next-level CRT mask emulation effects. With this hardware capability and Mikes stellar FW updates, ts truly going to be the ultimate scaler.



https://twitter.com/retrotink2/status/1 ... 6539465729

[fernan1234]

What a God among men

[subcons]

Badass. FPGA has been such a game changer for retro gaming.

[Ed Oscuro]

agh, why did he have to go with the clown emoji array...almost a deal-killer! I will have to replace it with something a bit more palatable like a wojak or a rare pepe

in all seriousness, though, good to see the parts shortage is finally lightening to the point that this may be possible (i.e. Gartner's reports) to deliver in a reasonable timeframe.

[strayan]

Hopefully by the time this thing is released we’ll have have OLEDs with much higher sustained brightness to compensate for the scanline masks.

[ZellSF]

Probably be >$200 though it's harder to say.
4K will ONLY output 4K.
5X will have support for all the other resolutions

(200$ more than the Retrotink5X)

I want sharp scaling (lanczos), anime scaling (Anime4K) and post-processing AA (SMAA). For a scaler that can take 720p/1080p input and scale them to 4K, those are essential IMO, they can do so much to improve image quality for games that use those resolutions.

I'm only expecting the 4K output to be used for CRT filters though, which would be a bit of a waste.

[Josh128]

^^

Ehh, Mike's shown he can be a pretty creative guy, aside from being a genius engineer. Im sure he will find some creative uses for the additional resolution.

Im not sure the reason he is saying the 4K will not output lower resolutions though. I'd hope that he would reconsider that, but it may be due to financial/business and/or unnecessary replication of work decisions. Either way, Mikes a smart guy and has his reasons--- Im 100% behind whatever he decides to do if its best for him.

[PearlJammzz]

Well, I am all about just one scaler for 4k if he can somehow make it better or cheaper that way. Maybe locking to 4k is how he is able to provide 4k without having a 1000 dollar FPGA attached?

Luckily the RT5X isn't going away and will be plenty for a lot of people. a 4k scaler is pretty niche as far as scalers go and it'll likely be 500+ dollars making it even more niche. I am sure if there is more than a static 4k output on the table Mike will eventually take advantage of that like he has for the RT5X.

[Guspaz]

Supporting so many different output resolutions has been a real headache on Mike's part, and takes up a lot of space on the hardware, so limiting it to 4K is the way to streamline things. This product will cost a lot more than the 5X in order to do 4K, so it only makes sense if you will use it on a 4K display. If you need lower resolution output, that's what the 5X is for. He plans to continue supporting and updating the 5X since it will be the more mainstream product.

[mikechi2]

^^

Ehh, Mike's shown he can be a pretty creative guy, aside from being a genius engineer. Im sure he will find some creative uses for the additional resolution.

Im not sure the reason he is saying the 4K will not output lower resolutions though. I'd hope that he would reconsider that, but it may be due to financial/business and/or unnecessary replication of work decisions. Either way, Mikes a smart guy and has his reasons--- Im 100% behind whatever he decides to do if its best for him.

Appreciate that man. Yeah the main limiting factor as Guspaz says is time and headache. That's the one lesson I learned from the 5X. It makes no sense (for one dev) to get spread thin trying to accommodate a 100 different obscure use cases and still get nitpicked about them. I'd much rather do one main thing (at least initially) and do it extremely well. Other goodies can and probably will come later down the road, but I don't want to promise a bunch of different things and then not deliver on launch

This is especially true since the pipeline needs some optimizations specifically for 4K, which may or may not translate well to lower resolutions (doesn't mean the hardware can't do any arbitrary resolution, it just again more time and testing).

I want sharp scaling (lanczos), anime scaling (Anime4K) and post-processing AA (SMAA). For a scaler that can take 720p/1080p input and scale them to 4K, those are essential IMO, they can do so much to improve image quality for games that use those resolutions.

I'm only expecting the 4K output to be used for CRT filters though, which would be a bit of a waste.

Realistically, the initial feature set is around retro games and CRT filters. Again, focus on one thing first haha. I'm not sure Anime4K is practical on a real-time FPGA, but other stuff can and will be looked at later down the road.

[Unseen]

This is especially true since the pipeline needs some optimizations specifically for 4K, which may or may not translate well to lower resolutions (doesn't mean the hardware can't do any arbitrary resolution, it just again more time and testing).

There is always the option to boot separate bitstreams for 4K and <4K output, if you have sized your flash generously enough.

[XtraSmiley]

For the love of god, I hope he is able to incorporate low-lag rotation as a feature for this one. I don't mind it costing $$$ more, but it HAS to have this feature for us!

[Unseen]

For the love of god, I hope he is able to incorporate low-lag rotation as a feature for this one.

