#Infinite compile problem (optimization)

1 messages · Page 1 of 1 (latest)

obsidian oasis
#

Hey !

I'm encountering a problem when compiling a map with hammer.
In fact, I can well compile the map as "fast", but when I try as "default", that doesn't work : the compile time seems to be infinite.
In fact, the compile log file didn't get edited since 4 hours

Compile log : https://hastebin.com/lihozukexi.http

Does anyone have any idea please ?

Can this problem be caused by too much func_details ?

Thanks

lyric lagoon
#

Maybe you have messed up the compile parameters

prime basalt
#

What are your compile settings?

#

Is this your map?

#

If you are editing the map, it must be renamed as follows: rp_truenorth_v1a_"Your Map Name Here". It cannot be named v2, v3, etc, or anything other than Truenorth.

turbid cipher
#

its just slow portalflow

#

because look at that, 24k numportals

obsidian oasis
obsidian oasis
prime basalt
#

I was asking if this is a decompiled map or not your own?

obsidian oasis
#

so original one yes

prime basalt
#

Ah, my apolgies then. Yeah, the numportals are way way too high

obsidian oasis
#

How can I reduce numportals ? with func_details ?

prime basalt
#

yeah, or propper to turn them into models

#

Although wallworm for 3DSmax or Blender Plummer would be better

obsidian oasis
obsidian oasis
prime basalt
#

https://dev.wallworm.com/

Wall Worm

Wall Worm tools, utilities, plugins and scripts for creating Source Game Engine Assets inside 3ds Max. Documentation, tutorials, tips, videos and more.

obsidian oasis
prime basalt
#

There is also this for blender

prime basalt
#

Making them into models will also decrease

obsidian oasis
obsidian oasis
prime basalt
#

Blender is a third party tool that is free to use

#

The best way to make models is to either use propper, or better, import the vmf into blender using plugins and then exporting that vmf as a model or two.

rain sparrowBOT
prime basalt
#

Thanks Ido

prime basalt
obsidian oasis
prime basalt
#

Using the -notjunc option will skip this optimization at the price of possible visual inconsistencies.

obsidian oasis
#

oh ok I see

obsidian oasis
#

and, how many numportals should I target as an objective ?

turbid cipher
#

so, compiling (portalflow) is more or less logarithmic

#

i.e. 9... is going to be the longest step, and could take upwards of over twice as long as it took to even get to 9... in the first place

#

its not impossible to compile with 24k numportals but its probably not going to be anything that can be completed in a remotely timely manor

#

the main problem with gmod in particular is that its vbsp doesn't support any commands that would make this easier to do (i.e. -autoviscluster, available in CSGO)

#

so you are more or less going to have to do a deep analysis and figure out where you can start shedding off numportals (or at least optimize them away with a viscluster)

#

NOTE: do not just go around placing viscluster randomly, it will break optimization if you do (and other things like water and such)

obsidian oasis
#

Hmmmm I see...

obsidian oasis
#

because it seems that the map already has a lot of func_detail

#

so I think that I can't really add more func_detail

turbid cipher
#

well no

#

I'm just saying this is kind of a hard place since you are using gmod

obsidian oasis
#

hmmm

obsidian oasis
turbid cipher
#

impossible to say really

#

the correct answer is "as many as you need"

obsidian oasis
#

ah 😅

#

there is no difference between func_details and props in terms of numportals right ?

turbid cipher
#

no

#

if you want to get to optimizing the number you need to really learn how they actually work/how to optimize them

#

tophattwaffle's optimization videos for instance

obsidian oasis
turbid cipher
somber perch
#

IDK if it is an issue like it was in GoldSrc, but also make sure a face does not have an infinitely stretching texture. I think that cause similar problems for me.

turbid cipher
#

its not

somber perch
obsidian oasis