Flashing speedtrees and lightning

[hutten]

Hello,

Currently I am running TRS22 build 123794. I do experience some speedtrees flashing at some distance (approx. 500 m, so probably one of the LOD levels) and "lightning" effects around them. These at speedtrees by author RMM. The effect ceases when the distance to the trees decreases.
I have seen the same in a previous build.
My setting are all at the maximum. The video card is RTX 3070.
Anyone?

Paul

 

[JCitron]

Hello,

Currently I am running TRS22 build 123794. I do experience some speedtrees flashing at some distance (approx. 500 m, so probably one of the LOD levels) and "lightning" effects around them. These at speedtrees by author RMM. The effect ceases when the distance to the trees decreases.
I have seen the same in a previous build.
My setting are all at the maximum. The video card is RTX 3070.
Anyone?

Paul

There's another discussion regarding the same issue. We were told by N3V Dev team that it's a graphics driver issue but I've seen it with the latest drivers and the problem occurred in TRS19 as well.

 

[hutten]

Hello John,

Thank you. I am glad that I am not alone here.
I do not see the effect in TRS19 build 117009.

I just found the thread where this was discussed recently. The thread was started by ElStoko.

Paul

 

[HPL]

There's another discussion regarding the same issue. We were told by N3V Dev team that it's a graphics driver issue but I've seen it with the latest drivers and the problem occurred in TRS19 as well.

Like said in the other thread, if this issue exists in 123794 and not in 122411 (neither in 117009), it is NOT related to the video driver.

 

[hutten]

If the latest version(s) of the game use(s) other features of the graphics system, that statement is not necessarily true, I think. That is just a general way of thinking, I haven't the slightest idea of what features could be involved.

Paul

Edit: I just now installed the latest "studio" version of NVIDIA's RTX driver, 546.01. I see no more white flashes in Kickstarter Country 2 in build 123794. I did not test de "game ready" version of the driver.

 

[JCitron]

If the latest version(s) of the game use(s) other features of the graphics system, that statement is not necessarily true, I think. That is just a general way of thinking, I haven't the slightest idea of what features could be involved.

Paul

Edit: I just now installed the latest "studio" version of NVIDIA's RTX driver, 546.01. I see no more white flashes in Kickstarter Country 2 in build 123794. I did not test de "game ready" version of the driver.

That's good to hear. There are features and bug-fixes in the drivers not related to things that N3V uses. Reverting back to an earlier driver is not an option for some if they have a newer video card that wasn't out when the earlier driver is released.

 

[JCitron]

Like said in the other thread, if this issue exists in 123794 and not in 122411 (neither in 117009), it is NOT related to the video driver.

I said that before.

 

[hutten]

Hello,

That is just what I do not agree with, John. The driver update has resolved the problem for me. It shows that build 123794 and build 122411 use the graphics system differently. That is my interpretation. It IS dependent on the driver.
How else would you explain the result?

Paul

 

[wreeder]

Hello,

That is just what I do not agree with, John. The driver update has resolved the problem for me. It shows that build 123794 and build 122411 use the graphics system differently. That is my interpretation. It IS dependent on the driver.
How else would you explain the result?

Paul

Wait and see if it stays resolved. It has been known to return after appearing to work for a while. I agree with John. I find it almost impossible to be that a driver can cause just some trees to have this problem with level 2 LODs and not effect anything else. I think it far more likely that there is a problem with the level 2 LOD meshes and or the textures used. I plan on trying to remove the level 2 LOD from the config of the trees to test my idea. Plus I'm using Win 7 with a much older driver that had no problems until the first TLR beta build of Trainz Plus. I did get the newest driver for Win 7 to install but then Trainz Plus would fail to start with failure to initialize the renderer error. I check the driver and it appeared to be a beta to GPU Z. But there is another one out now. I'm going to clone my boot drive tomorrow and then update this PC to Win 10 so we will see what happens.

 

[JCitron]

Hello,

That is just what I do not agree with, John. The driver update has resolved the problem for me. It shows that build 123794 and build 122411 use the graphics system differently. That is my interpretation. It IS dependent on the driver.
How else would you explain the result?

Paul

Yes and no. The drivers may have "fixed" the problem this time but what's going to occur next time? There's still an issue with those particular assets that are causing the problem. Why is this only certain trees by a certain content-creator? I would've looked there first but that's too "hard". I would say definitely a driver issue if this issue occurred across all Speed Trees or other kinds of assets, but I can't put this 100% down to a driver causing the problem.

Speed Trees make hard use of the GPU instead of the CPU. The assets are converted from meshes to binary code and interpreted directly by the GPU. This saves a lot of time rendering but does put more pressure on the video card to process the data.

If something is amiss with these assets, then they're going to poke something incorrectly and have an issue. The driver update may have changed how the data is written at least for now, but I don't consider the issue resolved because the issue occurred in previous versions of TRS19 and TRS22 and then it disappeared and reappeared again. I saw this in the early days of the Multiplayer Surveyor. We were told to ignore the problem at that time by the Dev Team. Hmm... They knew something was wrong then.

