240p test suite for DC,PCE,Wii,SNES,GC,MD and SCD

[FBX]

@FBX
You can also use the NES-version if your N64 flashcard has got a NES-emulator. I used both, the one from page 9 and the NES-version. I'm not sure if the color patterns are really accurate through using them through the N64's emulators. It's at least very helpful to set sharpness and aspect ratio correctly.

That won't do. It needs to be the internal palette spread of the N64's graphics so I can adjust A/D accordingly. The NES would not cover the entire N64 range.

There's an emulated version for the N64 on page 9 of this thread.

I tried it out using the Everdrive 64 v.3. It works, but again is emulation of the SNES RGB spread rather than using the N64's complete RGB spread. It runs at about 10% speed due to being emulated too.

Edit: Although only emulating the SNES palette, it did show my A/D of 160 for my amped N64 was a bit too high. Lowering A/D to 140 and a gamma of 19 gave me best coverage of the RGB spread without overbrightening mid-tones. Still though, if I want to get even more accurate, I'll need to see an RGB spread of the full N64 range. Nevertheless, I'm going to update my profiles with the new settings.

[Guspaz]

Doesn't the SNES cover the full RGB range, from 0 to 255 on each colour channel? How does the colour range differ from the N64?

[FBX]

Doesn't the SNES cover the full RGB range, from 0 to 255 on each colour channel? How does the colour range differ from the N64?

SNES is 32,768 colors, which is 0 to 31 on each channel.

The N64 is supposed to be 0 to 255 if what I read online is correct (16.7 million total colors).

But at any rate, I suppose it would get difficult to discern 254 and 255 on a given channel, so probably the emulated SNES bar pattern is sufficient.

[Artemio]

I just bought a n64 flash cart, I'll see what I can do.

[Guspaz]

Doesn't the SNES cover the full RGB range, from 0 to 255 on each colour channel? How does the colour range differ from the N64?

SNES is 32,768 colors, which is 0 to 31 on each channel.

The N64 is supposed to be 0 to 255 if what I read online is correct (16.7 million total colors).

But at any rate, I suppose it would get difficult to discern 254 and 255 on a given channel, so probably the emulated SNES bar pattern is sufficient.

It's analog output, the range should be the same, with less steps in the range for the SNES. I'm not a video expert, but the 240p test suite doesn't seem to have any problems outputting the SMPTE test bars at their full range, which is supposed to include the full range, IIRC.

[FBX]

Doesn't the SNES cover the full RGB range, from 0 to 255 on each colour channel? How does the colour range differ from the N64?

SNES is 32,768 colors, which is 0 to 31 on each channel.

The N64 is supposed to be 0 to 255 if what I read online is correct (16.7 million total colors).

But at any rate, I suppose it would get difficult to discern 254 and 255 on a given channel, so probably the emulated SNES bar pattern is sufficient.

It's analog output, the range should be the same, with less steps in the range for the SNES. I'm not a video expert, but the 240p test suite doesn't seem to have any problems outputting the SMPTE test bars at their full range, which is supposed to include the full range, IIRC.

Well but more steps gives me more accuracy on the 16.7 million color consoles like the N64, Gamecube, and Wii.

At any rate, the Wii and Gamecube test suits also seem to use the 32-shade test pattern from the SNES design. Perhaps Artemio can elaborate on this?

I admit the bars would be rather small doing a 256 step, especially in low res modes. Perhaps ideally we'd have a separate "High End" color bar test pattern, where 32 of the top end colors for each channel are shown. That would REALLY allow me to fine-tune A/D using a pattern like that.

[Artemio]

Well but more steps gives me more accuracy on the 16.7 million color consoles like the N64, Gamecube, and Wii.

At any rate, the Wii and Gamecube test suits also seem to use the 32-shade test pattern from the SNES design. Perhaps Artemio can elaborate on this?

I admit the bars would be rather small doing a 256 step, especially in low res modes. Perhaps ideally we'd have a separate "High End" color bar test pattern, where 32 of the top end colors for each channel are shown. That would REALLY allow me to fine-tune A/D using a pattern like that.

Yes, Dreamcast, Wii and Gamecube use a 32 shade pattern. They use:

Code: Select all

index	16b hex	pcnt
0	0	0%
0	9	4%
1	11	7%
1	19	10%
2	21	13%
2	2A	17%
3	32	20%
3	3A	23%
4	42	26%
4	4B	30%
5	53	33%
5	5B	36%
6	63	39%
6	6B	42%
7	74	46%
7	7C	49%
8	84	52%
8	8C	55%
9	95	59%
9	9D	62%
A	A5	65%
A	AD	68%
B	B5	71%
B	BE	75%
C	C6	78%
C	CE	81%
D	D6	84%
D	DF	88%
E	E7	91%
E	EF	94%
F	F7	97%
F	FF	100%

Right now the Dreamcast for example, uses 16 bit color. I could make a pattern in 24bit color for such a test, and check how that behaves under the scope. But which values would be useful for you? I could do the same for the GC and Wii versions.