Rotation requires at least one frame of lag unless you have some way of looking into the future: For a 90 degree clockwise rotation, the first pixel of the first line in the output is the first pixel of the last line of the input, so you cannot start the frame on the output until the frame on the input is basically complete.

[orange808]

For the love of god, I hope he is able to incorporate low-lag rotation as a feature for this one.

Rotation requires at least one frame of lag unless you have some way of looking into the future: For a 90 degree clockwise rotation, the first pixel of the first line in the output is the first pixel of the last line of the input, so you cannot start the frame on the output until the frame on the input is basically complete.

About tate rotation:

There is no affordable video processor solution anywhere on the market that performs the task of rotation, scaling, and padding with only one frame of complete latency. Warp processors are difficult to operate and many have troubles with irregular signals. That's unacceptable, because coin op game PCB's are a major use case for tate mode.

So, one frame of latency would be the definition of low lag. Ease of use, compatibility, affordability, and availability would be nice as well. Used pro video gear can be difficult to find--and potentially expensive.

If I understand correctly, OSSC Pro (or Morph for people that want to throw out all their expensive gear and start over buying a new expensive switch setup) will probably be the machine(s) for people that want to use tate mode.

[Josh128]

^^

But as pointed out above, even a theoretical "best" implementation is going to be a minimum of +16ms lag, so even these newer machines can not get around that. With rotatable LCDs/OLEDs these days, the bona-fide superior solution is to rotate the display, therefore bypassing a $400+ scaler and at least one frame of lag.

[ZellSF]

For a lot of people, rotating their displays isn't viable.

Definitely hoping we will have a scaler that does it well at some point.

[Ed Oscuro]

It's fascinating to me that flat panels and cheap VESA mounts from Amazon make rotating setups lighter and more compact than they ever have been, but ask an expensive new scaler replace heavy stacks of ancient, out-of-production legacy video processors and some folks will say that's wasteful of resources. But hey, I won't say that it would be bad to add a rotation feature to a scaler. It would certainly be nice to have.

[ZellSF]

It's fascinating to me that flat panels and cheap VESA mounts from Amazon make rotating setups lighter and more compact than they ever have been

Sure rotating a light display can be done for cheap (that said; Amazon is not that cheap worldwide), but rotating a 50"+ TV if you don't have the option to wall mount? Probably not that cheap, or easy.

Plus you know, a giant PITA to rotate the screen whenever you want to play a certain game. People here buy automatic switches so they don't have to do simple cable switches. I'm guessing automatic solutions are much farther from cheap and easy.

Hell on my monitor that I can rotate I spend like 10 minutes getting it perfect again after touching it for any reason.

Not to mention the cable management you might have to do to accommodate rotating the screen in both directions. If you don't care how your setup looks this might be easy, but if you do...

I wouldn't be surprised if anyone wants to buy an expensive scaler not to have to deal with any of that. And 16ms of input lag is nothing if your display is lag free in the first place. Only reason I care about input lag as much as I do is because my display has 42ms of input lag to start from. You start adding to that and things get bad quickly.

[fernan1234]

At least counterclockwise rotation should be faster too, many games support it either by default or via dip switch.

All these years I've hoped for a fast scaler than can scale 240p tate games into a 480i output for 15khz CRTs, or higher scan rates for PC or multiformat CRT monitors. With CRTs even with the additional delay of clockwise rotation everything should feel pretty responsive still.

[mikechi2]

At least counterclockwise rotation should be faster too, many games support it either by default or via dip switch.

All these years I've hoped for a fast scaler than can scale 240p tate games into a 480i output for 15khz CRTs, or higher scan rates for PC or multiformat CRT monitors. With CRTs even with the additional delay of clockwise rotation everything should feel pretty responsive still.

A big challenge with rotation is inefficient memory access compared to regular readout. I'm not going to say rotation is off the table. There's probably enough memory bandwidth to do rotation for 240p sources, but it probably won't be slated for the initial release. Again, don't want to overpromise and under-deliver

[XtraSmiley]

If OSSC Pro can do it, I have hopes for Mike's 4K device!

I have CRTs, I have arcade cabinets, I have an LCD I can rotate. That's not the point. I'd like this feature as well, everything has it's use!

[mikechi2]

If OSSC Pro can do it, I have hopes for Mike's 4K device!

I have CRTs, I have arcade cabinets, I have an LCD I can rotate. That's not the point. I'd like this feature as well, everything has it's use!

Hehe, damn, I was going to suggest selling you a Tink branded rotatable VESA mount

[XtraSmiley]

If OSSC Pro can do it, I have hopes for Mike's 4K device!

I have CRTs, I have arcade cabinets, I have an LCD I can rotate. That's not the point. I'd like this feature as well, everything has it's use!

Hehe, damn, I was going to suggest selling you a Tink branded rotatable VESA mount

Who's saying I wouldn't buy it...

[neorichieb1971]

Whilst I love these products, it does seem a bit "Nintendoish" with the baby steps.

