Jump to content
IGNORED

FPGA Based Videogame System


kevtris

Interest in an FPGA Videogame System  

682 members have voted

  1. 1. I would pay....

  2. 2. I Would Like Support for...

  3. 3. Games Should Run From...

    • SD Card / USB Memory Sticks
    • Original Cartridges
    • Hopes and Dreams
  4. 4. The Video Inteface Should be...


  • Please sign in to vote in this poll.

Recommended Posts

Imagine if someone bought a new OLED TV and spent three years playing the Super NT on it with scanlines always enabled. Those same scanlines would be permanently etched into the screen when viewing normal non-retro content. That's not a design flaw, that's a feature! :rolling:

 

I've already been doing this for months with the NT Mini and OLED.

 

FWIW every week I shift the image up a couple of pixels, then down, and left and right by way more than a couple of pixels. And on the OSSC (mostly 16:9 480p Wii currently) I alternate between top scanlines and bottom scanlines.

 

I'm showing no signs of burn-in but I watch a variety of content.

  • Like 1
Link to comment
Share on other sites

 

I've already been doing this for months with the NT Mini and OLED.

 

FWIW every week I shift the image up a couple of pixels, then down, and left and right by way more than a couple of pixels. And on the OSSC (mostly 16:9 480p Wii currently) I alternate between top scanlines and bottom scanlines.

 

I'm showing no signs of burn-in but I watch a variety of content.

 

That's some dedication! Probably not necessary, but I commend you for trying to keep your set in great condition. You should throw the OSSC's vertical scanlines (I think they're meant for shmup TATE mode) in the mix just for the heck of it. :lol:

Link to comment
Share on other sites

With up-scaling you only really run into trouble with modern TV's when dealing with old SD resolutions like 240p.. Some modern TV's treat it correctly and others treat it like 480i which just makes 240p look terrible. If a newer television has to truly upscale a non-integer resolution then there will likely be some lag generated but between HD signals like 720p and 1080p and even the standard 4k resolutions the TV's don't really have much work to do and lag created by those conversions is minimal. I'm certainly no expert in this area but I've witnessed it enough and watched MLIG videos enough times through to get the just of it I think.

Link to comment
Share on other sites

If it's true that modern tv's aren't adding upscaling lag anymore that puts my mind at ease about my NT Mini not being obsoleted once I get a 4k tv!

It is not true, they certainly in many cases are getting better than they once were but it all depends what input signal needs to be upscaled / multiplied to 4k. Standard 4k resolution is essentially 4 1080p screens in one panel and up-scaling from 1008p is a simple 4x multiplyer or assigning 4 pixels on a 4k panel to each pixel of a 1080p signal. This would not be true upscaling like say the framemeister does so well, it would be more like line multiplication like the OSSC does. similar with 720 which needs a simple 3x multiplier to get to 4k and a 1.5x multiplier to hit 1080p, these are easy for a TV to upscale. TV's just do terribly with older lower resolution signals in general, which is why 240p signals look so terrible when fed straight into a modern set. The framemeister is specialized to do what modern TV's do so poorly but you pay for it in $$$ and in lag, the framemeister still adds some lag because of the buffering it needs to do to properly upscale those signals. TV's should upscale 240p the same way they deal with 720p and some TV's do like my Samsung but many modern TV's see 240p signals as 480i which really mucks up the upscale work it does.

Link to comment
Share on other sites

I started experimenting with my Sony 55 inch LED display. I set both Brightness and Color to the middle position of 50, and then started testing the Super Nt in both Full RGB space and Limited RGB space. Full mode showed all colors, but looked horribly washed out. Limited mode looked much more vibrant and 'inviting' to play. A good spot to check for this is outside Link's house in A Link to the Past. Switching between the two modes, Limited looked much better. Furthermore, when I set my TV to detect color space automatically, it chooses Limited mode for the Super Nt.

