TileMapLayer3D Discussions and Feature Proposals

[DanTrz]

👋 Welcome!

We’re using Discussions as a place to discuss features and overall experience. Issues should NOT be logged here. Add them to the ISSUE section.

I hope that you:

• Ask questions you’re wondering about.
• Share ideas and Feature request

[Luceid71]

Hello!
Copying my comment from your recent youtube video over here as requested.

This tool is fantastic, keep up the great work! I have a couple suggestions for new mesh types if that's alright.

1. Box mesh is great, but we might want to make double sided walls, and a lot of those small faces are going to go to waste as they get stacked on top of each other. Having a mesh that is purely just a double sided quad would help. Additionally it might be really nice to paint each side of the quad differently.
2. Also, for pure convenience, a "smart" cube mesh could help block out objects like houses and such much faster. So it would start with making a full cube at the location, then any cubes placed adjacent to it would automatically delete their common wall, creating a larger convex shape. That way you could easily paint 3d structures without having to paint every side individually.
3. Adding to the youtube comment, I think it would probably be smart to standardize how back faces are dealt with sooner rather than later. For example, if I draw a regular wall, then later decide I actually want to draw on the back face as well, it would be convenient to just be able to draw on the other side and have that tile automatically update to a double sided quad. This would also work perfectly for the cube suggestion above. Once you have the exterior blocked out, the interior would all be culled faces. Then you could start painting inside to update those faces to double sided quads.

Thanks again for the excellent asset. I think it's great you opened this page up for feedback.

[DanTrz]

@Luceid71 I think I got it working. Manual for now. But allow painting on backfaces of flat tiles.
If you could test it that would be awesome mate.. then merge that into main.

No setting to toggle..Just paint on backfaces

[DanTrz]

@Luceid71 The latest version on the main branch (v0.4.0) has most of the improvements in this discussion ready.

[Luceid71]

Hey Dan, apologies for the delay. I was also taking some time off for the holidays.

I love the implementation of all the features discussed here! The box mesh is such a huge time save, and painting on both sides of a tile works exactly as expected.

The only thing I noticed was that adjacent box meshes are not updating and deleting their common wall. This would lead to quite a few wasted quads in the scene needing to be processed that the player will never see.

Anyway thanks again for the awesome update. Love it!

[DanTrz]

Hey @Luceid71 - There is no easy way to merge them live in a non-destructive way. And I want to avoid destructive behaviour, as that will break tiling operations such as Undo, Erase, Replace, etc.
Since we are now using MultiMeshInstance3D with Chunks, the system is well optimised, and we reuse the same mesh. All instances share the same mesh data, and each chunk renders in a single draw call.

I wouldn't worry too much about performance unless you see a real case issue. I have huge scenes and usually no significant problems. In any case, if you have a scene with over 500,000 BOXES Meshes, you could merge them into a single Mesh. The plugin allows "Bake/Merge" operations.

[Luceid71]

That makes a lot of sense, I hadn't considered the non destructive aspect to it. In my head, I was thinking each cube was an entity in a cell, and that entity would add / remove walls depending upon it's neighbor. But the entities in this system are walls, not cubes, so it wouldn't be possible without a giant rewrite. The cube mesh is just convenient tool for multiple walls at once.

And as for performance, yeah you're totally right. I have a tendency to fall into pre mature optimization haha. I won't worry about it then.

[TheClave]

Hi Dan, first of all, I really like the new updates and the progress on this plugin. I have experimented with PBR textures and would like to prepare a pull request, but I would like to get a design-wise opinion of yours. PBR textures allow us to use normal maps, roughness maps, metallic maps etc., everything which Godot provides in it's renderers.

My current setup identifies the loaded TileSet if it's a CanvasTexture which supports 2D normal maps and specular maps and reads them into an updated shader. I am not sure how to move forward here, because these options come to my mind:

1. There is only one "Load TileSet" button and it detects if the loaded TileSet is either a StandardMaterial3D, OMRMaterial, CanvasTexture or a simple Texture2D. There are no visible changes in the GUI, everything works like it does now.
2. There are multiple buttons like "Load Normal Map", etc. but this will probably overload the UI, even though this might be more user friendly.

Furthermore, depending on the actual resource used, not everything is supported. If a CanvasTexture is used, metallic rendering is not possible. This means, that to have more control over each value, it'd be beneficial to have sliders instead - at least if we use a simple CanvasTexture for that.

Probably the easiest way would be to simply support StandardMaterial3D for now and go with the first option, because this offers the most customizations and control over the Material sliders. What do you think?

Cheers,
Johann

[DanTrz]

First of all. Thanks for taking tbe time to work on this. It is a great idea. Im with you on StandardMaterial and one button

[DanTrz]

@TheClave You able to join the discord to talk about your proposal? We have a Server. https://discord.gg/e8kjnk4Y