What are you working on?

[Barnes]

uhh, John, it looks very cool

realtime lava surfaces lighting

[hogsy]

I still find odd that you get rid of quakec for a c-based gamecode though (imo quakec is the best thing of quake engine 1! )but, of course, you'll have your motivations!

http://oldtimes-software.com/blog/archives/269

Finally got round to finishing a post that covers our reasoning a little, or mainly mine. Feel free to give it a read

[toneddu2000]

@Irritant: I love the passion you're putting on CRX engine! Can't wait to see final results!

@Barnes: it seems a cool effect but maybe a video could show all of its beauty

@hogsy: very interesting lecture. I still find, but maybe because I'm just a noob at programming, that it doesn't seem to me that second block of code (the C-one) is unmessy compared to the first one. I still think that a clean and working code can be done even in quakec. With fteqcc, infact, you have most of the worktools of C (structs, arrays) but you don't have to face some obscure (at least for me ) treats like pointers!
But anyway, thanks for that article, really appreciated!

[Barnes]

@Barnes: it seems a cool effect but maybe a video could show all of its beauty

Ok
http://www.youtube.com/watch?v=Gjkny7aujzM

[toneddu2000]

Thanks Barnes! Parallax mapping is an awesome effect! What kind of performance drop do you have setting it on/off?
PS: I understood that lava content emits dynamic light! That's why I couldn't see any difference from the image you posted! But, what's the difference between content_lava and a wall, for example?

[Spiney]

Ooh nice, cone stepping, the only parallax mapping I actually like visually
How long does it take to preprocess the CSM?
I rendered out some in xNormal before, regular CSM was okay (still fairly slow) but the relaxed variant took forever to render.

[Barnes]

I spent a few hours in order to alter the height map in csm maps. As a result, the size of re-texture increased by 500 megabytes. If we compare the speed - csm is approximately two times faster than pom or relief mapping. But if you use optimized algorithms for relief and pom and then the speed is approximately equal. CSM slightly faster (5-10 fps) with optimized shader.

[toneddu2000]

As a result, the size of re-texture increased by 500 megabytes.

[Barnes]

csm map is rgba texture -
Red: inverted heightmap
Green: sqrt (Cone_Ratio)
Blue: df/dx
Alpha: df/dy
But shader use only RG channels

[Barnes]

fog lights (jail5 relight)

[Irritant]

Wow that is stunning!

[toneddu2000]

so beautiful!Compliments!

[hogsy]

I still think that a clean and working code can be done even in quakec.

Clean and functional code is definitely possible with QuakeC, I don't doubt that. I don't think there's anything necessarily wrong with QuakeC in its own right. But I think for us, we primarily felt there were more benefits in the long-term to write it out in C, as opposed to rewriting it out in QuakeC, purely because we didn't want to have to endure dealing with custom compilers that might run into issues down the line for us. Having rewritten it in C, we have everything and more we could want that custom compilers for QuakeC try to add and it takes away the need for us to do any potential work on the compiler ourselves. My article probably went too much into the personal feeling of it and less so the actual logical reasoning behind the decision, so I thought I'd just go over that real quick.

I think the best way to summarise is that switching to C was done because it meant less work for us in the future, rather than now. To be honest though, had I started on OpenKatana today, I wouldn't have opted to make that switch until OpenKatana had been finished because it essentially added an extra year to its development. It's why I emphasized my inexperience, back when we started the project, at the start of the article.

That article was pretty rushed since I've wanted to get it out the door for ages. I think the above would've been a much better way to explain things without sounding like I'm just trying to flak QuakeC for being an "inferior language".

[qbism]

bsp2 (and 2bsp) map support thanks to reading quakespasm code, plus bsp transparency from Makaqu. Possibly bugs, and only one transparency pass for now.
Download exe from the blawg http://super8.qbism.com/2015/01/bsp2-beta-release-humongous-map-support/

[ceriux]

that looks sexy.