 

But here's the thing: the blacks are a bit crushed in Limited mode on the TV. On my Datapath Vision E1S captures, Limited mode for the Super Nt crushed the blacks AND the whites. Furthermore in both modes, the upper 4 blue shades were crushed together in both direct capture and the live feed on my TV. Take a look at these color ramp tests (Full RGB on the left, Limited RGB on the right):

EP7bpyb.png

It's strange how my card crushes the upper whites in limited mode, whereas my TV does not. Both show the same level of crushing of the blacks though. But take a look at the upper blue shades. Even in full mode, the upper 4 shades are crushed together, and this happens on my TV as well.

Here's that same color test pattern in Higan for reference:

v7Un5sI.png

Edited by Karbuncle
Link to comment
Share on other sites

I started experimenting with my Sony 55 inch LED display. I set both Brightness and Color to the middle position of 50, and then started testing the Super Nt in both Full RGB space and Limited RGB space. Full mode showed all colors, but looked horribly washed out. Limited mode looked much more vibrant and 'inviting' to play. A good spot to check for this is outside Link's house in A Link to the Past. Switching between the two modes, Limited looked much better. Furthermore, when I set my TV to detect color space automatically, it chooses Limited mode for the Super Nt.

 

But here's the thing: the blacks are a bit crushed in Limited mode on the TV. On my Datapath Vision E1S captures, Limited mode for the Super Nt crushed the blacks AND the whites. Furthermore in both modes, the upper 4 blue shades were crushed together in both direct capture and the live feed on my TV. Take a look at these color ramp tests (Full RGB on the left, Limited RGB on the right):

 

EP7bpyb.png

 

It's strange how my card crushes the upper whites in limited mode, whereas my TV does not. Both show the same level of crushing of the blacks though. But take a look at the upper blue shades. Even in full mode, the upper 4 shades are crushed together, and this happens on my TV as well.

I am a stickler about things like input lag, and while I see what you are pointing out here and I saw the color differences earlier I don't see that sort of thing at all until someone points it out to me. You sir have one heck of a discerning eye.

Link to comment
Share on other sites

I am a stickler about things like input lag, and while I see what you are pointing out here and I saw the color differences earlier I don't see that sort of thing at all until someone points it out to me. You sir have one heck of a discerning eye.

 

Oh it wasn't just me. I just had noticed some complaints starting to float around on various forums about the Super Nt's colors being off, and I started looking into the matter myself. Sure enough, there seems to be some issues going on that I'm curious about the exact nature of. Right now, it's a trade-off decision: Full mode gives you more range of colors, but they looked washed out, while Limited mode looks more appealing, but crushes blacks (and in some cases upper whites). And then there's of course some crushing going on in the upper blues in both modes. /shrug

 

 

Link to comment
Share on other sites

With up-scaling you only really run into trouble with modern TV's when dealing with old SD resolutions like 240p.. Some modern TV's treat it correctly and others treat it like 480i which just makes 240p look terrible. If a newer television has to truly upscale a non-integer resolution then there will likely be some lag generated but between HD signals like 720p and 1080p and even the standard 4k resolutions the TV's don't really have much work to do and lag created by those conversions is minimal. I'm certainly no expert in this area but I've witnessed it enough and watched MLIG videos enough times through to get the just of it I think.

 

While this certainly isn't a scientific case study, I wanted to try and put this to bed. Anyone using a somewhat modern HD panel should see similar results as far as differences between DIGITAL source resolutions. (not saying your particular display will be as quick as these results)

 

This is my Genesis+OSSC to a 4K set. The following source resolutions being dealt with by the TV are all digital. No analog conversion on the TV's end, just upscaling digital - digital.

 

240P Passthru (240P source)

post-45470-0-37032100-1518501537.jpg

 

Line 2X (480P source)

post-45470-0-78699500-1518501542.jpg

 

Line 3X (720P source)

post-45470-0-76370400-1518501549.jpg

 

Line 4X (960P source)

post-45470-0-36855900-1518501559.jpg

 

Line 5X (1080P source)

 

post-45470-0-05032000-1518501565.jpg

 

While there may actually be some microsecond difference between all of those, as far as real world gameplay goes. There's no friggin difference.