And yes, that was selected because it is a 3% difference and because having more colors would not fit in the resolution. It would be a 1 pixel wide bar for each color in 320x240, and two pixels wide in 640x480.

Do you really want just colors 223 to 255 on each bar? In the same arrangement and width they are, I could make it so you can swap between the color patterns.

EDIT: I need to test t on the GC/Wii, since I am already using 32bit color on those by default, but I never tried encoding a texture deeper than 16 bits.
The DC does support it, but I would probably need to create the image in code and not import it as a texture, since the texture tools don't seem to support 24bit textures.. Or I'd need to modify the tools (once again) to do so. WIll look into it after your feedback.

[FBX]

Do you really want just colors 223 to 255 on each bar? In the same arrangement and width they are, I could make it so you can swap between the color patterns.

Actually that would be awesome if you can pull that off! Please use the N64 and/or Gamecube one as the testing ground as I have immediate access to those.

[Artemio]

I don't have an n64 cart in my hands yet. Neither have I started development on a suite for it.

I was currently updating the Dreamcast version, but can switch to Gamecube and test there. I'll PM you a beta version when I have something to test with.

[FBX]

I don't have an n64 cart in my hands yet. Neither have I started development on a suite for it.

I was currently updating the Dreamcast version, but can switch to Gamecube and test there. I'll PM you a beta version when I have something to test with.

Awesome! Thanks a bunch! This will make fine-tuning A/D on the Framemeister a breeze!

[Einzelherz]

Artemio, is a 720p capable PS2 version still unlikely?

[Artemio]

I don't have a modded ps2, and I believe there is no open source sdk for PS2 either.

Right now I'm updating the Wii/Gc/DC versions. When I finish I'll look into either ps1/saturn/n64.

[Unseen]

I don't have a modded ps2, and I believe there is no open source sdk for PS2 either.

ps2sdk appears to be open source and there are various ways to launch homebrew software on unmodified PS2s, for example with a memory card that was prepared with FMCB.

[NormalFish]

I don't have a modded ps2, and I believe there is no open source sdk for PS2 either.

Right now I'm updating the Wii/Gc/DC versions. When I finish I'll look into either ps1/saturn/n64.

PS1 version would be fantastic. Looking forward to it.

PS2 would also be nice, of course, but it's easier to boot a PS1 iso than a PS2 one on my unit.

[Artemio]

I don't have a modded ps2, and I believe there is no open source sdk for PS2 either.

ps2sdk appears to be open source and there are various ways to launch homebrew software on unmodified PS2s, for example with a memory card that was prepared with FMCB.

Thanks a lot, I'll put it on the list as well.

[Ikaruga11]

So what will the updated GameCube and Wii 240p Test Suite do?

[Guspaz]

Any idea why the PAL GameCube image is shifted a bunch to the right as compared to the NTSC GameCube, even when the PAL GameCube is outputting an NTSC resolution like 480i60?

[Artemio]

So what will the updated GameCube and Wii 240p Test Suite do?

They'll have the audio sync test, the vertical scroll test with grtaphics, frame perfect audio feedback in the Manual lag test, a few bugs corrected, help updated and probably the color ramp requested above.

Any idea why the PAL GameCube image is shifted a bunch to the right as compared to the NTSC GameCube, even when the PAL GameCube is outputting an NTSC resolution like 480i60?

No, so far all tests on PAL work without shift in my PVMs on my US/JP Gamecubes. I don't own a PAL one, but there have been no reports so far. I wouldn't discard it is wrong on such a setup, but I've had no feedback in that sense.

Does PAL work correctly on your NTSC Gamecube? How does the other homebrew behave n your PAL gamecube? Genesis DX?

[Ikaruga11]

They'll have the audio sync test, the vertical scroll test with grtaphics, frame perfect audio feedback in the Manual lag test, a few bugs corrected, help updated and probably the color ramp requested above.

Oh nice. That sounds awesome.

Pardon my ignorance, but what is the purpose of each new feature and how to test for it? Do I need special equipment?

[Guspaz]

No, so far all tests on PAL work without shift in my PVMs on my US/JP Gamecubes. I don't own a PAL one, but there have been no reports so far. I wouldn't discard it is wrong on such a setup, but I've had no feedback in that sense.

Does PAL work correctly on your NTSC Gamecube? How does the other homebrew behave n your PAL gamecube? Genesis DX?

Sorry, I should have been more specific. It's not a problem specific to the 240p test suite, I just figured you'd be a good person to ask

The PAL GameCube shows the offset to the right on the GameCube boot menu (which is PAL, obviously), it shows it in GBI, in NTSC Windwaker, and in the 240p test suite in both PAL and NTSC video modes.

The NTSC GameCube doesn't have the offset, nor do any of my other consoles, all of which are NTSC.

I could understand if PAL had some sort of offset, because the timing is quite different: my PVM even loads a different set of geometry calibration settings for PAL (if I switch between PAL and NTSC I can see the numbers in the service menu change). What's confusing is why the PAL GameCube *still* has that offset when outputting an NTSC resolution.

