Wow 2.5 rocks!

(BoltyBoy) #1

I have to say, what ever Ydnar has fixed in 2.5 q3map2 it has made some dramatic changes to light compile times.

Currently my near completion map has gone from around 30 hours with -fast - filter -super 2 -bounce 5 to, get this… 2 hours !!! Yes I was also totally in disbelief! That’s like … ah man …like …wow…i mean totally…amazing!

What was the bug? A missing semi-colon? hehe

I have found the lighting now needs a little extra loving care and attention as 2.5.3 seems to highlight bad lighting more so than before, but nothing that I’d classify as a decreased quality.

I’d post pics to compare but don’t think I could face the old 30 hour compile!

(hummer) #2

I must be doing something wrong then… 2.5.3 takes much longer + eats way more system resources that 2.3.38, which I’m still using…

I only have 128 megs of ram though… Hard drive chugs away on VRAM…

Lighting takes a few hours on 2.5.3 rather than a few minutes… that and some shadows don’t show up properly :confused:

(ydnar) #3

2.5 is more memory hungry than previous versions. The tracing code is all new, and that’s basically how it goes. RAM is cheap… :slight_smile:

As for incorrect/missing shadows, do you have screenshots?

y

(hummer) #4

/me kicks his grad student budget and $30,000 debt…

Maybe I’ll use that Fry’s gift certificate and pony up for some RAM…

I’ll work up some screen shots… I deleted 2.5.3, but I’ll reinstall it.

(ydnar) #5

Ah, stay away from Fry’s. They hardly have the best merchandise or prices. Try here instead…

y

(hummer) #6

Suno f a biznitch… had it all written then clicked “back” when previewing :confused:

Ugh… anyway… lets try again…

1. Wide shot of problem area
2. Close up of problem area… a inverted beveled-bevel for sidewalk corner meets with regular brush
3. Same
4. same, different area
5. same, different area
6. Close up of darker area… curve and non-curve meet, and shadow is misaligned…

----- 2.3.38 below, same compile options------

1. Light / no shadow
2. light shadow, not choppy
3. Wide shot of similar area
4. Close up of above
5. Darker area, ok
6. Shadows slightly off, but hardly noticeable
7. Same situtation as 6 from 2.5.3… but shadows line up…

So, whats going on? Some compile option I’m leaving out in 2.5.3?

(ydnar) #7

I assume the patch meshes on the corners are in func_group entities. There was a bug in 2.3.x that allowed lightmap merging across entity boundaries. The behavior was corrected in 2.5.

Ungroup the func_group entities and the lightmap across that sidewalk will be merged properly.

Couple things:

  • Lose the high ambient light value. Use radiosity or q3map_skylight for nicer uniform sky lighting.

  • Use -patchshadows if you aren’t already.

y

(hummer) #8

A bug? Seems more like a feature than a bug :wink:

Anyway, I’ll give radiosity or q3map_skylight a go… I had my compile options listed in my first post attempt, but forgot to replace them :confused: I was using -patchshadows for these compiles…

Thanks :slight_smile:

(ydnar) #9

Trust me, it’s a bug.

A bug you’ll be glad is gone when you realize the shadow detail on the sidewalk in your map is so low because the 128x128 lightmap is stretched across all of the contiguous brush faces–when you realize all you have to do is break one or more sections into its own func_group entity and have shadow detail reappear.

y

(BoltyBoy) #10

I’ll second that motion. The shadows on my coniguous faces were being stretched over a length of 3000 units or something silly like that.

(Davros) #11

ill second that :clap: , huge terrain with caves, trees and buildings all in under 10 minutes. thankyou yndar :chef: