RetroTINK 5x-Pro

[mikechi2]

The old RT2X devices used 4:2:2 sampling at the input source resolution, which results in a noticeable decrease in color fidelity. OTOH, the RT5X does have an internal 4:2:2 data path but at a much higher sampling resolution so that even the chroma samples have as much, and in some cases, more data points than standard 4:4:4 sampling at a given resolution.

The 4:2:2 data path lends itself to an efficient integration of the SDRAM frame buffer and MA deinterlacer. For 'generic' modes, this is not noticeable in 99% of cases (once I fixed some stupid chroma shift bugs in the release firmware). Optimal sampling that try to create "emulator sharp" pixels do indeed need 4:4:4 handling at the output side for the best results. The latest firmware uses a nice trick inspired by Extrems to accomplish this in some modes

The 4:4:4 datapath would make a substantial difference for higher resolution modern computer content (i.e. 1080p), but the RT5X was not designed for those use cases.

But of course, there's always a few people that go: zOMg four tOO toO, it suX without bothering to understand the nuances, lol.

[mikechi2]

Completed preliminary testing of the new 540p FW from Mike on my HS420. I'll post some pics later but for now I'll get to the gist of it. It works. I ran into a few issues but I believe all of them are related to my particular set (if someone else has a 36" HS420 I'd love for them to try it). I first tried the 540p mode on my 32" LG QHD monitor, and it worked great. Tried it via direct HDMI on my HS420 and no banana. This didnt surprise me though, I believe my set has some kind of issue on the HDMI port. It only accepts (if I recall) 480i, 480p,720p, and 1080i-- and the 480p is "jumpy" as if its about to lose sync or something. Going into the YPbPr ports and everything works, so thats what I did.

I had the Portta HDMI to component converter that Fudoh recommended so I fed that from the 5X and went into the set with component. Worked like a charm. Was able to adjust in the svc menu to get perfect screen filling goodness on both Sega Genesis and GCN. No visual anomalies, no issues going that route. Only thing I couldnt do is lag test with the TS because I had no way to chain it through the 5X to get the component out that my TV would accept, which is disappointing, but I can say that both GCN and Genesis felt absolutely lag free in frame locked mode. Im sure the obligatory 2ms-4ms of lag from the 5X was there, but lets be honest, thats essentially zero lag in real world applications.

Below are some pics of FZero GX and Sonic 2 that I took. The first pics are unscaled, ie how the 480p image fits in the 540p frame, the close ups are after having filled the screen through the service menu, and last of course are the full screen shots. Note that I actually still had a bit more stretching ability in the menu than what is shown here, but this is what I settled on during this quick testing. Looking forward to someone whose set will accept the HDMI signal, it pisses me off that mine doesnt. I bet theres something wrong with the HDMI port/circuitry on my set. Also, my set shut off on me several times, I dont think its ever done that before, but I used a different power strip to power it this time and Im sure its either an issue with that or the set needs some TLC. I will test more later.

Spoiler

Spoiler

Spoiler

Spoiler

Spoiler

Spoiler

Awesome, glad that worked!

Yes the aspect ratio is incorrect with TV's default settings since the 480p occupies a subset of the full 540p vertical height. The RT5X can also interpolate the 480p -> 540p to fill the frame, but drawing it without scaling and then adjusting the deflection controls on your TV is a better solution, imo (especially if you want to run the scanline generator).

[Josh128]

Awesome, glad that worked!

Yes the aspect ratio is incorrect with TV's default settings since the 480p occupies a subset of the full 540p vertical height. The RT5X can also interpolate the 480p -> 540p to fill the frame, but drawing it without scaling and then adjusting the deflection controls on your TV is a better solution, imo (especially if you want to run the scanline generator).

Yes, thanks so much for taking this seriously and working on it for the community Mike. I really think this feature will make at least the Sony HD CRTs a great option for both 240p and 480p sources when they were acceptable at best for 480p and downright dreadful for 240p previously. As I mentioned earlier the digital input of my set doesnt like the 540p signal but the analog in works great with it. Im very anxious for others with these sets to try their digital ports and see if its just an issue on my model, my particular set, or what not. My Time Sleuth doesnt support 540p, but it does support 240p and my set doesnt take that either when going straight into the HDMI port, it only takes (that I know of) 1080i, 720p, 480i, and a jumpy, weirdly scaled 480p.

