pouët.net

smash: thats great news , I've got a 6950 here I'll give it a try on that too.

smash: any idea how that works on ati gpus? I have a radeon 2600 here (8xxx series equivalent) and a 4xxx series ati at work.

smash: ye. what about pixels that are not preculled and are marched behind the actual framebuffers depth. those are the ones with lots of potential to waste performance due to double calculation. can't help it tho.

sort front to back :)

I thought that's what the z-Buffer's for ;)

The true reason for having z-buffers is to prevent heavy pixel shaders running too many times rather than actually sorting anything!

I was joking.

I know.

For real life speed test you can try MF Real.

Z-writing hurts only if your scene is dependent of early z-culling.

I just stumbled upon this one: http://nippe.teknopapu.net/index.php/archives/124#more-124
Thanks for the "inspiration: pouet" credits. *cough*
That's at least 85% the code I posted some pages ago.

In addition to that - I feel kinda elite that people started ripping off some of my really crappy code.

Well, he does pretty much admit that he's copied the code and fiddled with it until it works. And from his description, I'm not sure he understands what raymarching is :D

I need to get back onto my raytracing stuff. I have some fresh ideas that need trying out. But, too busy with iOS stuff.. for the past couple of weeks I've been chasing ultimate performance for image processing, figuring out obscure image formats and such :/ Complexity goes up with speed it seems. It's teaching me a lot though, I think I'll have some new ideas for optimising raymarching from this :)

to all : some of my raymarching scripts (including the one with reflected cubes i posted some weeks ago) can be downloaded here :
http://www.advsys.net/ken/polydraw.zip (check tigrou folder)

I try to keep it simple, hope they can be usefull for raymarching beginners :D

That fireball is so bad :|
I couldn't even move the noise function.

Contact him and see what happens, we love a good flame.

Ahem, I see you guys noticed my webgl experiment with crappy code "stolen" from this thread before I had a chance to stop by and give the link to you myself. Psonice, you're right, I don't fully understand raymarching yet but one can always learn, right?

Oh and flame on if you feel like it.

I'm already done with flaming you - you just owe me some beers and everything is fine. :D

Some guys here are faster then you think ;)

@las: "then you think"? Apparently you are too fast to write properly - slow down and then get your then/than's right.

las, sure, I can buy you beer if we happen to meet at parties. :D Anyway, I'll go back to my corner try to learn more about raymarching.

Lord Graga. Sir, Yes sir! You are right of course.
If you prefer german, just tell me - I prefer to write pouetish here - and it seems that you perfectly understood what I wanted to express.

eimink: Go ahead - start being awesome. :)

u iz puting teh bar 2 low d00d.

FART

you will never understand raymarching. It's a special kind of black magic invented by a few chosen coders from the demoscene. Some day, the game industry will catch up and they will be in endless moral debt.

hyde: damn right. Ever seen rasterbars in a PC game? 20 years, and they still can't do it!

eimink: damn right, but instead of taunting us and running off, why not join us here? Ask some questions, post up your experiments, flame everyone, and agree with hyde because he's always right (and if he isn't, he has a belt ready).

I recommend a simple challenge to start off, like raymarching a sphere without any copied code.

how do you for without for'ing?

Note: while's don't usually work on shaders.

goto!