View Full Version : LOD >500 Errors - A Possible Solution

September 22nd, 2015, 05:48 AM
Although the situation isn't quite as bad as it was I am still seeing cases where assets which should be passing the 500 poly test are being failed. Whilst I can now make sense of where the reported numbers in the error message are coming from I need to use a spreadsheet to do so.

Here is a proposed alteration that should be cheap to implement and a lot easier for content creators to get their head around, it may apply to traincars only - I haven't checked it through for mesh-table LOD:

Split the test into two passes.

First Pass
This looks at the main mesh only and checks it against the 500 poly limit. If there is a failure here the error will list a set of numbers which should coincide exactly with the polycounts reported by Max or AssetX. It will therefore be far easier for the creator to understand what the problem is and how to fix it. Most importantly there won't be any false positives here.

Second Pass
This looks at all of the meshes in the mesh-table EXCEPT the main mesh.
I believe that what happens at present is this:

Meshes with auto-create 0 or with no auto-create tag are OK
Meshes which are *.LM.txt are OK
This leaves meshes which are *.im with auto-create = 1, these are meshes which would ordinarily be visible at the lowest LOD.
If these meshes are < 300 polys there will be a warning that LM.txt should not be applied, hence they are OK.
If they are greater than > 300 polys then you will probably take the view that they are potentially inefficient.
If the meshes have a specified attachment point they may still be OK if they have been culled, either with a :cull suffix or by way of the attachment point being missing at this LOD.

You would need to go the whole way to avoid all false positives but we would probably be content with a warning here.

As far as I can see this would avoid any false-positive errors which exist at the moment and can prevent valid assets being installed. It would also be far easier to understand.