[gpz]

For those would like to check how some glTF models look like in Openrails, it is possible to display the official test models easily in the unstable version by:

• git clone https://github.com/KhronosGroup/glTF-Sample-Models.git

  into the TRAINS/TRAINSET/glTF-Sample-Models/ folder
  
• Add VS debug options command line parameters or run the RunActivity.exe like: -start -explorer "C:\Devel\MSTS\ROUTES\USA2\PATHS\tut6path.pat" "glTF-Sample-Models" 12:00 1 0
  
• When running this test, the animations can be switched on-off by [V], [Shift+,] and [Shift+.], go to the next/previous model with [Alt+PgUp], [Alt+PgDn].

It is just a test to ease the development, but also it may be a fun for some... It is like a toy store.

[ErickC]

gpz, on 05 February 2023 - 12:32 PM, said:

For those would like to check how some glTF models look like in Openrails, it is possible to display the official test models easily in the unstable version by:

• git clone https://github.com/KhronosGroup/glTF-Sample-Models.git

  into the TRAINS/TRAINSET/glTF-Sample-Models/ folder

I will be honest, I don't understand what that any of that means so I'm not going to even try to install them. But the models look impressive so far! I have a whole bunch of rolling stock that I'd be more than willing to donate for testing and development, if you're interested. I can send pretty much anything over in 3DS or FBX format, including a 70-tonner with a 3D cab.

[gpz]

Erick, thank you very much for your offering. Originally I didn't plan to deepdive into the exporters spiritual world, not even talking about the 3dsmax itself. If you feel like you would like to see how your models perform as gltf's by trying to export them to this format, I would gladly participate in a testing session, like I do with Robert. But I would rather stay at the coding side for now. I have quite a lot to do. For example currently I am rewriting the binary loader of the gltf's, because I am not satisfied with how I coded it originally. I try to modify the code to shovel all the binary data directly over to the GPU at load time, without any further midtime processing, and set up the buffer structures (ibo/vbo) accordingly later. Actually this is how it supposed to work in the first place, but it would have been hard to do this all together with all the other modifications, so I've chosen an easier path. (With this modification loading a gltf will be much faster than loading an s file.) When I am finished with this, and make all the necessary visual tests pass (the models available with the git clone above), I will ask James to review the code, and merge eventually.

[James Ross]

gpz, on 31 January 2023 - 02:51 AM, said:

Looks like the existing viewers have problems with the gltf-s having "../" in their texture paths.
...
Because of this I am thinking on invalidating this scenario for gltf-s. So the textures of the gltf-s will be needed to be stored either in the .gltf file's folder or in that folder's subfolder. What do you think?

Please follow whatever limitations are in the spec and other existing editors and viewers, making our code as strict as reasonable; we can always relax the rules later, but we cannot do the opposite :)

[Genma Saotome]

FWIW the Kronos group has a .dae to .gltf converter available in github. Here's the link to the source code for that... perhaps it might be of some use in creating a converter for other modeling source file formats we've used over the years.

[gpz]

To the related github PR I noted to myself, that there are some topics that need to be clarified regarding the glTF support:

• If Y elevation adjustment to be applied (the cars are placed below the rail-top by some centimeters). It is a heritage of the .s file placement, but it is possible to add a correction factor now. Needs an inspection if the models are off only by using them as trains, or also as scenery items.
  
• Looping animations whole loop is to be (a.) the total length of the animation clip or (b.) a predefined time, e.g. 1 second or (c.) some combination like the total length with the int(length)+1 is the number of the loops to be considered (for being able to include multiple rotations into a single logical loop).
  
• Externally animated nodes are to be tagged by their "name" or a defined extension like "OPENRAILS_animation_name".
  
• The accepted animation tags: should we keep the MSTS ones or define new names.
  
• LODs are to be internal, external or either. (See some relevant comments here.)

If anyone has any thoughts or guidances about the topics above, please comment here. Because once these things are settled, they will not be possible to be changed, we will stuck with whatever set up.

[Genma Saotome]

Am I correct in understanding that .gltf models must use either .jpg or .png textures?

[gpz]

Jpg, png or dds is allowed. The trendy new basisu-ktx cannot be supportex due to MonoGame limitations.

[Genma Saotome]

gpz, on 22 February 2023 - 08:40 PM, said:

Jpg, png or dds is allowed. The trendy new basisu-ktx cannot be supportex due to MonoGame limitations.

Thank you Peter. I obtained a Sketchup to .gltf export plug in... I need to fiddle with a copy of a model to use .png textures (I want to see what it looks like in .gltf). You want a copy? Very large (13k polys), many texture files, a few incomplete spots. Different than your average locomotive or freight car.

[gpz]

Dave, yes, sure, just give a copy to me, and I will take a look on that, and also if the loader faces any bugs, I will correct them (the loader part). The larger the test suit, more reliable will be the result! If after the gltf export you can view it in the Windows built-in 3D Viewer, that's a good start. It should look quite similar in the game too.

[Traindude]

I for one would love to see animation tags and hierarchical conventions deviate from standard MSTS practice.

For example, many third-party models have parts that are animated proportionally to the speed of the vehicle, but are labelled as "WHEELS". This screws up the axle/wheel count in the consist information in the extended HUD Consist Info section (SHIFT+ALT+F5). For example, on midneguy's brilliant CB&Q heavyweight coaches, there are animated "WHEELS" that represent axle-driven generators. When viewed in the Extended HUD, instead of the wheel count being "6-6" like it ordinarily would, it's "6-2-8".

Deviating from prior MSTS conventions would potentially allow better animations for things like articulated locomotives, idler axles within a rigid frame (no "BOGIE" required) and other things...

[Weter]

Yes.
We have jealousies, animated as wipers, cooling fans, animated as wheels/rods... All these looks ugly.
8-axle tankers needed to be configured as locomotives with 1watt power to have animated gear.

[superheatedsteam]

Thank you Peter and the team for all your work on this enhancement.

gpz, on 22 February 2023 - 02:34 AM, said:

[*]The accepted animation tags: should we keep the MSTS ones or define new names.

Will defining new new names cause duplication for existing tags. 3D cab views have more than external models.

http://www.elvastowe...post__p__280752

gpz, on 22 February 2023 - 02:34 AM, said:

[*]LODs are to be internal, external or either. (See some relevant comments here.)

I can see the advantages and disadvantages of both but if I had to pick one then it would be external. Would this allow overriding the LOD settings of any glTF model?

Cheers,

Marek.

[gpz]

superheatedsteam, on 23 February 2023 - 07:54 AM, said:

I can see the advantages and disadvantages of both but if I had to pick one then it would be external. Would this allow overriding the LOD settings of any glTF model?

I can think of something like that in the anyobject_LOD00.gltf for the root node an "extras" field could be defined, where the LOD configuration could be set, when to show the anyobject_LOD01.gltf, _LOD02:

"extras": { "OPENRAILS_screencoverage": [ 0.5, 0.2, 0.01 ] }

or

"extras": { "OPENRAILS_distancelevels": [ 200, 500, 1500 ] }

[Genma Saotome]

Curiosity here... would the new format allow modelers to craft large steam locomotives as a single model instead of the MSTS practice of dividing it into two?