Trees Of Daggerfall [PERFORMANCE PROBLEM/SOLUTIONS]

Discuss modding questions and implementation details.
User avatar
King of Worms
Posts: 2358
Joined: Mon Oct 17, 2016 11:18 pm
Location: Scourg Barrow (CZ)
Contact:

Re: Trees Of Daggerfall [PERFORMANCE PROBLEM/SOLUTIONS]

Post by King of Worms » Sat Sep 21, 2019 1:33 pm

Very nice, it will be CPU bound, so fingers crossed :)

User avatar
Hazelnut
Posts: 1585
Joined: Sat Aug 26, 2017 2:46 pm
Contact:

Re: Trees Of Daggerfall [PERFORMANCE PROBLEM/SOLUTIONS]

Post by Hazelnut » Sun Sep 22, 2019 4:36 pm

VMblast wrote:
Sat Sep 21, 2019 10:17 am
Update:
Ive added one more LOD group in between the highest and billboards, which makes transitions to be a bit smoother, so no more that much obvious snapping between LODs. Also Ive increased the resolution of the billboards, from previous 512x512 to 1024x1024, which gives a little bit clearer image of the trees of the billboards distance.

This decreased performance for just a little bit, from previous MOD iteration of just 1 lod and 1 billboard(512x512). Its not much, it is pretty much playable still, but it was the only way to balance all things together. Hopefully as @Hazelnut said, he could increase performance for just a little bit more from the present version. So finger crossed.
Sounds like good progress. I am surprised that adding the extra LOD level and upping the resolution of the billboards decreased performance noticeably though. Still that's speedtree and your department, I really don't know enough about it.

So the increase in performance I mentioned was an idea that I have for reducing the number of terrain chunks (i.e. 1 DF map pixel) that are rendered using speedtrees from 9 down to only 4. I was only really considering doing that if it turned out that 9 was simply too many for 60 fps on reasonable spec PCs, since it would be a lot more complex that what I've currently done and provided for you to test. (several times more complicated) Whereas the changes I've made already are pretty lightweight and minor so I'm very confident they will be accepted into the codebase. I don't mind coding it at all, it would be fun, but I never add needless complexity to code so will hold off for now.

Since I'm now back home (as of 2 hours ago) I'm about to tidy and submit the existing code changes to go into the next build to support your mod. I would like to wait a bit to see how they work for people once released, and also you've done a bit more of the natures in the mod. Then if it's still a bit too much of a performance hog for reasonable computers then I'll discuss implementing my ideas with Interkarma & TheLacus and see if they think it would be okay to add the required changes to terrain and mod systems.

I apologise if I gave the impression I could tweak things a bit to squeeze out more performance of your mod, unfortunately that's not the case.

Post Reply