PDA

View Full Version : Bogie LOD Problems



andi06
October 16th, 2015, 06:40 AM
This is a steam engine main bogie:

825

The image shows the motion at LOD 1, the red and white tripods are animated attachment points - the same wheel and axle mesh is attached four times at these locations.

Both the motion and the axles have their own two level LM.txt files with the RenderCutOff flag set.

Running this through 'Preview Asset' the polycounts at the preset distances are:
At zero metres: 6768
At 50 metres: 966
At 250 metres: 0 (In fact the entire bogie is cutting out at about 100 metres)

I would say that this asset has an effective LOD profile of 6768/966/0. But I cannot set trainz-build for this asset to 4.2 because validation ignores the final LOD of zero:

- VE109: The low-detail meshes total more than 500 polygons. This may have a negative impact on performance: 0: 6768, 1: 966

Your suggestions please Chris.

WindWalkr
October 16th, 2015, 08:41 AM
Validation hasn't been updated to use the new cutoff support yet. It will come.

chris

norfolksouthern37
October 16th, 2015, 10:48 AM
I have a similar situation I want to know about. It does not seem to involve any 'new' cutoff support as it has been in the lm format for a while just like the original post.

I have a mesh asset that is a radiator fan and it's mount. the fan is animated. I use this fan on almost every locomotive so it is useful to not have to add it to each and every mesh, it simply renders at an attach point as an attached effect and is controlled by the locomotive's script.

the mesh for the fan is around 506 polygons give or take for different fan configurations and i have it loading with an lm file with a renderCutOff flag set so simply not display it past a certain distance since it serves no visual purpose. I would think that the fan then has a LoD profile of 506/0 but i get two warnings:

! VE84: The file 'fan.lm' is provided in LM format despite containing only a single mesh. This may have a negative impact on performance.
! VE109: The low-detail meshes total more than 500 polygons. This may have a negative impact on performance: 0: 506

The same occurs on an end of train flashing device i have configured for some items. I need to display this item at close range, but once it reaches a certain distance it no longer matters. The items are too small to make lod models, they just need to be cut off. The referencing traincar says that the item should be .lm format since high poly .im files are not supported, but I would needlessly have to make another lod level in order to satisfy the 'lm format despite containing only a single mesh' and the 'low-detail meshes' problem.

These items will be at a build of 3.7

This brings up a further question - under lm, does render cutoff also cut the attachment points of that model or do they still exist even though the mesh is not rendered up to attachment cutoff value?

andi06
October 16th, 2015, 11:38 AM
It does not seem to involve any 'new' cutoff support as it has been in the lm format for a while just like the original post.
I think that Chris means 'new to TANE' since it was omitted from the original retail version and re-introduced with the earlier dev builds and HF2.


! VE84: The file 'fan.lm' is provided in LM format despite containing only a single mesh. This may have a negative impact on performance.
! VE109: The low-detail meshes total more than 500 polygons. This may have a negative impact on performance: 0: 506
If I understand correctly then both of these errors/warnings will diasappear as and when validation recognises that meshes with RenderCutOff specified in LM.txt have an additional 'virtual' LOD of zero polys.

Chris, perhaps you would clarify whether or not the projected updates will also take care of meshes which are suppressed by use of a.helper:Cull + AttachmentCutOff or by omission of a.helper from the lowest LOD of the parent mesh.

WindWalkr
October 16th, 2015, 08:30 PM
What andi said, mostly.



This brings up a further question - under lm, does render cutoff also cut the attachment points of that model or do they still exist even though the mesh is not rendered up to attachment cutoff value?

They should all cut off at that point; if it's not happening then I would consider it a bug.

chris

norfolksouthern37
October 17th, 2015, 08:38 AM
no problem. I didnt try it yet so I am just questioning the intended operation.

VinnyBarb
October 17th, 2015, 04:57 PM
That would indeed solve some of my issues with either fans and bogies. Good one Chris!

VinnyBarb

WindWalkr
October 17th, 2015, 08:30 PM
no problem. I didnt try it yet so I am just questioning the intended operation.

Yep. Some of this stuff is still being tested/fixed, so if you find anything odd, bring it up here and I'll confirm whether it's expected for some reason or whether it looks like a bug.

chris

whitepass
October 18th, 2015, 11:35 AM
Take a look at kuid2:58422:1035:1 it's old and I will be remaking it as I do not have the gmax files on my computer. Note: the car bogies are good just in the wrong place.

WindWalkr
October 18th, 2015, 07:51 PM
Take a look at kuid2:58422:1035:1 it's old and I will be remaking it as I do not have the gmax files on my computer. Note: the car bogies are good just in the wrong place.

Thanks. Interestingly, it appears that the bogeys are in the correct places; it's everything else which appears to be wrong.

chris

WindWalkr
October 18th, 2015, 08:27 PM
Okay, figured that one out. It appears to be an animated mesh being used wholly without animation. I've added software animation support to the loader to allow the bones to be applied before the (unanimated) data is sent to the GPU.

chris