shmups.system11.org

Oh, that looks interesting.

So yeah, it's best to drop the Vcc connection. We will power the ESP using its own regulator.
All GBS boards ship with a nice 2 pin power lead that can be soldered onto the ESP board on one end, and plugged into the "5V IN" port on the GBS.
It's still portable this way: Simply unplug the power lead and you can take the ESP adapter off

That said, I also recommend using all available ground pins to mate the ESP board with the GBS.
There's 4x GND on a NodeMCU, for example (I hope the other boards with that physical pin arrangements also have those as GND).
I'd have to check again, but I think the GBS has only one ground pin on the PLCC, right? That's too bad really, it should've been more ><

Using this setup, the ESP8266 is mostly transparent to the GBS power design.
Any signal pins are either inputs or open drain outputs (the GBS board pulls them up to 3.3V), ground is shared via the power cable and several board points.
Vcc from neither board has a defined path to flow and potentially try to equalize each other (only some negligible parasitics).

Just checked, the PLCC chip has 3 ground pins. I've connected them all to different ground points on my NodeMCU

Okay, here's the recompiled video. I, of course, had to grab some additional footage to round out the video, and got some PS2 via component input with 240p/480i switching. And then grabbed some straight up PS2 component 480i -> 960p via GBS. It looks very good for being deinterlaced. Artifacting is minimal, and this is absolutely a fast deinterlacer. IMHO, given the quality and no noticeable lag when playing, this definitely beats the DVDO VP30 on all fronts - for progressive and interlaced sources.

I also failed to edit out a PS2 reboot that resulted from it not detecting my HD Advance disk on the first try. The signal stays locked across the entire chain through the reboot, which is great.

Anyhoo, enjoy! View in 1440p.

https://www.youtube.com/watch?v=HIaX2WVpiX0&t=179s

:edit: updated video, with improved RE2 n64 clips and additional n64 resolution switching footage, uploaded 1/12/19: https://youtu.be/pAYj-13mJQg

First thing I noticed, just from your Resident Evil 2 footage. Your brightness is way too low and your contrast is way too high

Contrast being too high is a function of the elgato camlink. There are limited adjustments in its UI, when capturing via OBS. But, I messed with all of it for quite a bit on all consoles, mostly via the OSSC. Saturn looks good. PS1 looks good. N64 I couldn't get dialed in, just mostly went for getting black to be black. PS2 is all over the place, one title to the next. I looked at other samples of ICO footage on youtube, and it is all overly bright in the 480i footage. And, this is all after using the GBS auto gain feature.

At any rate, the purpose was to capture the transitional effects, and secondarily, how does it handle the deinterlacing.

True, and it does show them pretty well. Personally I'd get rid of that elgato though

As soon as I have $300 to drop on the next tier up...unfortunately probably not in the near future.

That's a nice show off, looks like it took a while to create, too!

The brightness and contrast issues are a pity.
So far, yours are the "reference videos" for gbscontrol, and they'd be cool to show off the capabilities.
Of course, people will take a look and immediately notice the problems, and they'll be chalked up to the scaler :/


This all sounds to me like your Elgato is expecting or producing limited RGB at some point.
To get black to be black with limited RGB, you'll have to drown out a lot of range, and this looks to be the case here.
Of course it's far too dark then overall, so you'll be pushing the contrast up to compensate :p

If I knew where exactly this goes wrong, maybe I could think of something..
I just haven't dealt with limited range RGB yet.
Is it something I'm expected to output into such a device? Is there anything to do about it at all? (Who invented this crap, and what's their address? :p)

Awesome to see ICO in the mix. Did you play this title?
I found, bought and finished it on PS2 (PCSX2 ;p) back in 2014 or so. Was a sleeper hit, I suppose

Edit: The game's FMV really appear to be mastered for limited range. Output is from PCSX2 with software (correct) rendering

Many ingame scenes also suggest they're meant to be limited range. Hmm..

Couple more things:
The hostname for the device will be gbscontrol.local from now on. If you have bookmarks, you need to change the url.
This change makes it so that both IP lookup services in gbscontrol answer to the same address (MDNS and the ESP8266 DNS server).
Effectively, this increases chances that the device will be found when in station mode.
It's still not perfect, mostly because MDNS isn't supported out of the box in many operating systems.

