RetroTINK 5x-Pro

[Josh128]

They are indeed! Similar to GPU shaders in MAME! Mike is really serious about making this a real shootout between his box and Marqs and Woozles! This is good stuff.

Slot Mask



PVM Mode



BVM Mode

[mikechi2]

Glad you like it. I'm still chasing the holy grail (to me personally) of an accurate slot mask, but it is doubtful that such a thing will be possible at sub 4k (or perhaps even 8k).

[Josh128]

Sure, there will be limitations at sub 4K, but I'll bet 1440p can come close to making it look authentic. Good thing the 5X can handle 1440p output. It'll be a good primer for what we can expect from the the RT 9X+ Pro.

[mikechi2]

Haha, to do a real 4k scaler with the proper horse power to do GPU shader effects would probably require a "real" company (employees, payroll, HR, accounting, legal) to support. I'm not sure if I'm up for that again. One round has already caused me enough gray hairs.

One thing that I truly love about all this is that it's still very much a "hobby garage" operation. So perhaps, I'll have to pass the torch on that one. But hey, that's what I said last year about even touching a FPGA.

[TooBeaucoup]

The 5X already looks so good where it's at. If scalers fell off the earth today and none were ever made again, I'd be happy enough at this point. Of course, I'll gladly take better ones if they come along 5-10 years down the line. LOL!

[Odolwa]

"Frame lock is automatically engaged for 480p, 576p, 720p and
1080i sources since they derive from modern equipment that
generally follow standard refresh rates."

The manual has this on page 8, but when playing 480p GameCube games with progressive scan and component cables, I can change Frame Lock to Triple Buffer. Why is that?

[mikechi2]

Someone at some point requested that 480p sync be switchable

[thebigcheese]

Those modes look way better than just basic scanlines! Pretty cool stuff. On the one hand, I'd still love to see a smoothing filter for N64 games; on the other hand, I could probably just use one of these new filters instead to overcome how awful N64 looks.

[orange808]

Glad you like it. I'm still chasing the holy grail (to me personally) of an accurate slot mask, but it is doubtful that such a thing will be possible at sub 4k (or perhaps even 8k).

What you have looks really great.

Thanks for pouring so much extra time into the firmware (and adding all these really cool niche features). I'm just guessing, but I think you may have smothered the next Framemeister in it's crib.

[Odolwa]

Someone at some point requested that 480p sync be switchable

Is there a difference between the two? I don't see a difference on 480p when switching between the two options.

[bobrocks95]

Someone at some point requested that 480p sync be switchable

Is there a difference between the two? I don't see a difference on 480p when switching between the two options.

Triple buffer has more lag but is generally more compatible with displays. Unlikely to be a problem for 480p sources. Though the manual you were just quoting from explains this...

[Odolwa]

I didn't notice a difference between the modes in 480p, but it does make a difference when switching from 240p to 480i.

https://youtu.be/rz3aVuJuvG4

[Odolwa]

Why did EposVox say that the Retrotink 5X cannot do 4:4:4 color subsampling in his review on YouTube?

https://youtu.be/bP9cXiWOGt4?t=1029

[bobrocks95]

You, uh, already pretty much asked that question 2 pages back and got an answer...

viewtopic.php?p=1463507#p1463507

[Odolwa]

Partially.
By default it does 4:2:2, but 4:4:4 was introduced for some of the optimal modes in the latest (v1.55) firmware :
https://www.retrotink.com/post/retrotin ... re-updates

You can read more about it here :
https://www.retrorgb.com/retrotink-5x-f ... -v155.html

It can do 4:4:4 with the latest firmware, so it's just a software thing, not a hardware limitation then? I noticed that 4:4:4 is implemented for 1080p (Over) for SNES 256 mode, but N64 320 mode doesn't have it. Does 480p from the GameCube get 4:4:4 by any chance?

[strayan]

Does 480p from the GameCube get 4:4:4 by any chance?

GameCube is not 4:4:4 to begin with.

[N64]

Messing around with emulators on a soft modded original Xbox, and I noticed something weird. The horizontal axis in 480p is kind of blurry, but the vertical axis is razor sharp. Any ideas what could be causing this? Here is the test pattern generated by the emulator (each black/white line should correspond to 1 pixel in 240p which is being 2X scaled in the emulator to the 480p resolution of the Xbox):

https://i.imgur.com/RV0FTlR.jpg

I can scroll through the horizontal scaling factor in the emulator while viewing this test pattern but every single value on the horizontal looks like this. Toggling 10x11 pixel aspect ratio in the emulator didn't fix it either. I then checked out some Xbox games running in native 480p and I can see evidence of it in some thin vertical lines or text in menus now that I know what to look for.

As far as h.sampling, I can switch back and forth between generic 16:9 and DTV 858 on the 5X but both look the same. I haven't noticed anything strange like this on the Wii/GC or PS2 in 480p over component, but maybe it's a quirk with the way the 5X handles the Xbox's video output in 480p? Something to do with a conversion between 720 and 640 width and pixel aspect ratio? Toggling the widescreen setting in the Xbox dashboard didn't appear to change anything with the emulators running in 480p. Does the Xbox have some kind of horizontal anti-aliasing filter applied by the video encoder similar to the N64?