I can go broke just buying these hardware upgrades and cables. With analogue/retrotink my SACD and game collections and Japan trips on hold, I'm pretty much going into the dark side of credit

[orange808]

^^

But as pointed out above, even a theoretical "best" implementation is going to be a minimum of +16ms lag, so even these newer machines can not get around that. With rotatable LCDs/OLEDs these days, the bona-fide superior solution is to rotate the display, therefore bypassing a $400+ scaler and at least one frame of lag.

Sure. That's obvious. Everyone knows that.

Here's the thing: with a zero lag sample and hold display that refreshes instantly, I get the frame on screen without display lag or raster/scanout latency. In this case, that means one total frame of latency from video processing and the display. The display still finishes "drawing" each frame at the same time a CRT would, because the "scanout" takes microseconds instead of ~17ms (faster than electron gun scanout).

So, it's not the absolute deal breaker you're presenting. Depends on your display configuration and the trade offs you're willing to accept. Consumer options for extremely low lag displays are extremely limited, however. It's a pain in the ass to buy a specific panel and install a DIY replacement board.

There are things that can (and will) be done to remedy that. At some point, a frame buffered tate mode option will be practical enough without latency worries.

---------------

Honestly, the best solution to lower lag and increase signal compatibility on consumer displays is adding VRR output from a video processor. On an LG OLED, that knocks latency down to ~6ms. You give up BFI, but it's worth it if you have a situation where you need lowest lag or your source has an odd refresh rate. Unfortunately, getting that implemented is easier said than done. Cost concerns aside, I get the impression that small developers can't get their hands the resources/tools to develop VRR output. Seems hobby devs are locked out.

Anyhow, gamers should want and value "VRR" output (triggering VRR on the display) more than 4k in my opinion, but I am probably an outlier. I dislike using VRR when the signal isn't "variable", but it would be a good option to have when needed.

Personally, I could care less about 4k. An HDFury with a Darbee in the chain will scale up 1080p into a very perceptually sharp image with no lag. ¯\_(ツ)_/¯ But, I have one or two video machines that others may not.

[mikechi2]

Whilst I love these products, it does seem a bit "Nintendoish" with the baby steps.

I can go broke just buying these hardware upgrades and cables. With analogue/retrotink my SACD and game collections and Japan trips on hold, I'm pretty much going into the dark side of credit

Heh, I only wish the jump to 4K was a "baby step". Looking at it now, I would say the jump for 1080p to 4K is significantly more than 480p to 1080p.

In seriousness, realistically you have to crawl before you walk, and walk before you run. I, and every other dev I'm sure, are grateful for the community's support for making each only these steps possible.

[ZellSF]

Anyhow, gamers should want and value "VRR" output (triggering VRR on the display) more than 4k in my opinion

Because it has lower lag on one line of TV screens most people don't have?

[orange808]

Anyhow, gamers should want and value "VRR" output (triggering VRR on the display) more than 4k in my opinion

Because it has lower lag on one line of TV screens most people don't have?

I like turtles.

[bobrocks95]

Whilst I love these products, it does seem a bit "Nintendoish" with the baby steps.

I can go broke just buying these hardware upgrades and cables. With analogue/retrotink my SACD and game collections and Japan trips on hold, I'm pretty much going into the dark side of credit

Heh, I only wish the jump to 4K was a "baby step". Looking at it now, I would say the jump for 1080p to 4K is significantly more than 480p to 1080p.

In seriousness, realistically you have to crawl before you walk, and walk before you run. I, and every other dev I'm sure, are grateful for the community's support for making each only these steps possible.

Tons of things in tech where you can either buy now or wait for the next thing that's always just around the corner. Continual improvement is expected and really appreciated. It's not like the 5X suddenly becomes obsolete and worthless the second a 4K version comes out either- a Framemeister still sells for $300-400 and it's completely obsolete at this point.

Anyhow, gamers should want and value "VRR" output (triggering VRR on the display) more than 4k in my opinion

Because it has lower lag on one line of TV screens most people don't have?

Plenty of brands support VRR at this point. If you're buying a new TV for gaming and it doesn't support it, that's on you by now.

[ZellSF]

Anyhow, gamers should want and value "VRR" output (triggering VRR on the display) more than 4k in my opinion

Because it has lower lag on one line of TV screens most people don't have?

Plenty of brands support VRR at this point. If you're buying a new TV for gaming and it doesn't support it, that's on you by now.

Probably should've quoted the entire post so people got the context, but I thought it wasn't necessary in a so short topic.

Lag was part of the conversation, the implication was that gamers should care about VRR because VRR modes are lower lag, except VRR isn't inherently lower lag. That's just specific to one line of TVs, not all TVs with VRR support.

Gamers should care about VRR, just not for fixed refresh rate content.