Irrlicht Engine

Asset-pipeline and Vulkan format branches are getting merged soon.

It looks like your model has "shared vertices" in some places, which basically makes the normals fuck-up. It basically makes your tetrahedron think its a sphere. If you want a tetrahedron lit like a tetrahedron you need not 4 but 12 vertices, for distinct faces. Btw, here's my explainer on...

Are you going to use a shader afterwards or are you going to rely on built-in irrlicht materials?

If yes then bump mapping / derivative mapping may be a better alternative.

In my own private fork I will replace all setRotation/getRotation calls on nodes to take quaternions (normalized, no scale) instead of Euler XYZ angles. In the fork you can already provide whole pre-calculated matrices instead of translation,rotation and scale separately. There's also an atomic-coun...

That's actually on my private roadmap.

Why are you still outputting depth to a separate color Render Target?

Rendering large models, like 1GB meshes of complete cars seems to just load the GPU so much. I guess its still rendering all the sub-meshes even if they are not finally visible. We've actually improved the performance so much that we've canned occlusion queries and just keep normal per-node frustum...

@devsh So the render process runs, loading of meshes to CPU and GPU, load/update textures to GPU, update animations CPU values, issue render commands to GPU... GPU starts rendering, shader units access meshes and textures... so we can't update those meshes or textures until the rendering has finish...

Its is very useful because a single GPU now has many command dispatch schedulers (Nvidia is 2-4 cores per scheduler), which allows for async compute or async rendering in general. Independent tasks can be executed simult. to other tasks that would undersaturate the GPU. The drawing commands you send...

Just a little sidenote. I'ver recently implemented and tested my idea of single-pass stencil-light-volumes, and it works. Depth write disabled Depth function (GREATER) [reverse the direction of Z-compare to the other objects in the scene] Frontface: stencil func GL_EQUAL to reference value 0 mask 0x...

New version is out. Stable-ish custom memory allocators, and a proper implementation of GPU memory streaming. There will be a new version sometime this week that will touch up a few things with Resizable Address Allocators, but nothing major. This is the last version before asset-pipeline (which is ...

if you want to use right handed coordinate system its easy.

You just need to patch all of the cameras+CMatrix class to swap Y with Z.

Also make sure to change backface culling to front-face or change the order of triangle indices in all meshes.

Nice stuff, but seems that your Blender Exporter or something hates your smoothing groups.

The approach you're taking about The main thread pushes the scene tree nodes in a queue, the worker threads fetch a node from queue and do the OnAnimation and OnRegisterSceneNode. is exactly what I've outlined. Will cause latency issues, unless the other cores in the system are running no other thre...

I think you'll find that OnRegisterSceneNode and OnAnimate are quite hard to parallelize since you're dealing with a scene-tree... also kicking stuff off to worker-threads may give you worse performance because of latency (time for worker thread to get its job, time you need to wait for worker threa...