pouët.net

Yeah, next year .. we're going to be totally awesome!

yeah, next year for me too ;D

of course. that's why i mentioned the consolebuffer restrictions (which is basically saying the same thing).

trc_wrm, fragment & preacher: I've read this one so many times.. I better have an incredibly awesome top three next year! :D

Deal :)

Yup, deal :) But no one said fragment, Preacher and I aren't collaborating.

BTW, any pointers on ways to generate good ANSI subsets for RGB color space mapping? I've already read Sol's TextFX stuff, but I'm wondering if there are approaches other than Euclidean distance / least-squares mapping that look better. I run into problems, for instance, when doing fade-in/fade-outs: I get some aweful color changes.

I think it is better not to do a direct RGB mapping at all. At least I've gotten much cleaner results by mapping the effects to a few handmade gradients. Textures and effects are calculated "grayscale" and then colorized. Also the amount of different characters used is very small.

Try different color spaces and different distances (libcaca uses HSL with a wacky distance function for one). But frankly you do not want accurate distance based fitting, otherwise most everything would have ugly grays or other unpleasant color mixes. Better to tweak distance function for what looks better to you. Or you might even setup you're color mapping table by hand, might as well use a smaller by nicer mapping. I guess the goal would be to to have transitions from one solid color to another be in a the most linear progression in character density.

What's this crap I hear about gradients in textmode, the 16 available colors are more than enough ;)

@fragment,paulius: Looks like I need to experiment a lot more :)
@britelite: That also deserves consideration, but I suck at direct ANSI art..

you suck at art in general, niels ;)

Hmmm, I think you're right!

also, you cant count fingers, unless you have a freaky thumb..

:D

nice art :)

but on the color space I did both research and experimentation for my converter. Tried HSL / HSV and RGB distance and other stuff... HSL is much worst tan it would seems (unless you bake in some crazy had-tweaked bias).

Finally I stumbled on this very useful piece of info, which I found very useful. It's well explained, concise enough, and you even get some code!

Very interesting link BarZoule!

Here is a comparison I made with a palette based on the euclidian distance metric and the optimized distance metric:



The optimized metric has somewhat better purple/blue handling. I still get some odd colours though..

For comparison, here is the original:

Nice tests!
I'm intrigued as o what libCaca would do with those pics.
From my experience libCaca has very low fidelity but looks good nonetheless. It's generally much more luminous and saturated.

libCaCa rulez! Sol is the one Man when it comes to TextMos!
pure love for that dude! will use your stuff next year to show off what a Console is able to do!

Right. So, where are the results?

Next year?

I'm more interested to see the prods; not necessary to give the results yet.

Would it be ok to release them today after the deadline? Or should we wait the results? I could release mine anytime..

@optimus: I'm pretty sure all of us could. Still, it would be nice to have the results too..

So what are you waiting for? : )