3000th commit - IrrlichtBAW (GIT repo, v 0.3.0-gamma1)

[devsh]

Native binary format coming which will save and load animated meshes, static meshes, all texture types etc.

As well as some documentation.

[devsh]

Added Multisample Textures and Texture Arrays.
https://github.com/buildaworldnet/Irrli ... b2a3e2defb

Have your deferred in MSAA!

[devsh]

Showed in an example how to do a programmable MSAA resolve with a shader
https://github.com/buildaworldnet/Irrli ... 49a377d46a

[devsh]

And added another example on how to do cubemap shadows!
https://github.com/buildaworldnet/Irrli ... 7007ab21f0

[devsh]

Example with shadows got upgraded and optimized:
https://github.com/buildaworldnet/Irrli ... 5f76791858

1+ Million Polygons, 625 mobs, moving light, 56FPS

[devsh]

My modified take on Stencil Routed K-Buffer

[devsh]

Added our own binary format, compression and encryption coming soon.

We made a comparison in loading a 18mb binary .x file

DEBUG:
60000ms for .x
40ms for .baw

RELEASE:
1800ms for .x
14ms for .baw

This means we are hitting the disk-read-speed bottleneck.


Known issue: Loading skinned meshes from a .baw file leaks memory.

[devsh]

Fixes:
Fixed all memory leaks.
Added safe bitfield export and import on serializable classes

Core:
Moved to C++11
Added AES128 Encryption, LZMA compression and GCM auth to BAW format
Added per-context state tracking for non-shareable opengl container objects such as active VAO, FBOs, Samplers and XForm Feedbacks

Extensions:
added a matrix AVX vs. SSE benchmark as an example
provided a 'command-line' convert2BAW tool for converting all supported mesh formats to .baw

Also we have branches for :
C++11 threading and atomic fast locks
VAO cache for multithreaded drawing and in the future, mesh loading and streaming to GPU
this branch also includes c++11 macros and base classes to make uncopyable and unassignable classes.
as well as the change from map<> to unordered_map<> for the per-context sampler cache

[devsh]

Compute Shaders COpenGLExtensionHandler utility functions are in, proper shader API coming soon.

[devsh]

Added an extensions repository with 1 sample extension (actually used in example 23.Autoexposure)
https://github.com/buildaworldnet/Irrli ... master/ext

[devsh]

A decision has been made to move ahead with providing a dual rendering backend of OpenGL and Vulkan, with OpenGL support to be dropped as soon as our Vulkan driver is mature and Intel stops supporting the integrated GPUs which don't support Vulkan on windows.

This is because we will be porting the engine to android.

[devsh]

Added a FAQ
(its mostly the result of all the questions in this thread)

https://github.com/buildaworldnet/Irrli ... W,-the-FAQ

[Virion]

Haven't been to Irrlicht forum for years. But wow, Irrlicht-Vulkan combination would be great.

[devsh]

Try it

[devsh]

Merged the Mesh Optimizer branch, its available in the engine via the API and in the convert2BAW tool.

It does everything that AMD Compressonator does to meshes, but more slowly, and uses a different method for normal attribute compression.

Will add support for glTF 2.0 and the AMD Compressonator tool in the future.
(also want to add support for keyframe dropping, animation optimization and compression -- hopefully there is a 3rd party library for that)