Compute terrain patch size and TER resolution

[bigben]

I decided to run some tests on the effects of patch size in the computer terrain node to see if there is a usable relationship between patch size and TER resolution using the default fractal detail. I'd be interested to hear anyone else's thoughts on this. 

First test sequence is still rendering but it's quite clear that patch size is going to be an important setting for low resolution (e.g. SRTM3) terrains, firstly for the appearance of the fractal detail and even more importantly the results of subsequent displacements.  I have gone nuts at times trying to get good displacements on TER files (one of the main reasons you haven't seen many finished images from me) and it looks like I should have done this a long time ago.

[rcallicotte]

I've always wondered about this and the few tests I've done really didn't show me much, but I think that might have something to do with the fact I don't know what to look for.  But, nothing obvious.  Or is it?

[bigben]

It's obvious on a 90m resolution terrain.  Try patch sizes of 1,10, 45, 90 and 180 and compare.  At the moment it's looking like somewhere around 1/2 x terrain resolution, but I haven't looked at all of the images yet... or toggled smooth surface and repeated... Also using reasonable quality settings. Detail 1, AA 6, GI 2,4

[rcallicotte]

Okay, I might know what you're talking about, but what is a "90m resolution terrain"?  Obviously, it's 90 meters, but I don't understand the resolution part.

[bigben]

90m per pixel of the terrain ... so a 4097 x 4097 terrain would be 368730m wide. This is roughly the resolution of SRTM elevation data

or in TG0.9 terms, the terrain spacing value.

[gregsandor]

hey bigben,
Are patch sizes in meters?  If you set patch size to 10, then your features will be 10m, set to 100 then features are 100m, etc.?

[bigben]

hey bigben,
Are patch sizes in meters?  If you set patch size to 10, then your features will be 10m, set to 100 then features are 100m, etc.?

I'm working on that assumption. The developers seem to be sensible most of the time