I unified the deinterlacer section and removed the duplicate data from the presets.
It makes it a little easier to support the various options and presets from now on.
If you have saved customized presets, they are invalid now though.
Sorry for the inconvenience, but it helps me manage the clutter

Finally, I think the 240p/480i transitions can be better still.
I found 2 more ways to reduce the glitches

Excellent stuff! Don't worry about invalidating custom presets, this is all beta stuff after all. I'm just excited to see how far this inexpensive scaler can really be pushed

Good info and explanations. The camlink is a good cost-effective capture device which doesn't have any issues with what the OSSC and GBS outputs, although it will in no way accept what the VP30 outputs. The whole image is diagonal and skewed. I went round and round with elgato support for months until they finally said "sorry we can't support all applications. But, obviously the cost-effectiveness comes with some trade-offs.

I've got an old Avermedia C127 as well but it only does 60fps up to 720p, and doesn't like things that are even slightly out of spec. I believe I tested it with the GBS CFW directly (C127 has a VGA input), and didn't have much success, but maybe it's worth retrying.

At any rate, sorry if I'm not giving the best possible representation. I do what I can with what I have or am able to scrape together.

No, I never played ICO. I grabbed it to check out PS2 240p via the OSSC initially, but here it is coming in handy for this test as well.

I updated to the latest CFW - move out the deinterlacer section from presets, this invalidates user… (UI - looks nice )

On my Megadrive (only system tested last night) I was getting some breakup, like loss of sync. Were there recommended resistors to add in the cable? I have had this problem before (along with Master System Mk2) when I was using a Sync Splitter on older CFW revisions but it seemed to stop after a while so I left it. I think it is my end rather than the CFW. Now I don't run a Sync Splitter (done cap change and 100ohm resistor between sync & ground). On the info box on GBS Control I saw 'SOG 4'.
I hope to do further tests tonight.

From a quick check my Megadrive SCART cable seems a pure connection from the Composite Video pin on the Megadrive.

One interesting this which I am sure was introduced around 3 CFW's ago, on my monitor selecting 1280x9?? says 'Out of Range' I was pretty sure it worked before. Not a problem as the other resolutions look really good (BenQ 912? 15KhZ compatible 5:4 ratio).

Thanks

Great. I measured the actual value of Vcc from GBS nearby SDA/SCL that was 3.18V, still slightly lower than appropriate voltage.
I'm considering that if I use just powering the ESP with usb port, than I can easily use and which should be through to GBS.
I didn't tried yet, but if I can use the usb power, the Vcc line will be no longer needed.




It seems that D1 pin layout is slightly different rather than mine, it cannot be able to share same PCB . Maybe I should to make separately .
Yes, this PCB is not just for me. It is lazy to make the several piece of gbscontrol for my friends who is owner of arcade game center.
If this PCB is guaranteed(?) without issue, I'm concerning share my kicad file.

NoAffinity:
I analysed the video some more, and man, that Do-Don-Pachi game sure kills the deinterlacer!
The game is on PS2, right? I'll try to get this. Nothing else I have shows this level of fail all at once.

Also, I think the scanlines filter adds too much contrast compression these days. I see it on my test display as well.
In some games it's not at all noticeable, but others show some kind of dull look.
I'll be improving this somehow.

Lastly, I mostly fixed the 480i transitions. This will be pushed soon

Higgy:
The MD remains my hardest to support console.
I really recommend you get CSync working on it, or use a sync stripper again.
This console stretches all the video specs to the limit with the very wide sync pulse, "as much as possible" active video
and chroma extending far below the black level on Composite Video.
The sync problems you get now are probably from the SOG detection choosing a different level than in earlier revisions.
As such, sending "wsog X" (X being 1 to 8 ) over the terminal can restore the old functionality. But yeah, get CSync

znpfg:
3.18V is a problem. Something isn't right there, or the measurement equipment isn't ideal.
You should be seeing at least 3.25V on the worst part of the GBS board, but normally around 3.31V.