This having to use the analog/component in has opened up the question of whether the HDPT (HD Pass Through) input lag elimination option that is known to work with a digital source, still works when using analog in. I wasnt able to test this initially but I have components coming (hopefully today) that will allow for it, and I just realized I can still do a photo test of the final component signal split between the HD CRT vs a small SD CRT running the 240p Suite to get the answer to that question, so hopefully I can get to that very soon!

[incrediblehark]

I got my 5X Pro in the mail today, way ahead of what I was expecting! I did some preliminary testing of the 540p mode on my 40XBR800. Only a couple of pics right now, and they aren't very good, but are mainly to show some of the quirks I noticed. I'll probably be up all night playing with this and trying as many systems as I can!

So first off I want to say the 540p mode displays on my particular set. This model tv has DVI, so I use transcoders for both this input and compnent. My results were exactly the same for both inputs tested, except I could see a scrambled image for unsupported resolutions on the component input wheras DVI simply had a blank screen. I assume this has to to with analog vs. digital.

When set to 540p, I was able to get full vertical without any service menu adjustments other than Vol. 2's recommended JUMP setting. If you set the 5x sampling to 16:9 generic, it will fill the horizontal as well and you'll get a full screen image as a result. These were the easiest adjustments to get a 540p image in a full screen. The downside is that you get uneven horizontal spacing using this method. Its close, but I noticed it while running the color bleed check in 240p test suite:



When sampling is set to 4:3 generic, the spacing is perfect:



However, I was unable to get a full stretch to my display using HSIZ and HPOS. I can probably use a few other adjustments in there and get it closer to full screen though. I'm wondering if some sort of 540p horizontal sampling could correct this, as it would make using this mode extremely easy for anyone, requiring only 1 SM toggle of JUMP. this would allow the HDCRT to be used for other sources as well without needing to stretch the screen constantly, at least for Sony HDCRTs. I know the sample size is too small right now to assume these are typical results, though. It could just be my tv.

As another note, the other sampling modes did not give me good results, so I stuck with generic.

Here are 2 more comparison shots, the first is with 16:9 generic, the 2nd is 4:3 and stretched as far as I could using only HSIZ and HPOS (maxes at 63). It is very close. Again I apologize for the poor quality:



You'll notice in these photos I do not have scanlines enabled. The reason is that when enabled, I got extreme flickering/strobing effect which made the scanlines unusable. This did not occur when set to 480p mode, however. I got nice scanlines with no flicker and a full screen, perfectly scaled image in 480p. Not sure why this was the case with my set and not Josh's, I'm wondering if it is actually outputting as 1080i as it states in the service menu? Would love to hear more thoughts on this. Unfortunately I was unable to capture this in pictures or the short video clip I recorded.

As for lag, unfortunately I do not have a Time Sleuth, so all I can go by is the 240p suite. I haven't connected 2 displays to compare and only went by the obviously unreliable manual lag test. In those tests I got about a half frame better result compared to 480p.

Some more thoughts/requests for the 5x pro:

The design is really nice, I like the ease of use and gives me a more compatible picture than the OSSC. I'm looking forward to testing on my 1080p lcd as well to see how nice it can look there, as I have never been able to do better than 3x with the OSSC. Right now I do like the OSSC's scanline options a little more, and its reverse LPF for my Super Famicom, but other than that so far I prefer the 5x.

I also got the remote overlay from RetroLabs and highly recommend it! Ordered mine ahead so I'd have it when I finally bought a 5x.

Don't know how possible it would be as a feature request, but would be nice to move the OSD or have it positioned closer to the center of the screen, especially with this 540p mode and setting overscan you lose a lot of the OSD. The same would be true of the lightgun border, if there could be a way to increase size to make it visible in this mode we could see use of the Sinden lightgun on HDCRTs some day. I don't own one but just thinking into the future. I'm sure Mike Chi has already thought of these as well, and its not like this 540p mode is the main focus of the 5x so I'm definitely not complaining.

I plan to post some more pics too sometime this week. Thanks again Mike for putting out a great product, as well as supporting your userbase and their crazy demands!

[vol.2]

Josh,

You can ignore the reply I sent you in PM. I think I sorted out what was going on in the schematic, and the component input can bypass the DCR, so long as the input signal is 1080i/540p.


I think this is what I was using for the PI: hdmi_timings=640 1 276 82 92 480 1 58 6 18 0 0 0 60 0 36787500 1

[Josh128]

You'll notice in these photos I do not have scanlines enabled. The reason is that when enabled, I got extreme flickering/strobing effect which made the scanlines unusable. This did not occur when set to 480p mode, however. I got nice scanlines with no flicker and a full screen, perfectly scaled image in 480p. Not sure why this was the case with my set and not Josh's, I'm wondering if it is actually outputting as 1080i as it states in the service menu? Would love to hear more thoughts on this. Unfortunately I was unable to capture this in pictures or the short video clip I recorded.

Interesting. This sounds like your set is displaying 1080i instead of 540p. There should be absolutely no jitter/flicker/line crawl in 540p, but you will get that in 1080i. And if for some reason one field is displaying the game lines and the other field is displaying the blank/scan lines (when you enable scanlines), it would make for a horrible, flickering/flashing image because one field is black and the other has spaced game lines. 480p should look great for you, but it will have a baseline of 14-16 ms of input lag + the 4 from the RT5X. Only 540p/1080i allow for 0 lag with the HDPT enabled.

[thebigcheese]

IIRC even when working in 4:2:2 it works hard to align the subsampled channels so that they look very close to 4:4:4 anyway.

Is 4:4:4 even a thing on TVs, or is it more of a monitor thing?

I know that LG OLEDs, at least since the C9 and onward, will to 4:4:4 if you set it to PC mode. Beyond that, no idea.

[dojima]

444 exists on just about every modern TV, but you're not noticing it unless you're using it as a monitor, i.e., sitting very close to it. I find that letting the TV perform its normal processing will look better than forcing 444 at normal viewing distances.

[PearlJammzz]

Has anyone tested the 540p mode of GBI? I wonder if that'd help with some 540p signal settings. I believe that mode was created specifically for these HDCRTs.

[Jenrai]

Got one the other day. Hoping for a shipping notification soon.

[mikechi2]

Got one the other day. Hoping for a shipping notification soon.

This batch was a bit larger than the first, so it may take a while to plow through all of them. Hopefully you weren't too far down the list.

[Jenrai]

Got one the other day. Hoping for a shipping notification soon.

This batch was a bit larger than the first, so it may take a while to plow through all of them. Hopefully you weren't too far down the list.

Managed to get my order in the first five minutes but I'm not really sure how many people managed to order within that timeframe. I did have issues with paypal at the start but I got a confirmation e-mail so I assume everything went fine.

[incrediblehark]

I noticed that when I switch to CRT Simulate for deinterlacing 480i it disables the scanlines for 240p content. Is this as intended? I saw this while playing Virtua Fighter 2 during resolution changes.

[Josh128]

I have some fantastic new test results to share. I received the second Portta DAC recently and thus was able to try out the Time Sleuth via the RT5X Pro 540p mode. Just as I suspected, the component input on the HS-420 is just as fast as the HDMI input!! This is great news for anyone (like myself) that has trouble using the digital input of their Sony HD CRT. All sub-HD modes yielded 3.25 ms or less of input lag when using the 5X in framelock mode. 480p was the fastest at ~1.6ms. Note that I used the "bob" deinterlacing method on the 5X for this test. I'd also like to note that even in framelocked mode, switching from 240p to 480i and back was seamless and instantaneous, but going to 480p did require a very brief re-sync.

This should put to rest the idea out there that the component input is slower than the digital input on these sets. Thanks again Mike. If you decide to include this resolution in a publicly available release, I suggest to consider including a 1080i mode as well for enhanced compatibility for those who wish to use their digital inputs and have a set with a picky digital in such as mine. Mission accomplished, this new FW just made a lot of previously near useless Sony HD CRTs awesome sets for 240p/480i/480p gaming!

[vol.2]

Awesome Josh. Thanks for the time test!

Also, sorry for the misinformation about the component input. I finally remembered that someone had told me in a different thread that the component input wouldn't get bypassed with HDPT turned on, and I went through the schematics and figured out that they were wrong, and then I forgot about all that and only remembered what the person had told me! Well, at least I set myself straight rather than blundered forward until corrected.

IAC, if anyone wants to see this, BOTH the DVI/HDMI (1,2,3) input and the YPbPr input (41,42,43) have a secondary wiring hit that goes directly to the Component IF chip, and the the VDO outputs (25,26,27) bypass the DRC chip and go to the DAC board which is directly connected to the RGB drivers.

Spoiler

I believe the same is true for any SONY HDCRT that is 4:3.

[Josh128]

Hey no worries, Ive also read similar info on Reddit. There may be sets out there from different manufacturers that do vary input lag from analog to digital for all I know. What I do know is that this HS-420 doesnt!

[BazookaBen]

This is great news! Almost makes me want to find another HS or SFP set.

Question about the 480i mode though, is it line doubled to 960i inside of 1080i, or is it deinterlaced into 480 inside of 540p? I think 960i should look better.

