[osg-users] Announcement: VulkanSceneGraph and SceneGraphTestBed!

Paweł Księżopolski pawel.ksiezopolsk at gmail.com
Mon Jun 4 03:07:00 PDT 2018

Hi Robert

I am pleased to hear that you decided to move into the Vulkan development.
Creating VirtualSceneGraph may shape graphics development for 
many years to come.
It may have great impact on open source projects, because in my opinion
the current state of Vulkan open source development is rather weak.
Few game engines implemented its Vulkan backend ( and I suppose 
that none of them uses Vulkan to its full potential ), 
there are few small game engines made by hobbyists/students with 
very narrow objectives in mind, lots of better or worse demos 
showing how some Vulkan features work, plentitude of tutorials showing 
how to draw a single triangle and that's all. But there is certainly lack of 
universal, feature full industry quality graphics library similar to 
OpenSceneGraph for OpenGL.

Eleven years ago I started working as a graphics developer using
OpenSceneGraph everyday in my professional career. I took part 
in a few significant industry scale projects and made 
few submissions to OpenSceneGraph, with the most important 
one beeing the osggpucull example. 
With that example I wanted to start a discussion about how to implement 
new features from OpenGL, that didn't fit well into OpenSceneGraph 
architecture, but it looked like there was no demand for it at the time.

When Vulkan came out I decided to start my own project in which 
I try to use many good ideas from software I used for years 
and combine it with Vulkan architecture :


My project goals are :
- use multithreading as a first class citizen, so that applications 
  scale well with growing number of CPU and GPU cores
- create architecture that is as close to the metal as possible, but not closer
- enable rendering to multiple windows, like OpenSceneGraph does
- work on multiple platforms ( currently rendering on Linux and 
  Windows is implemented )
- use modern C++ features that make programming more robust 
  and efficient, skip features that are irrevelant
- explore possibilities of creating scene graph that is stored and processed 
  on GPU side ( this feature is in its early infancy, but when properly 
  implemented, it may speed up applications a lot )

I spent much time learning, thinking and exploring many dead ends while 
implementing my library and I am willing to offer my knowledge to help you 
create VulkanSceneGraph. I am convinced that synergy coming 
from common work may speed up creation of VirtualSceneGraph a lot. 

If you are interested, I may start with describing my library design in 
greater detail as a food for thought, because documentation 
for pumex library is outdated or missing at the moment.

Also you can reach me through my email address : pawel.ksiezopolsk at gmail.com

Looking forward to hearing from you
Paweł Księżopolski

Read this topic online here:

More information about the osg-users mailing list