If you're planning to install modded boards in arcade cabinets, I strongly suggest using the power supply method via VIN on the ESP board (as described earlier).
You say you want to use USB power though? If you have a reliable USB connection in those cabinets, that's great as well (basically the same method).

I'll wait till you push your latest changes to the deinterlacer and i'll put together a demonstration video of various different systems.

@rama I had an idea for a possible GitHub wikipage. It might be useful to have a page with a table that has every single hardware modification mentioned and what board it applies to, and whether it's necessary it not.

I thought it would be useful to see everything in one place - even mods that are bad/depreciated. I know I've personally had to read though this entire thread three or four times over the past few years just to reorient myself with the progress of the project.

It could be helpful to newcomers.

I made a fairly comprehensive table using standard GitHub markdown. If you're interested in using it let me know and I'll share.

@rama - yes, its dodonpahci on ps2

Sent from my SM-G955U using Tapatalk

Yes, please do! I've had to do the same multiple times, just to keep up with the newest hardware modifications. I've even asked Rama himself a couple of times which capacitors to replace and whatnot

Ok I put the page up in GitHub. It's incomplete so please add/correct mistakes.

Rama - I got my Megadrive sorted. I modified the PCB for CSync. I cut the Composite track to the AV plug and joined in CSync which was right next to it. My SCART cable was moulded at the Megadrive end so I could not change the wire there. My AV cable only has 7 wires so CSync was not connected.

I've just been playing Sonic and picture Sync is great.
I used a 470 ohm resistor and 10uF Cap in the SCART plug housing.

Update - I played between 15-20mins and just got 1 quick sync looking glitch and 1 vertical glitch that passed quickly left-to-right. So overall really good. Does 'Vertical Masking' work ? I could not get it to cover up the Megadrive areas where you get small amounts of flashing pixels, especially at the bottom of the screen. The Horizontal masking works well (I did not know about this feature, but it adds that extra bit of polish to the screen ).

hmm, I have a weird problem. Since updating, the brightness seems to be way too high now...

Ok I fixed it. For some reason I had to manually adjust the brightness in the gbscontrol menu. Anyway, I put together a demonstration video showing 240p/480i transitions

https://youtu.be/s7FooVlXegg

RockeTim:

Nice, I'm all for more documentation. I don't have the time to create all that content, so if others did it, that'd be great!
I don't see a link to your work though?

AndehX:
Nice video! Looks like the color space issues are reduced as well
I'm not sure why you get an overly bright image now.
I've worked on the deinterlacer transitions, so maybe there's a new problem there now.

It's fine, once I dialed it in and saved a custom preset. If I load the defaults though, they're way too bright

Great video there, AndehX! What did you use for that video? Now you guys all got me thinking about upgrading my capture device...

Sent from my SM-G955U using Tapatalk

PEXHDCAP, although any recent capture device that supports HDMI should be ok. Elgato HD60 Pro should work fine afaik

@rama here is the link to the page: https://github.com/ramapcsx2/gbs-contro ... od-Library

If anyone wants to help, feel free to update the page, or (if you don't have a GitHub) put the change in here and I can update it as I'm able.

If possible I'd like to have links to pictures for each mod to include in the description as well.

Hey, great job, RockeTim! I will try to provide some links, to hosted pictures already buried in this thread. Will do that on Saturday.

Question - I think I missed this one which sounds like a good idea. What is the "wire to ground on component shielding?" Is the normally grounded outer shell of the component inputs not properly grounded on the gbs?

Sent from my SM-G955U using Tapatalk

Ah, the page was on my Wiki! And that's even the reason I made it public :p

Okay, I'll edit some details.

Edit:
I removed a few items that I didn't know of and that didn't make sense to me.
Particularly "Add wire to ground on component input shielding" is an odd one. The metal on those inputs is very well grounded
I also reordered it a little, so the list starts with important items.

Thanks a lot, RockeTim!

I was repairing a launch Wii yesterday and found something very interesting.

The CPU heatsink legs have ferrite rings around each of them.

I was trying to think of why Nintendo did this, and I started thinking about how our heatsinks work as an aerial and pickup Wemos wifi and other crap.
Maybe they wanted to try eliminate as much RF as possible before dumping it into the ground plane?
Regardless it was removed in later revisions