First Happy new year!
2013 was pretty great (first DX11 release, Node13, lots of interesting projects, birth of a new tool ;)
Hope 2014 will follow, and to be honest, it's starting quite in many interesting ways.
First thing I'm reasonably happy on adoption for DirectX11, still some parts are missing, but most times people going into it don't go back (which is a good sign).
There's still some nodes missing, but well I'm kinda more or less on my own writing it (thanks to the few people who contributed nodes), on top of various projects, so it's a lot of time (gladly I use it for projects, so at least I can add some bug fixes/Few more nodes, but nodes are generally so specific they not really much point pushing to core).
If a few wannabe c# programmers decide to pop in and help add some of the missing nodes (Text Geometry/XFile writer/Player...), please feel free to ping me, I'll happily help with basics of coding a Dx11 node (it's much easier than writing a DX9 one believe me ;)
Also of course thanks to people who started to post dx11 contributions, I believe that helps adoption, and at least I can spend more efforts on the runtime instead.
On the same way I'd love few people to write help patches and more examples, it also takes time, I rather enjoy to explain routines, but same thing = time.
So now 2014 started on a quite fast paced mode (I spent bit of time between xmas and new year too), let's see what's going on:
2013 was pretty great (first DX11 release, Node13, lots of interesting projects, birth of a new tool ;)
Hope 2014 will follow, and to be honest, it's starting quite in many interesting ways.
First thing I'm reasonably happy on adoption for DirectX11, still some parts are missing, but most times people going into it don't go back (which is a good sign).
There's still some nodes missing, but well I'm kinda more or less on my own writing it (thanks to the few people who contributed nodes), on top of various projects, so it's a lot of time (gladly I use it for projects, so at least I can add some bug fixes/Few more nodes, but nodes are generally so specific they not really much point pushing to core).
If a few wannabe c# programmers decide to pop in and help add some of the missing nodes (Text Geometry/XFile writer/Player...), please feel free to ping me, I'll happily help with basics of coding a Dx11 node (it's much easier than writing a DX9 one believe me ;)
Also of course thanks to people who started to post dx11 contributions, I believe that helps adoption, and at least I can spend more efforts on the runtime instead.
On the same way I'd love few people to write help patches and more examples, it also takes time, I rather enjoy to explain routines, but same thing = time.
So now 2014 started on a quite fast paced mode (I spent bit of time between xmas and new year too), let's see what's going on:
- Port to core runtime to SharpDX is pretty much complete, nicer, faster, simpler. That means support to latest versions, and support up to Win7 is there.
- Texture was bit of an issue, but thanks to directxtex.codeplex.com , few exports and P/Invoke, now have a portable Texture reader/writer, with tga support on the way, and BC7 encoder, if you survive the "very" slow export, in order to enjoy the very fast import ;)
- Most of old nodes are also done (i'd say 80% of the nodes are ported)
- Many type of nodes are now templated (using T4 templates), so most geometry/layer and others, are just reflected and node is generated from that, makes refactoring a hell of a lot easier).
- Shader nodes have been reasonably revamped.
- Lot of new interesting/experimental features on the way too ;)
- Many high level nodes, so non shader experts can start to use their gpu in a reasonable way :)
Now of course there's still work to do, Pin system (once greg will have sorted the convolution issue), will also get some revamp, Resource management is getting along more improvements, and still some more questions about some other bits (see below).
Please note that most work as been done to make API simpler. API now uses less code, is faster, and I find using it is also less code. Writing a node with new API is really simple, there's much less cooking around.
First session of rolling out (pre alpha early bird to not use for production you've been warned) small release is incoming, so will be able to see how things are getting along.
There is still some bits I'd like to work on, in no particular order:
- Code editor: Time to really start to think as fx files as projects, and just not a window with unusable code completion.
- Improve elements decoupling: Some more core, vvvv being so tied to MEF it's gonna be fun, but there's some big rationale behind it, which deserves a post on it's own.
- Get rid of property injection : So i can really get proper inheritance control , and nodes could even be reused.
- More high level nodes: No comment, everyone likes high level nodes
- New transforms : The funny thing is actuallyafter some thoughts, moving transforms to a new type opens a hell of a lot of new doors, more on that later, that deserves another post.
- Continue to simplify: It doesn't mean lot of small files with 2 lines of code each (no offense to java people, which actually can't since their header is already filled with imports ;) Simpler is better, I don't need the most amazing configurable runtime, I need a lightweight, easy to use and fast runtime!
- Some more top secret things are also on the way ;)
As usual, thanks for reading and happy new year again !
No comments:
Post a Comment