As far as being able to stretch the raster enough, I don't know how the buffer works on the retrotink, but if you could decrease the size of front and back porches, you would have an easier time getting the raster to reach the edges of the screen

You'll notice in these photos I do not have scanlines enabled. The reason is that when enabled, I got extreme flickering/strobing effect which made the scanlines unusable. This did not occur when set to 480p mode, however. I got nice scanlines with no flicker and a full screen, perfectly scaled image in 480p. Not sure why this was the case with my set and not Josh's, I'm wondering if it is actually outputting as 1080i as it states in the service menu?

I actually remember seeing in the scanline'd sections of Axiom Verge when I played it on my 30HS420. I never thought about it much beyond that because I didn't use that tube for emulation.

But, is there something like a "v-hold" setting in the service menu? Making minor adjustments to that should change where the scanlines land, and could possible give you proper "double strike" 540p. I don't know how that would effect the picture when a 1080i signal is sent though.

[vol.2]

^^^^^

AFAIK, that mode doesn't exist yet. It's been requested, but I think it's just on a big list of possible future things. What's been asked for is line-doubled 480i to 960i inside of a 1080i frame. (or at least that what I and perhaps Josh were asking for).

[Josh128]

This is great news! Almost makes me want to find another HS or SFP set.

Question about the 480i mode though, is it line doubled to 960i inside of 1080i, or is it deinterlaced into 480 inside of 540p? I think 960i should look better.

As far as being able to stretch the raster enough, I don't know how the buffer works on the retrotink, but if you could decrease the size of front and back porches, you would have an easier time getting the raster to reach the edges of the screen

When you set the 5X to 540p, the input resolution doesnt matter, the set always gets 540p. As far as what happens with different input resolutions, they basically look exactly like they would natively, even 480i (when using "bob" deinterlacing). Bob deinterlacing retains the "flicker" look that you get on an SDTV with a 480i source. So its essentially still shifting the 480i image up and down in the 540p frame if thats what you are asking. Because you are stretching the raster, it should look identical to 480i doubled to 960i.

As far as stretching the image, as Mike has said, you dont want to do it with the 5X. You want to integer scale into the 540p frame and then adjust the raster on the TV. This is the only way to get perfectly even scanlines with no shimmer or distortion when scrolling vertically or horizontally, and that is how this mode currently works. Additionally, do note that I had plenty enough room in the svc menu to stretch the raster vertically and horizontally, no problems there.

As far as the 960i resolution, there should be no need for it for these sets. The 540p mode + SVC menu can do whatever you need with 240p, 480i, and 480p. Its perfect.

[BazookaBen]

Oh ok, if bob-deinterlacing already looks in effect the same as 960i, then yeah, no need to send an interlaced signal.

With stretrching, I thought I read someone was having issues bring it to the edge of the screen, but hopefully everybody gets that worked out. Though on the horizontal axis, I think you technically could reduce the size of the porches, depending on how the line/frame buffer works. Reducing vertical porches would require using the full frame buffer I guess, and if you reduce them too much, you'd leave that 33 kHz window for zero lag.

Sorry if you guys already talked about it and I missed it

[Josh128]

It does. I may try to post a video showing 480i and how it looks later so you can see. Also, the 960i mode of this TV features some strange vertical compression/banding and also doesnt fall within the 33.75 KHz HDPT range either, so its definitely a no go anyway, IMO.

[Guspaz]

Why would you want to simulate 480i on the display, instead of using motion-adaptive deinterlacing to get proper 480p output on the TV using the 540p mode?

[BazookaBen]

Also, the 960i mode of this TV features some strange vertical compression/banding and also doesnt fall within the 33.75 KHz HDPT range either, so its definitely a no go anyway, IMO.

Oh, by 960i I meant 1080i with Retrotink, just "effectively" 960i since the rest would be overscanned.

Sending an actual 960i signal gets intepreted as 480p I'm pretty sure, and gets all kinds of fucked up. Unlike feeding a direct 480i signal and using the "interlace" DRC mode to effectively show 960i, which looks halfway decent in my memory, but has the ridiculous amount of lag.

And no need to get pictures, I understand what you're describing. And I imagine it would be really hard to get pictures that accurately convey what it looks like in person

Why would you want to simulate 480i on the display, instead of using motion-adaptive deinterlacing to get proper 480p output on the TV using the 540p mode?

I'm a purist and typically don't want software "guessing" about how to alter the image. I'd rather just have the raw lines, for all their faults.

