Moderator: Moderator Team
- Atari Super Hero
- Posts: 778
- Joined: Mon Jan 19, 2009 12:45 pm
- Location: The Netherlands
Something I look at the Indie games designers or on the pixelation site and see very nice
examples. The whole gfx seem to get to a higer level, but then again the whole retro machine
seems to be in the lift again. There is a certain learning curve at this moment that is very steep.
I also think that the tools we have today are better, look at the different drawing programs.
Or 3D sculpting stuff and convertors.
I for one love to see it happening, new gfx masters, super coders, stunning new music...
Some of the effects I now see are really amazing.
Sometimes I think, if this was the level 20 years ago, what gems would we have had
then in those days. But then again, we had loads of fun with what we had
( I have just to much Atari stuff)
The whole demo was very good. It was nice to see a graphic-centric demo on ST. They are quite rare nowadays.
Thanks for the positive comments on that demo, this means a lot ! All the graphics in this demo were made by Mic of Dune. The first slideshow features pictures which were done using Neochrome Master, using the multiple palette functionality. The second one uses a technique similar to Cyg's, with a STe-specific color switching method enablig to use 56 different colors in a single scanline, with uniform distribution of the color changes. I just released the code (source and binary) to convert images to a new format I designed for that purpose, including the file viewer on the ST(e), and the conversion code which can run on a PC, Mac or even an Atari. Everything is explained and available for download on my blog.
This is a very nice bit of work. And nice to see the explanation and code to go with it.zerkman wrote:I just released the code (source and binary) to convert images to a new format I designed for that purpose, including the file viewer on the ST(e), and the conversion code which can run on a PC, Mac or even an Atari. Everything is explained and available for download on my blog
The greedy search is a powerful way to do this and it's something I considered when I started looking at my own convertor again this year, but a few things stopped me short of trying it. I was worried that it could get trapped in some local minima after a random amount of time and it would be difficult to tell how much opportunity it missed - and require extra complexity (noise, momentum or something) to avoid or nudge it out of traps. It's also quite complex to implement without mistakes which hide well inside such algorithms and limit their performance in obscure ways. I had some experience with this kind of algorithm when optimizing geometry for games consoles 10+yrs ago.
So I spent my time trying to find a way that solves as simultaneously as possible, avoiding searches, forking patterns and perturbing costs around. The solution I came up with still uses heuristics and still becomes 'trapped' (each decision in the refinement limits future decisions, so the decisions have to be very well steered) - I don't think there is any complete / exhaustive solution to this so any method will be some sort of compromise.
Anyway your system has producing *really* great results - likely hard to beat. I'm still working on my solver now and again and improving it but it's hard to say if it will be ever better than this. Especially hard to tell what the gap is until I change the colour count per line and distribution which is fundamentally limiting the image quality.
I'm not sure how Cyg's reduction method works as he's been pretty quiet about that - I'd imagine it has to be something along the same lines as yours or maybe he's doing something completely different again.
AGT project https://bitbucket.org/d_m_l/agtools (source) https://bitbucket.org/d_m_l/agtools/downloads?tab=tags
BadMooD p/l: http://www.youtube.com/playlist?list=PL ... oOGiLtcniv
Quake II p/l: http://www.youtube.com/playlist?list=PL ... 5nMm10m0UM