I figure there must be some sort of difference in the sync timing on the PAL cube in general, regardless of resolution and framerate.

[Extrems]

Check Options > Screen Position in the IPL.

[Guspaz]

Well, serves me right for getting a GameCube and not playing around in the menus, I didn't even know such an option was there!

It does get a bit weirder though: I went into the position settings, and found that it was set to 32 (the max to the right). Except when I went in there, the screen re-centered itself so that 32 is correct center for PAL.

However, NTSC is still offset, and so even though I left the setting at 32 (and never hit confirm), when I switch back and forth between PAL and NTSC, PAL is centered, while NTSC is offset to the right. Before going in there, both PAL and NTSC were offset to the right.

I set it to "0", the neutral value between -32 and +32, and while PAL is now offset to the *left*, NTSC is correct.

Considering that I intend to treat this like an NTSC GameCube (I want RGB, not PAL), this is perfect!

Thanks Extrems!

[Artemio]

Actually that would be awesome if you can pull that off! Please use the N64 and/or Gamecube one as the testing ground as I have immediate access to those.

Done:



Not a very good picture, but you can see the slope on the oscilloscope.

Sorry it took so long, but I had to make sure the levels were right with the scope, and I needed to find my composite cables for that after creating the patterns. I made it so you have the default 16 bit color bards when entering the test, pressing left gives you 00 to 1F, and pressing right gives you E0 to FF. Sending you the link to the test via PM. I'll release it tomorrow if everything goes well and I finish the Audio Sync test that is still missing.

Pardon my ignorance, but what is the purpose of each new feature and how to test for it? Do I need special equipment?

Audio Sync is to verify if your audio is on sync, you'd need either specialized hardware or simply video editing software to check in a recorded video if they are in sync as generated by the console. The other two need nothing, and are described on the help.

Sorry, I should have been more specific. It's not a problem specific to the 240p test suite, I just figured you'd be a good person to ask

No problem man, good thing Extrems knew what was happening.

In general news, I'll probably have news on the SNES physical release soon, just as with the Genesis one I am not making any money of it since the intention is to have it sold for as little as possible.

[FBX]

Thanks a bunch for working on this Artemio!

I tried it out on the Gamecube and it is fully functional. The dark spectrum to the left worked perfectly! However, it seems near impossible to discern colors on the high end, which perhaps I was too enthusiastic about the theory of the concept. On the high end, would you be willing to go from C0 to FF in 32 steps? This makes the graduation 2 per bar if my math is correct.

[Artemio]

I imagined so, that's why I was surprised by the request.

I'll try it and send you only the red part to check if it works. If it does, I'll do the others (I have to create them by hand, since they are textures). In such case, I guess I should move the range for the low end to make it symmetric? or leave these and also the other two?

[FBX]

I imagined so, that's why I was surprised by the request.

I'll try it and send you only the red part to check if it works. If it does, I'll do the others (I have to create them by hand, since they are textures). In such case, I guess I should move the range for the low end to make it symmetric? or leave these and also the other two?

I figure it will be fine to make them symmetric. Basically what I'm after on the high end is a range that is more detailed than the full range, but still distinct enough to calibrate by eyesight with. I would say use green instead of red as the solo test, because it is easier to the naked eye to spot variations from. Basically use your best judgement. If C0 to FF isn't distinct enough, try doubling once more and that should definitely do the trick.

[Artemio]

I imagined so, that's why I was surprised by the request.

I'll try it and send you only the red part to check if it works. If it does, I'll do the others (I have to create them by hand, since they are textures). In such case, I guess I should move the range for the low end to make it symmetric? or leave these and also the other two?

I figure it will be fine to make them symmetric. Basically what I'm after on the high end is a range that is more detailed than the full range, but still distinct enough to calibrate by eyesight with. I would say use green instead of red as the solo test, because it is easier to the naked eye to spot variations from. Basically use your best judgement. If C0 to FF isn't distinct enough, try doubling once more and that should definitely do the trick.

I though about exactly the same thing regarding green after doing red. I just sent you the test with a few green tones and the full red from C1 to FF. Let me know what you think. I have a scope and a vectorscope here that I use to figure out the values, and by eye I can in my bigger PVMs. The idea is to have something useful.

[FBX]

Tested the new wip. I think I've got the perfect solution that will work great as a universal option on all the 16.7 million color consoles:

Default = 32-step full range.
Pressing left shows exactly the lower half of the range in 32 steps (00 to 7F I believe)
Pressing right shows exactly the upper half of the range in 32 steps (80 to FF)

So we get symmetrical and useful screens at the same time. Agreed?

BTW thanks a bunch for indulging my OCD on this!

[Artemio]

Ok, so steps of 4 in the details, instead of 8 as it is for 16 bits.

I'll send you a test with greens only, ok?

We all have some degree of OCD here =)

[FBX]

Ok, so steps of 4 in the details, instead of 8 as it is for 16 bits.

I'll send you a test with greens only, ok?

We all have some degree of OCD here =)

Sounds great!
Thanks again!