not the same old max_map_draw_indexes problem

(rgoer) #1

A jk2 mapping buddy of mine, Wedge, has run into a weird max_map_draw_indexes situation. He’s been working on a map that is almost entirely constructed out of patch meshes, and it’s been compiling just fine up until a few days ago. He’s using q3map2 2.5.5 (not sure which version of 2.5.5, though), for the record.

Anyway, he started messing with -patchmeta to maybe get the fps down, and now all hell breaks loose when he compiles. It crashes during BSP, giving him the old max_map_draw_indexes routine–but he hasn’t added any new geometry to the map since his last working compile. I was under the impressions that -meta and -patchmeta decreased the number of triangles in a BSP–so why would he hit the max_map_draw_indexes limit with -patchmeta if he wasn’t before?

I sent him the link for 2.5.5-test-13 and told him to recompile with verbose mode on and to log the BSP section, in case there’s any helpful info in there… but has anybody got any idea why this kind of thing might happen?

(ydnar) #2

It’s actually pretty simple.

Every triangle in a mesh (brush face, meta, terrain, misc_model, or patchmeta) requires 3 indexes.

Using -patchmeta basically hardwires all the patch meshes into triangle meshes, taking up index space.

Your friend’s map must be pretty heinous to be hitting this limit. I can’t say I’ve ever heard of anyone hitting it before.

y

(rgoer) #3

Hey, thanks Y. Here’s how his map looks, if you’re curious:

(UniKorn) #4

Nice map, but nothing but patches :slight_smile: He can try to make his patches less complicated (for instances 5x5 instead of 7x7 etcetera) That way he’ll use less triangles => less index space.

(Emon) #5

He could do it without -patchmeta, add -subdivisions 16 and make it an ASE. I did that for a friend’s model he made in Radiant.