 

So, feel free to set your SuperNT to 720P if you want.

post-45470-0-37032100-1518501537.jpg

post-45470-0-78699500-1518501542.jpg

post-45470-0-76370400-1518501549.jpg

post-45470-0-36855900-1518501559.jpg

post-45470-0-05032000-1518501565.jpg

Edited by keepdreamin
Link to comment
Share on other sites

Update: I found on my Sony display its possible to 'counteract' the washed out look of full RGB mode on the Super Nt by manually boosting saturation. I balanced everything to the limits of the color ramp test pattern on my display and came up with the following settings:

 

RGB mode = Full

 

Picture = 90

 

Brightness = 50

 

Color = 57 (highest it would go before crushing colors together)

 

It looks pretty damned good now under these settings. Not perfect, but pretty damned good.

Edited by Karbuncle
Link to comment
Share on other sites

TV's should upscale 240p the same way they deal with 720p and some TV's do like my Samsung but many modern TV's see 240p signals as 480i which really mucks up the upscale work it does.

My Pioneer Kuro plasma handled 240p properly. Of course the upscale was soft since it was going to 1080p and not 960p. You had to put the set in 'game' mode to get it to display 240p properly.

 

All of the current LG sets, OLED and LCD, seem to handle 240p properly. None try to 'deinterlace' the signal.

Link to comment
Share on other sites

 

While this certainly isn't a scientific case study, I wanted to try and put this to bed. Anyone using a somewhat modern HD panel should see similar results as far as differences between DIGITAL source resolutions. (not saying your particular display will be as quick as these results)

 

This is my Genesis+OSSC to a 4K set. The following source resolutions being dealt with by the TV are all digital. No analog conversion on the TV's end, just upscaling digital - digital.

 

240P Passthru (240P source)

post-45470-0-37032100-1518501537.jpg

 

Line 2X (480P source)

post-45470-0-78699500-1518501542.jpg

 

Line 3X (720P source)

post-45470-0-76370400-1518501549.jpg

 

Line 4X (960P source)

post-45470-0-36855900-1518501559.jpg

 

Line 5X (1080P source)

 

post-45470-0-05032000-1518501565.jpg

 

While there may actually be some microsecond difference between all of those, as far as real world gameplay goes. There's no friggin difference.

 

So, feel free to set your SuperNT to 720P if you want.

 

Wow, impressive results. What's the supposed input lag of that monitor? From your tests it appears to be <1ms.

 

I just did that manual lag test on my TV (that supposedly has a lag of 21ms according to rtings) and it actually reported exactly 21ms.

  • Like 1
Link to comment
Share on other sites

How does the Super NT compare to a SNES fed through an OSSC to a 1080p tv?

Image should be more stable on the Super NT (jitter may affect the OSSC slightly), audio shouldn't hum. OSSC currently has better scanlines (you can adjust strength; which may adjust thickness in some modes, not sure.

 

I'd recommend 4x with black bars, or 5x with slight overscan, on both, even though 720p has better scanlines you'll get 1.5x scaling on the vertical by the display.

Link to comment
Share on other sites

For what it's worth, I just had a quick look at the Higan source code, and it does some magic with the ouput:

auto Interface::videoColor(uint32 color) -> uint64 {
  uint r = color.bits( 0, 4);
  uint g = color.bits( 5, 9);
  uint b = color.bits(10,14);
  uint l = color.bits(15,18);

  //luma=0 is not 100% black; but it's much darker than normal linear scaling
  //exact effect seems to be analog; requires > 24-bit color depth to represent accurately
  double L = (1.0 + l) / 16.0 * (l ? 1.0 : 0.25);
  uint64 R = L * image::normalize(r, 5, 16);
  uint64 G = L * image::normalize(g, 5, 16);
  uint64 B = L * image::normalize(b, 5, 16);

  if(settings.colorEmulation) {
    static const uint8 gammaRamp[32] = {
      0x00, 0x01, 0x03, 0x06, 0x0a, 0x0f, 0x15, 0x1c,
      0x24, 0x2d, 0x37, 0x42, 0x4e, 0x5b, 0x69, 0x78,
      0x88, 0x90, 0x98, 0xa0, 0xa8, 0xb0, 0xb8, 0xc0,
      0xc8, 0xd0, 0xd8, 0xe0, 0xe8, 0xf0, 0xf8, 0xff,
    };
    R = L * gammaRamp[r] * 0x0101;
    G = L * gammaRamp[g] * 0x0101;
    B = L * gammaRamp[b] * 0x0101;
  }

  return R << 32 | G << 16 | B << 0;
}

If you have enabled "colorEmulation" in Higan it will use it's own gamme ramp to simulate the non-linear analoge signal.

I havn't received my Super NT yet (It will arrive tomorrow!) so I can't comment on the colors compared to Higan yet.

However, everyone needs to pay attention to a lot of factors before making incorrect statements:

 

1. Higan runs on windows. This means the several factors can affect the HDMI output. What colorspace have you selected in the Nvidia/AMD/Intel control panel?

Nvidia supports both RGB/YUV color space in both limited/full color space. For doing a fair comparison you should make sure that you have selected Full RGB color space output.

Also, make sure that you have not activated any calibrated ICC color profile. That will also change the output from the PC.

Also make sure that you havn't enabled any fancy settings in Higan like shaders etc. (quite obvious I think....)

 

2. Higan is not the reference for comparison. The real SNES hardware is the real deal. I guess the real SNES also differs quite a bit between chip revisions (Early Revs up to the 1-Chip etc.).

 

3. NTSC-U vs NTSC-J vs. PAL.... These will probably look different as well on a high-end Sony BVM monitor...

 

In order to fix this properly, someone should make a video test-rom for the SD2SNES and run it on several real SNESs. The analoge RGB signal levels would need to be sampled (or checked using an oscilloscope) for all red/green/blue values from black to 100% color. red/green/blue might have different gamma ramp curves.

The results should be averaged using several hardware consoles to rule out variation in old capacitors etc.. And this also needs to be done for Both NTSC and PAL consoles...

 

Then these gamma curves/lookup tables/math formulas can be applied to the pixel processing in the Super NT. You would then need built-in presents for NTSC-J/U and PAL or raw linear, that the user can select :)

 

I guess @kevtris has already looked into this already. Maybe you can fill in something here @kevtris?

 

 

-Eicar

Link to comment
Share on other sites

 

Wow, impressive results. What's the supposed input lag of that monitor? From your tests it appears to be <1ms.

 

I just did that manual lag test on my TV (that supposedly has a lag of 21ms according to rtings) and it actually reported exactly 21ms.

 

It really isn't < 1 ms though. It's a TCL 4K set with around 14ms of actual measured lag. The manual lag test just shows it's quick enough of a display as one actually needs for real world gameplay.

Edited by keepdreamin
Link to comment
Share on other sites

For what it's worth, I just had a quick look at the Higan source code

 

I'm pretty sure the output from Higan is the incorrect one in this testing methodology then.

 

But I wouldn't say the output is actually wrong. If you check "colors" in v106 the colors are pretty much the same as the SuperNT

Edited by Kismet
Link to comment
Share on other sites

Just read on Nerdly Pleasures that the SNES PowerPak will not load games on the Super NT. I'm going to get a SD2SNES when I can afford it, but I was really hoping to use my PowerPak in the meantime. Has anyone heard any word on this issue?

I have no idea what's causing this issue, but SNES Powerpak appears to be depreciated. It hasn't been available in a long time. I got the Super Everdrive because it was cheaper and it had automatic saving. Still though if it takes somebody loaning their Powerpak to Kevtris to get it fixed, I'm sure somebody will step up.
Link to comment
Share on other sites

I havent gotten my Super NT yet, so I havent downloaded the 4.0 firmware yet. What size SD Card is needed?

I would recommend at minimum an 8Gb card. That will be enough to fit a full no-intro SNES set should the jailbreak drop in the future. Otherwise, anything will do as the firmware isn't huge.
Link to comment
Share on other sites

Can we drop this already? Banks don't understand tech. The second someone starts talking to them about field programmable gate arrays not performing correctly the mid 40s heavy smoker chick I'm talking to is going to roll her eyes and stamp a paper. It's a bank not a patent office.