Here is the same test pattern but running directly into the TV over component 480p, bypassing the 5X. The image is blurrier as a whole because the TV doesn't upscale from 480p nearly as well, but the horizontal axis still looks a little blurrier than the vertical axis leading me to believe this could be something the Xbox is doing.

https://i.imgur.com/jMo70i0.jpg

[mikechi2]

I think this is just the Xbox. The video output quality isn’t as clean as a ps2, unfortunately, which is a real shame.

[mikechi2]

Does 480p from the GameCube get 4:4:4 by any chance?

GameCube is not 4:4:4 to begin with.

Indeed and neither is Wii.

444 for 480p is possible if a manual phase adjustment was implemented. Haha. But I think it’d need to be tweaked every time you connect a source. The current implementation is pretty good though IMO. Especially with the DTV 858 auto phase.

[N64]

I think this is just the Xbox. The video output quality isn’t as clean as a ps2, unfortunately, which is a real shame.

I have HD Retrovision cables coming later this month for the Wii and Xbox though I don't expect that will have any affect on this. A real shame indeed, running XBMC Emustation makes it a fun and user friendly emulation box in addition to running Xbox games.

[Odolwa]

Does 480p from the GameCube get 4:4:4 by any chance?

GameCube is not 4:4:4 to begin with.

Indeed and neither is Wii.

444 for 480p is possible if a manual phase adjustment was implemented. Haha. But I think it’d need to be tweaked every time you connect a source. The current implementation is pretty good though IMO. Especially with the DTV 858 auto phase.

Is there a way to get 4:4:4 for the N64 or NES?

[vol.2]

Is there a way to get 4:4:4 for the N64 or NES?

Unless I am misunderstanding you, I think the issue is that you do not yet have a clear understanding of chroma sub-sampling. If you are using anything in the RGB color-space, there is not sub-sampling done at all.

Sorry if that's not what you are asking, but if it is, you should read this carefully: https://en.wikipedia.org/wiki/Chroma_subsampling

[anexanhume]

Has anyone tried 540p with a HDCRT? I’m curious if it’s genuinely 0 lag and treated like 1080i on Sony sets with the HDPT. I was going to wait for the Morph, but this may change my mind.

[vol.2]

Yes just read back in this thread over the past 2 pages

[Josh128]

Has anyone tried 540p with a HDCRT? I’m curious if it’s genuinely 0 lag and treated like 1080i on Sony sets with the HDPT. I was going to wait for the Morph, but this may change my mind.

viewtopic.php?f=6&t=67984&start=1095

[anexanhume]

Has anyone tried 540p with a HDCRT? I’m curious if it’s genuinely 0 lag and treated like 1080i on Sony sets with the HDPT. I was going to wait for the Morph, but this may change my mind.

viewtopic.php?f=6&t=67984&start=1095

Thanks, I didn’t go back far enough. Now I’m super excited for my 34XBR960 possibilities. Thanks for your testing!

[Josh128]

Sure thing, just know that its likely that you will have to convert the HDMI out of the 5X to component video to get it working. I suspect thats going to be the case with most if not all Sony HD CRTs. Best get one of those $20 Portta converters on the way so you'll have it when you get your 5X. Report back when you are able to test!

[vol.2]

Has anyone tried it with a DA4 chassis yet? I think the 34XBR960 is basically the same time period as the DA4 IIRC and shares the input section? Not 100% sure.

When experimenting with the HS510 and the Raspberry Pi, I had zero difficulty in outputting a 540p signal to the DVI input. I'm not sure why that would not work with the RT5X, but I also don't know what timings are actually being output to the set from the tink.

I guess if Mike were to chime in with the timings, I could compare them to what I got working with the Pi and see if there's anything really different going on.

[anexanhume]

Has anyone tried it with a DA4 chassis yet? I think the 34XBR960 is basically the same time period as the DA4 IIRC and shares the input section? Not 100% sure.

When experimenting with the HS510 and the Raspberry Pi, I had zero difficulty in outputting a 540p signal to the DVI input. I'm not sure why that would not work with the RT5X, but I also don't know what timings are actually being output to the set from the tink.

I guess if Mike were to chime in with the timings, I could compare them to what I got working with the Pi and see if there's anything really different going on.

I know that the HDMI input has less lag than component on the 910/955/960 in like for like resolutions. From looking at the schematics, that makes sense as the HDMI input is passed directly to the DAC and converted to tube level inputs from the amps there on. The syncs are also passed straight through from looking at the schematics.

The syncs appear to have a path through muxes to be passed straight through just as the digital path is, but I’m sure the tube forcing digital processing on an analog input precludes that from being functionally possible.

[Josh128]

I know that the HDMI input has less lag than component on the 910/955/960 in like for like resolutions. From looking at the schematics, that makes sense as the HDMI input is passed directly to the DAC and converted to tube level inputs from the amps there on. The syncs are also passed straight through from looking at the schematics.

The syncs appear to have a path through muxes to be passed straight through just as the digital path is, but I’m sure the tube forcing digital processing on an analog input precludes that from being functionally possible.

That is not true for the 36HS420. Ive tested with a Time Sleuth direct to HDMI compared to converted with a DAC to the YPrPb inputs. The input lag was identical.