Re: Mode 5 colour limitations?
Posted: Fri May 13, 2022 7:03 pm
That is if you can stand how interlacing looks on your TV, at least with a CRT. SNES doesn't natively have deflicker filters for interlacing either.
That is if you can stand how interlacing looks on your TV, at least with a CRT. SNES doesn't natively have deflicker filters for interlacing either.
This goes against what I've seen with my own testing though. I've tested on a number of sizes of consumer CRT TVs via composite, and mode 5 was noticeably more crisp on all of them.Dwedit wrote: ↑Fri May 13, 2022 6:06 pm Old games used "Psuedo Hires" mode to do 50% transparency with another layer, so on a consumer television connected by Composite, it won't look very high-res. But it does work fine for white text against a dark background, Secret of Mana's hires text was quite readable.
Was pseudo-hires also more crisp? Did you see any vertical stripes when playing a SNES game like Kirby Dreamland 3 that uses pseudo-hires to fake more translucency?
Pseudo-hires blends into transparency with a bit of a grainy texture. In mode 5, individual pixels are too fine to be discerned via composite on a CRT, and adjacent pixels do tend to blend together or have shifted colors. However, that doesn't mean the increase in resolution is imperceptible. To say "it won't look very high-res on a consumer television via composite" is inaccurate, at least from what I've seen.
Hmm, pretty interesting how a feature that sounds like it would be good for rendering such complex characters on paper ends up not too useful due to the video cables and TVs most people were using back then.
Chrominance can't, in fact. NTSC composite is limited to somewhere around 140 pixels of chroma per scanline, so even just the 256px mode easily exceeds the permitted chroma bandwidth.rainwarrior wrote: ↑Sat May 14, 2022 1:29 pm In my tests, the hi-res luminance definitely comes through a bit on composite, but chrominance doesn't at all, as far as I can tell.
I think all of these examples would probably look very clean on RGB (or S-Video).
The US version of Desert Fighter, called Air Strike Patrol, has equivalent stuff in English, but since the characters are much simpler there is no deficiency of detail to speak of. I don't know if there's nearly as much value in trying to increase letter detail like this for English?
Oh I wasn't actually talking about these specific examples, I was just talking about CJK characters in general, I'm sure there are CJK characters out there where even a crystal clear 32x32 isn't enough to properly render them.rainwarrior wrote: ↑Sat May 14, 2022 11:34 pm I think all of these examples would probably look very clean on RGB (or S-Video).
The composite colour signal is the biggest problem here. There's a hard limit on how much detail you can encode, and various TVs will take stronger (blurrier) and weaker (error artifactier) approaches to decoding the signal.
Typical RGB computer monitors of the time should have had no problem with this resolution, if you could somehow connect an RGB SNES signal to them. I don't know much about consumer TVs with RGB from that era.... They might have their own filters that would reduce effective resolution, or have an unfocused picture, or other problems, but I think something truly built for RGB should easily be capable of having a nice clear picture?
I actually wonder how much clearer hi-res would be in S-video on the SNES, since the colour and luminance are entirely separate signals in S-video, and CRTs that support S-video need to be able to tell the difference in order to benefit from the higher quality.rainwarrior wrote: ↑Sat May 14, 2022 11:34 pm For a related example... Apple IIe has an 80-column mode, basically a 7x8 pixel grid at 560px horizontal resolution. On a monochrome monitor it's perfectly clear. On composite, it's a lot more compromised. If you set text mode it will disable the colourburst, and the 80 columns are legible but very blurry. If you render the same pixels but in graphical mode, which leaves the colourburst active, the text is illegible due to colour artifacts.
In theory my TV could probably render much cleaner 80-column text in text mode (colourburst off), but I don't think the TV is built to be able to disable the composite filter... so monochrome can remove the colour artifacts, but it still bandlimits the luminance and it can't make the text any sharper. A true monochrome TV on the other hand doesn't have that filter, and would display it very clean.
So... in general I tend to think that almost all of the problem is in the composite signal process.
I found these comparisons of SNES Composite vs S-Video vs RGB:Nikku4211 wrote: ↑Sun May 15, 2022 11:00 amI actually wonder how much clearer hi-res would be in S-video on the SNES, since the colour and luminance are entirely separate signals in S-video, and CRTs that support S-video need to be able to tell the difference in order to benefit from the higher quality.
Remember that as nice as s-video is, because there's no longer chroma and luma cross talk, that chroma is still a lower resolution than luma. Luma or Y can be as high as you want, but chroma cannot.iNCEPTIONAL wrote: ↑Fri May 27, 2022 12:36 am I would be very interested in finding out how clear the difference is between normal-res and high-res (both pseudo high-res and actual high-res) when run through a decent connection like S-Video, Composite SCART, RGB SCART (particularly when compared to default Composite), also on modern TVs/monitors/displays too, in PAL vs NTSC, when viewed in 8:7 pixel perfect mode vs 4:3, and even on systems like the 3DS, Switch, SNES Mini and modern emulators, etc.
Well, I don't know what any of that means on a purely technical level or how it ultimately affects the end picture of the game on the screen I'm playing it on, but, personally speaking, I'm just talking about what I can see with my eyes (and similarly with anyone else making such a judgement if they're going to tell me about what they observed if they happen to run some tests themselves), as in, what visibly looks cleaner, clearer, sharper, has better visible colours, etc. I mean, other than some stats on paper telling me higher resolution is in fact an increase in resolution, or that S-Video and SCART are better than basic Composite, or some expert telling me the PAL signal is better than the NTSC signal, and those kinds of things, I'm ultimately judging all of this based on how it looks as I sit there playing or observing these games on whatever device and screen. And that's all I personally truly care about at the end of the day as the end user/gamer (even though I am working on making a SNES game). So, from my perspective, as I said, I'd be very interested in seeing what the difference is between normal-res and high-res in all the various examples/cases I mentioned above.turboxray wrote: ↑Fri May 27, 2022 8:28 amRemember that as nice as s-video is, because there's no longer chroma and luma cross talk, that chroma is still a lower resolution than luma. Luma or Y can be as high as you want, but chroma cannot.iNCEPTIONAL wrote: ↑Fri May 27, 2022 12:36 am I would be very interested in finding out how clear the difference is between normal-res and high-res (both pseudo high-res and actual high-res) when run through a decent connection like S-Video, Composite SCART, RGB SCART (particularly when compared to default Composite), also on modern TVs/monitors/displays too, in PAL vs NTSC, when viewed in 8:7 pixel perfect mode vs 4:3, and even on systems like the 3DS, Switch, SNES Mini and modern emulators, etc.
At some point I'm absolutely sure I will test some of those myself. I'm also happy to look at, read about and listen to other people's results if they perform the same tests, especially if their tests have already been performed and the results are already available to study. The vast majority of everything I ever learned in this life was from looking at, reading and listening to other people's previous examples, such as from text books or via Google searches or in a YouTube video, so I'm totally comfortable with checking such stuff out on the Internet for example, even if I didn't directly perform the experiment myself. In fact, that's literally one of the main reasons I come into this very forum, to see if I can find any prior examples posted in here, and if I can't, to enquire about them and see what comes from that. And I don't have to do it myself--I didn't have to go build a SNES myself just to learn how it works, yet I'm able to ask questions about how it works, look for information online, and indeed find answers--although I eventually will do at least some of the tests around how the different graphics of SNES games look on various setups first hand (you can be sure of that).