 

That's an assumption that banks do not understand tech. You don't even need to go into FPGA or anything of that sort when disputing. Analogue would fire back and explain that what you've disputed is irrelevant. Returns are only accepted for defective units. Analogue would receive the unit, report to the bank that it was not defective and you would be held accountable. Your explanations are like getting pulled over by a cop and saying "no officer, the drugs you see in my car do not belong to me. They belong to my second cousin's former roommate who was a chemist and was testing them...." and thinking "he will never arrest me because I explained something over his head.

 

Bottom line, do not assume people are stupid. All those people I explained who are sitting in prison thought everyone else was an idiot.

Link to comment
Share on other sites

 

While this certainly isn't a scientific case study, I wanted to try and put this to bed. Anyone using a somewhat modern HD panel should see similar results as far as differences between DIGITAL source resolutions. (not saying your particular display will be as quick as these results)

 

This is my Genesis+OSSC to a 4K set. The following source resolutions being dealt with by the TV are all digital. No analog conversion on the TV's end, just upscaling digital - digital.

 

240P Passthru (240P source)

post-45470-0-37032100-1518501537.jpg

 

Line 2X (480P source)

post-45470-0-78699500-1518501542.jpg

 

Line 3X (720P source)

post-45470-0-76370400-1518501549.jpg

 

Line 4X (960P source)

post-45470-0-36855900-1518501559.jpg

 

Line 5X (1080P source)

 

post-45470-0-05032000-1518501565.jpg

 

While there may actually be some microsecond difference between all of those, as far as real world gameplay goes. There's no friggin difference.

 

So, feel free to set your SuperNT to 720P if you want.

You know I had left out the idea of the television receiving a digital 240p signal through the HDMI port like through an OSSC. I would think many if not all TV's would handle that one fairly well since it is already converted to a progressive digital multiple of 4k signal (TV has very little work to do to display it in 4k). This is what really makes the OSSC so amazing.. I think the issues come with many TV's when they are making the Analogue to Digital conversion and up-scaling. Many TV's see 240p coming in through composite as a 480i signal and when they upscale it it looks generally terrible and there is some more lag than a digital to digital multiple scaling method like happens with any signal the OSSC can deliver to a television. If the tv need only upscale a digital signal that is a multiple of 4k like all of these sources are in your examples I should think there would be no perceivable difference and likely no perceivable lag added. I think the real problem with built in up-scalers has really always been how good or bad they are at de-interlacing signals and many treating 240p analogue input as 480i (interlaced).. This is the task the Framemeister does so much better than a TV's built in up-scaler.

 

If you are dealing with a set that has say 16ms of input lag like my samsung 4k set and using an OSSC that adds really no perceivable lag to the setup then you should not really be noticing any and I really don't (every person is different in their sensitivity to input lag). I'm not sensitive enough in "most" titles to notice 1-2 frames of lag. However, when I play games on the same TV using my Framemeister I can sense that input lag & I'd say it is about 3-4 frames in total in that setup.

 

I also own a BENQ gaming monitor which handles all modes on the OSSC and everything I've ever thrown at it. It is advertised to have 1ms or nearly no input lag for an LCD and is a 1080p set. I think they make a 4k version too.. When I play on that screen through my OSSC it is pure bliss and I can not tell a difference lag wise from my CRT setup. This BENQ set is also I think one of the best ways to play if you are using the Frameister as the entire setup is only about 1.5-2 frames of lag and I can't really sense any lag in that setup. Despite being only 27" display my preferred method to play retro games on a modern display is this monitor... I'm also pretty close to it.

  • Like 1
Link to comment
Share on other sites

[...]

Don't bother replying, he's already been banned from this topic.

 

Anyone else still waiting for their units? Kind of nutty how it seems most of Europe got theirs already... but if you're in the US you might still be waiting...

UPS has been sitting on my package since Saturday. They won't release it until scheduled delivery date... :roll:
Link to comment
Share on other sites

For the record, mine is the dreaded VA7. The RF and Mono audio sounds like it was pushed through a grunge guitar pedal, but you can get perfectly balanced clean stereo sound out of the console by plugging a cable into the headphone jack and feeding it directly to the sound system with the volume set to 5.

 

Yeah, even the VA7 via stereo jack is pretty muffled compared to previous revs. You could always get it Mega Amp'd.

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Loading...
  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...