|
ForumsSega Master System / Mark III / Game GearSG-1000 / SC-3000 / SF-7000 / OMV |
Home - Forums - Games - Scans - Maps - Cheats - Credits Music - Videos - Development - Hacks - Translations - Homebrew |
Author | Message |
---|---|
|
Dithering on SMS
Posted: Sun Mar 12, 2023 12:24 am Last edited by mark_eire on Sun Mar 12, 2023 11:00 pm; edited 1 time in total |
In appreciation of dithering effects on SMS I wanted to note a list of some SMS games which make use of this effect.
Online discourse primarily notes use on the Megadrive, but SMS appears to use dithering quite extensively too! Please post with other game examples below. As I've struggled to find a list elsewhere I figured why not start here. Noted examples simplified into; "Blending" to create the appearance of additional colours, and "Transparency". Some SMS games with Dithering (example usually the first found in game but not necessarily the only or best one):
Notes on observing various dithering effects:
Additional notes:
As I only have an RF SMSII, I'm curious if others find Composite direct from SMS1 has the "blend" effect, or if the visible "checkerboard" appearance remains. The effect appears almost certainly lost in RGB. Is the Composite output of a modded SMSII the same as an SMS1 taken directly from the video port? Thanks for reading. |
|
|
Posted: Sun Mar 12, 2023 12:42 am |
On NTSC, the vertical purple and black lines in Shinobi stage 2 ("File No. 4523 BLACK TURTLE") do weird luma-chroma crosstalk effects. | |
|
dithering
Posted: Sun Mar 12, 2023 7:13 am
|
Just an example of dithering with 16 out of 64 colors at 256x192. That's probably at or near the best of what the SMS could display.
|
|
|
Posted: Sun Mar 12, 2023 7:44 am |
Except for the tile count. Random dithering sometimes looks worse on SMS than ordered. | |
|
Posted: Sun Mar 12, 2023 11:21 pm Last edited by mark_eire on Thu Mar 16, 2023 7:01 pm; edited 1 time in total |
That's a fantastic example of a complex video effect! Well beyond my simplistic categorisation of blending dot patterns and transparency. Do you know how Shinobi is intended to look? Did you understand this effect to be NTSC only, or had you just not seen PAL? I tested PAL RF SMSII today out of curiosity, while stationary I had rows of alternating Purple and a "new" lighter shade of purple between. Sort of "flickers" when I scroll the screen. Fascinating!
That's an impressive looking image, but as Maxim noted I don't think I could tile that up for SMS! Viewing on PC I can't even make sense of what all the dithering on the rear white panels is intended to do. Maximum dither image right there.
Do you mind explaining the difference there? Is that between clearly defined grids and more scattered patterns (sorta random) to blend colours? Like the difference between the dithering pattern in the sky and letters in the Lemmings title screen? Noticed in a thread from 2018 Flygon also made a note that "vertical dithering looks terrible on SMS Composite". I'd no appreciation the vertical and horizontal were handled so differently.... Complex topic. |
|
|
Posted: Mon Mar 13, 2023 7:53 am |
There are a lot of articles about dithering but I suggest to play with http://www.haroldo-ok.com/RgbQuant-SMS.js/RgbQuant-SMS.js/demo/index.html to try out lots of different algorithm options - although it seems to be down now.
Shinobi is likely supposed to look like the arcade game played on an arcade monitor with RGB, ie no artifacts. It seems the lines in question were subtle dark greys in the original, purple on black being an adaptation to the SMS palette. By their nature, video signals “blur” horizontally but not vertically. Hence the use of vertical lines to get dither/transparency in some systems, but this seems to work better on 320px wide video than 256px. |
|
|
Posted: Mon Mar 13, 2023 8:42 am |
The new URL is https://haroldo-ok.github.io/RgbQuant-SMS.js/RgbQuant-SMS.js/demo/index.html | |
|
Posted: Mon Mar 13, 2023 6:01 pm |
I don't know how it's intended to look. On NTSC there's a ... uh, a beat pattern? where we end up with vertical cords that move across the screen much faster than the background itself scrolls. Given that they're "just" houses, it feels to me like that's not intended. I hadn't seen PAL at all. A picture or very short video of the behavior on PAL would be awesome, if you would be willing. |
|
|
Posted: Tue Mar 14, 2023 10:31 pm |
Thanks very much Maxim, and sverx for the updated link. That website looks great, I think I saw it some time ago regarding tiles but had completely missed the dithering options.