But then again, I guess it would be similar to what VR setups do for head movement when outputting 60fps content on a 120hz headset? That actually looks pretty good on my PSVR

[vol.2]

Why would you want to simulate 480i on the display, instead of using motion-adaptive deinterlacing to get proper 480p output on the TV using the 540p mode?

I don't want to play Final Fantasy XII deinterlaced. I think that would be weird. I'm so used to games like that interlaced, and I like the way it looks.

[Guspaz]

But then again, I guess it would be similar to what VR setups do for head movement when outputting 60fps content on a 120hz headset? That actually looks pretty good on my PSVR

Hmm, not quite. A motion-adaptive deinterlacer will try to use weave wherever possible (since weave deinterlacing with no motion re-creates the perfect 480p image), and uses an alternate technique like bob whenever there is motion (since motion causes artifacting with weave). It's not doing any reprojection like Oculus's Asynchronous Space Warp does.

Maybe to make it easier to understand, I'll explain the concept with a very simplified system (it'd be a lot more complex in practice). For any given scanline, if it's more similar to the previous scanline in the same field, use bob. If it's more similar to the previous scanline in the previous field, use weave. The idea is to try to get the best of both worlds for every part of the image.

I don't want to play Final Fantasy XII deinterlaced. I think that would be weird. I'm so used to games like that interlaced, and I like the way it looks.

The PS4/XB1/Switch/PC versions of the game must look very odd to you then

Does the HD CRT TVs do forced deinterlacing on 480i content? As in, could you not just connect the PS2 directly to the TV and let it display the 480i? I wouldn't think it was the kind of game where the bit of extra lag would be a problem. I don't think full-time bob de-interlacing will get you what you want, in my opinion it doesn't look that much like 480i on a 15khz display. But maybe it's close enough for you?

[Josh128]

Most HD CRTs automagically de-interlace 480i, though some have options to present it in "bob-like" 1080i which gives it a semi-authentic look. My old Hitachi used to be able to do that. This set, I havent found the option yet, if it exists. I agree with BazookaBen and Vol 2 though, I also prefer 480i games to present as interlaced (ie authentic) vs any kind of deinterlacing that makes them appear progressive.

Getting back to Ben and Vol 2's questions-- theres actually several de-interlacing modes in the 5X that give the effect he's looking for-- I was using "bob", but theres also "linear", and "CRT Simulate". I just went back and looked at all three, and while "bob" and "linear" both have the signature flicker of 480i, "CRT Simulate" goes a step further and weaves blank scanlines in each alternating frame, at the cost of some brightness, but IMO it looks EXACTLY like native 480i content on an SDTV. Very impressive, and the brightness loss is really not bad either, but it can be easily brightened up by the set if you so wish. I was going to video it to show it, and might still do so.

Also, happy to report that running my TV direct to power instead of using a surge protector, it did not shut down even a single time during this testing. Interesting.

[vol.2]

bascially, just as Josh says. when a game is really well optimized for 480i, it looks distinctive and sharp. i'm just so used to the effect that i love it, and this is primarily on games i've played many times and got used to that way. even on many games on ps2, but especially on Wii look much sharper in 480i than they do in 480p

[fernan1234]

Glad to hear more voices singing the praises that interlaced video deserves! I've been evangelizing for 480i for a while now, but it's been hard to go against the pro-480p conventional wisdom.

BTW, it looks like the 5X has already done a fair bit to bring down the prices of the Framemeister as some of us predicted . It should only go even further down from here.

[Josh128]

Yeah, for me, if I grew up playing a game in 480i, thats the look I like when playing today. My first 480p system was GC, and I got the cables right when they released, and I loved how crisp and sharp it was on the Hitachi Ultravision 36SDX88B. The fact that the 5X allows for both of these on the same set now with zero lag is just amazing.

[fernan1234]

Yeah, for me, if I grew up playing a game in 480i, thats the look I like to when playing today. My first 480p system was GC, and I got the cables right when they released, and I loved how crisp and sharp it was on the Hitachi Ultravision 36SDX88B. The fact that the 5X allows for both of these on the same set now with zero lag is just amazing.

The OSSC has actually been able to do this as well all this time by using 100% scanlines and alternating scanlines for 480i sources, but I think very few people used it. The GBS-C does it by just using bob mode instead of adaptive deinterlacing too. But the 5X also makes it simpler. Oh, some of the 2X models did as well.

Since the 5X's CRT simulate is just bob+scanlines, it should also have the advantage of being faster than the AD mode that most people use.