[Genma Saotome]

Somewhere between versions V2150 and the latest something changed in the display functions with the result that some objects are being displaced from their proper location, others will display -- or not display -- based on camera position, and a whole lot of jumpiness is observed.

I don't have a more recent version that V2150 to try and narrow things down and this is a route that has sat unused for quite some time. Here is a screenshot from that version, everything looks as it should:





Here is the same scene as seen in V2669:





Just to be on the safe side, I went in the route editor and verified nothing had been moved:




There is a second, perhaps related problem - depending on camera angle most of polys in these shapes do not display at all. What I see:




Move the camera slightly and I see this:





If any of the programers want me to to check other versions I'm willing... perhaps using dropbox to send me a archive of one or more versions.

[markus_GE]

Dave,

you can always go to Jame's site and get whatever build you want :)

Cheers, Markus

[charland]

I think this is a superelevation thing, the terrain creeping through the ballast. I turned mine down to level 1 (it wont go down to 0), and changed the radius to 500 and it got rid of most of this. Maybe changing the radius to 2000 or something will tell you if that's the problem or not... if the problem goes away, that's what's causing it.

Paul :-)

[That Genset Foamer]

Could be a bug in the instancing function. Are those rip-rap shapes all the same file?

[Genma Saotome]

I do not use superelevation and yes, there are many identical shapes so it could be related to instancing.

I took Marcus' suggestion and downloaded earlier versions to see if I could identify the specific version where this started -- I could not: The earliest version on James' website is V2231 (I downloaded 2132). It has the same problem:




While taking that screenshot I noticed something else: all of these shapes are "jumpy"... bouncing up and down a bit, sometimes side to side like you see here. The foreground most shape (cropped out) acted differently: Starting from a low angle, as I raised the camera the rip rap shape in the immediate foreground rose right with the camera.

=================================

Having done all that the light bulb finally went on and I returned to the latest version, turned off instancing, and ran the same activity. No jumpiness, to misplaced objects, all polys properly displayed.

So there is some sort of problem with instancing. Perhaps it is due to the large number of identical objects placed in the scene... I have a handful of rip rap shapes to use, they get placed in random order to minimize the visual repetition that would occur if only one shape is used, but even with a handful of different shapes there are going to be many instances of each present in this scene. Might there be a display problem on account of that?

[James Ross]

 Genma Saotome, on 26 November 2014 - 10:58 AM, said:

I took Marcus' suggestion and downloaded earlier versions to see if I could identify the specific version where this started -- I could not: The earliest version on James' website is V2231 (I downloaded 2132). It has the same problem:

I archive the builds periodically but I have them going back to at least X1768 which you can always ask for. I'll see if I can narrow down when this started.

[James Ross]

Okay, so this is definitely down to some limit/wrapping in instancing but I haven't yet figured out what determines the limit. For the rip rap shown in this thread, it seems to be 47 instances are able to be grouped together without issues but 48 will cause problems. I've got the code to split instances up in to chunks/groups working but I need to figure out what determines the limit as I suspect strongly that it depends on the object itself (e.g. number of verticies).

[Genma Saotome]

 James Ross, on 27 November 2014 - 03:50 PM, said:

...as I suspect strongly that it depends on the object itself (e.g. number of verticies).

Dunno if this will help but here is the route-wide count of riprap shapes and how many polys are in each shape:

Filename	Count	Polys
BigRipRap	70	306
BigRipRap02	22	288
BigRipRap04	273	1260
BigRipRap05	25	764
BigRipRap06	123	1587
BigRipRap12	151	1184
BigRipRap16	133	1294
BigRipRap24	95	1448
RipRap  	135	460

Most of the placed shapes occur on either side of the view shown in the screenshots and so many could be right around the next bend, just out of sight but well within short viewing distances... certainly within long viewing distances. There are no LOD's and with them being TSM files instead of SU files there is no way I'm going to try to add LOD's.

[JohnnyS]

 Genma Saotome, on 29 November 2014 - 04:20 PM, said:

...and with them being TSM files instead of SU files...

Hi,

I have a hunch that this may be a TSM problem rather than an OR problem, in the past I had TSM built objects display such behaviour (displaced and misshapen polygons). After a long period of head scratching I chanced upon a forum post that suggested exporting a TSM model whilst in point mode or polygon mode was a no-no and would result in wierdness. Is it possible to export the models once again but this time ensuring you are in part mode and to see if this makes any difference?

Cheers,
John.

[James Ross]

 JohnnyS, on 30 November 2014 - 02:27 AM, said:

I have a hunch that this may be a TSM problem rather than an OR problem, in the past I had TSM built objects display such behaviour (displaced and misshapen polygons).

There is a chance that could be the case, but since the problem stops occurring with instancing turned off or when I limit instance groups to smaller numbers of objects, I am pretty sure there is something wrong beyond the object.