I'm hoping my idea on what counts as random vs ordered dithering as illustrated by the Lemmings screen is sort of right then. If I ever get further along trying to make SMS art myself these are definitely aspects I'd like to play around with. I do wonder what the intentional look for Shinobi was. I couldn't find the clearest images of the arcade version, but I take your point Maxim, maybe it was a palette limitation. The PAL image does seem to create new colours to my eye. lidnariq; I tried but failed miserably to record video from the CRT that was in any way legible via digital camera or smartphone. Got a picture though and I can explain how PAL Shinobi looked in motion. Stationary, you've got a repeating pattern of: Black line; light Purple; Brigher Purple; light Purple again; Black line. There is a very subtle thinner vertical black line between the purple shades. Looking closely in motion, the vertical bars are actually fairly stable, but the visible lines all become the lighter purple shade. The Black lines become more defined and the previously barely noticeable line between shades of purple look the same thickness as all the other lines. Any sign of the periodic pattern is completely gone in motion, just repeating black, purple, black, etc. I wondered if there is any periodicity to the underlying image. Tried to look on Emulicious with PAL and 1:1 Pixel Aspect ratio. At first I thought perhaps, but I think it was more of an optical illusion! Maybe.... |
|
|
dithering
Posted: Thu Mar 16, 2023 12:52 am
|
Even reducing the tiles to the SMS's 448 maximum (16KB RAM minus the sprite table itself), we would still get some pretty impressive images.
Remember that this image only uses up 56 kilobits (7KB) out of a typical 2,048 kilobit (256KB) ROM cart for the SMS. Even using another 2KB of additional tiles for the car for color blending (like the Sherlock Holmes and other FMV games), we still have ample room left over and enough CPU power to generate somewhat realistic sound for an opening. There are plenty of nay sayers on this forum, but the SMS was capable of far more than what we saw back in the day. |
|
|
Posted: Thu Mar 16, 2023 3:44 am |
How's that custom FP and VDP hacking going for you? |
|
|
Posted: Thu Mar 16, 2023 2:03 pm |
We have no nay sayers that said the full power of the SMS has been already unleashed back in the day. We're already seeing amazing things now and we're going to see more. |
|
|
Posted: Thu Mar 16, 2023 3:40 pm |
The very post above yours. I keep trying to get onto the Genesis/Megadrive fora, but the admins are not responding to new requests. I may just have to create a new one. Bottom line is that the SMS was capable of digitized images and much better audio and 3D than we saw. |
|
|
Posted: Fri Mar 17, 2023 12:30 pm |
Indeed, and we’ll see them on actual hardware, not on a fantasy world version of it that only exists in someone’s head. |
|
|
Posted: Sat Apr 08, 2023 12:40 am |
Rainbow banding as caused by the shinobi level is owed to the way color is encoded into the NTSC standard. The long short of it is that the NTSC color standard is basically a forwards compatible hack of the black and white 60hz standard that existed before it. It's forwards compatible, in that it's built ontop of the black and white standard before it, with information that can be ignored to return to the original black and white image. Long story short, the color information is encoded into the original black and white signal and must be separated using fourier transformations to extract the color resolution image inside. The real secret is that luminescence and chrominance share a carrier wave in NTSC. Separating them is a lossy process, and the rainbow artifact is produced due to noise in the signal.
Any color space standard that separates luminescence and chrominance will not yield the rainbow effect. So, the same stage in Shinobi, when viewed with an RGB cable, will not exhibit the rainbow pattern. This also goes for the S-Video standard, and SECAM. |
|