 

[hutten]

Hello John and wreeder,

Agreed, the assets may have unwanted properties. And yes, the problem may pop up again. Yet, I will go back (or rather, forward) to 123794 again and check more routes.

Paul

 

[wreeder]

Well, I took a look at one of the trees' config.txt:

kuid                                    <kuid2:211025:21199:1>
kind                                    "scenery"
username                                "st_rmm_yasen17_1987_18m"
visibility-radius                       5
trainz-build                            4.2
category-region                         "RU"
category-era                            "1990s;2000s;2010s"
category-class                          "FT"

thumbnails
{
  0
  {
    image                               "stlogo.jpg"
    width                               240
    height                              180
  }
}

mesh-table
{
  default
  {
    mesh                                "yasen17.srt"
    custom-render-plugin                <kuid:401543:1077>
    auto-create                         1
    mesh-scale                          0.35,0.4
  }
}
contact-email                           "RMM1@xxxxxxx"
author                                  "RMM"
license                                 "
     This asset was made by RMM.
     This is for personal use only not for sale nor redistribution only via N3V DLS!"

kuid-table
{
  0                                     <kuid:401543:1077>
}

I assume that the LODs are being produced by the speedtree library. However, I don't know how speedtrees work so I could be wrong. So is this a bug in the speedtree library or in the way that the E2 engine handles the output of the speedtree library?

 

[JCitron]

The trees are created in a Speed Tree development package and then they are compiled into binary libraries. I think the content-creator still needs to create the textures he wants to use and they are then referenced just like any other asset. If the textures have a faulty UVW map or missing a texture then we end up with the white flashes, according to my brother who creates models and artwork for a living.

 

[wreeder]

The trees are created in a Speed Tree development package and then they are compiled into binary libraries. I think the content-creator still needs to create the textures he wants to use and they are then referenced just like any other asset. If the textures have a faulty UVW map or missing a texture then we end up with the white flashes, according to my brother who creates models and artwork for a living.

So we are right, it is a faulty asset and NOT a driver issue.

 

[JCitron]

Yup. Why the latest driver works now beats me, but my warm and fuzzies tell me it'll reappear again at another date.

 

[wreeder]

Well, we know that the Speedtree library hasn't changed as it is obsolete and no longer being developed. The assets haven't changed unless N3V made changes to them. So the logical conclusion is that the game engine has had trouble dealing with rendering the trees that are produced by the Speedtree library several times in the past and managed to fix them. My guess it that this is related to attempting to implement DX12.

 

[Driver_Col]

There are a number of threads re the flashing/sparkling trees. and as I commented on 2 similar threads, I went direct to nVidia to check the status of my GTX1060 and the Driver was indeed out of date. (I mistakenly assumed that Windows Update covered all that stuff!). After the Driver had been updated, my trees once again looked normal! Regards. Colin.

 

[kwdubf]

You guys make me feel like a "babe" with all your technical prowess. I guess I had never heard of Speedtree before but I'm assuming that it has something to do with certain trees being dynamic instead of static? Can this be fixed easily. Remember, I don't have the computer knowledge of most of you. Or, is this just a matter of updating my "drivers" on my gaming PC. Any help or suggestions would be greatly appreciated. Ken

 

[JCitron]

You guys make me feel like a "babe" with all your technical prowess. I guess I had never heard of Speedtree before but I'm assuming that it has something to do with certain trees being dynamic instead of static? Can this be fixed easily. Remember, I don't have the computer knowledge of most of you. Or, is this just a matter of updating my "drivers" on my gaming PC. Any help or suggestions would be greatly appreciated. Ken

Update your video driver to the latest version. Apparently, that's the fix for this ongoing issue that has disappeared and reappeared off and on over the past versions.

Speed Trees are a technology that uses video cards to generate the trees based on specific libraries of tree parts that the content-creator used to make them. The older trees, also called billboard trees, use flat plains arranged in a star-shape with pictures of tree branches on them similar to those old-fashioned architectural models where the architect shows landscaping around a new building model that's usually placed under glass for everyone to see. The problem with these billboards is they can be very CPU intensive due to how they need to go from meshes, to meshes with images. In the past, this used to cause stuttering and other performance issues depending upon the complexity of the models. The issue we have now has to do with how the images are placed on the meshes due to changes in the rendering technology that cause them to look more like paper or cardboard instead of clear Mylar with tree branches on them.

With Speed Trees, as I said the trees are processed by a much faster video card and can actually look quite nice. They aren't perfect, well too perfect is a better way to say it, since they are computer generated rather than pictures of real trees. Since they are generated in the video card, there's less overhead for the CPU to handle other aspects for the computer to operate and also since they are generated in the video card, they render quicker due to being already processed and ready to display. Like anything else, they too can cause a performance hit if there are too many of them. As we say with Trainz, there can be too much of a good thing that can impact the performance of any route.