Distant feature billboard size estimation

[Beherith]

How does the engine do it? Because the distant billboards seem to be half the size of the features they replace at large distances.

[jK]

-_-
You are the 2nd one ...
It uses the S3O model-radius set in UpSpring.
Seems like modelers rarely set it to decent values ...

It's really time to display in debug mode feature collision volumes, too.

[bobthedinosaur]

So does that mean we can make a large hitsphere in s3o, and it would go off of that sphere size even if the unitlua shrank the sphere?

[Pxtl]

There are legitimate reasons for making the billboard radius not match the collision radius, like if you have some non-colliding extremities on the model.

[Beherith]

Ok thanks. Seems like ill have to switch to TDF defined hit volumes then. Because putting a tall tree in a large sphere screws gameplay, while having it in a smaller, sensible hitsphere makes for ugly ass distant billboards that you have no clue how they got so small.

Man every time I learn something new about the engine, I have to remake those 150+ features.

On a related note, does anyone have any sort of python or other scripted tool to read s3o-s?

[Beherith]

Also, will switching to "box" hitvolumes on 1k+ features per map be much slower than using hitspheres?

Burning features also use hit sphere radius for flame radius, which makes them look off.

[Argh]

will switching to "box" hitvolumes on 1k+ features per map be much slower than using hitspheres?

Somewhat, yes.