Hello there. Thanks for asking that question, as this is, at least in my opinion, one of the greatest problems in the world of libre game development. I have worked with quite a lot game engines before (Source, Darkplaces/Quake3, Sauerbraten, now Unreal Engine 3 for UT3). I do mainly level design stuff, as well as a bit of environment art. No really serious stuff though, due to the lack of motivation and nice engines

What I miss in my (open source based) gamedev toolchain? A lot, I would say. First of all, I had never the opportunity to work with a complete and, more importantly, working FOSS mapping pipeline. Working with GTKRadiant (the map editor for all Quake-based respectively all games using Quake3's map format, for example Darkplaces/Nexuiz) was a horrible experience. Sauerbraten was okay-ish for a while, but it lacks comprehensive editing tools (ingame editing might be cool for a while, but it pisses you off after a while, especially since you don't have multiple viewports etc.) and model integration. A big problem in the libre game development world might be that the majority of developers are pretty conservative for that matter. Working with CS2Blender wasn't too bad. I like its concept (it would be utter awesome if someone implements something like that for Sauer or some other engine). But while I only hooked up some test levels with it, the shader management pissed me off. Again, hooray for user friendliness.
What I'm trying to say is that there is a need for a solid base, meaning that, regardless of what technology or engine is in use, keeping a specific concept for the future pipeline in mind is the key. I don't want to shape basic stuff with complex mesh geometry, I'd rather go for simple shapes (like brushes in UE or Quake, or cubes in Sauerbraten). Then again, there should always be an easy to integrate complex geometry for detailing. When I compare the existing FOSS tech with, for instance, the Unreal Engine, which is pretty much a industry standard (and before someone whines about the effort etc. I quote: "One of the our important goals is to improve open source tools to better fit to the
professional gamedev indrustry."), then I actually see the problems in the engine, respectively its tools and environment, rather than the external applications, like Blender or the GIMP. Both are very nice examples of fine software (although they all have their flaws - every software has its flaws

). The root of the problem must be eliminated to solve it.
Lastly, my little wish list for Blender:
- A faster renderer/sculpting mode - I don't know what exactly is responsible for that, but working with a million of polygons can be really annoying, due to slowness
- A GLSL viewport - (Is it already in schedule for 2.5?)
- A flatten function
_________________
http://rgames.tuxfamily.org - My not very often updated Blog.
ah, thanks! i wasn't aware of that. i tried out a recent SVN build now. the patch of your link sounds great too (possibility to bake lighting only).
but we need better baking quality (-> baking doesn't seem to use any kind of anti aliasing).
Oh yeah OSA for baking would be nice
Add that to the wish list.(Totally forgot that)
Also a stretch vizualitation green for non strecht red for too little space blue for overstretched for UVs.
That the textures path is also searched for textures whenever you load a model. At least, when importing a .ac file and having the textured referenced as(the actual line in the .ac model that names the texture to load):
texture "stonegrass.rgb"
And setting the texture path to where the that texture is, it won't find it unless you run Blender from the texture's directory, or copy the texture into the same directory the model is.
This shouldn't be hard to add.
> Advanced image manipulation capabilities / image editing (2d drawing and manipulation software, would be nice to remove gimp / krita from pipeline), even more on the raster part (to remove inkscape from pipeline)... also with real-time and "turns" cooperative editing using verse and a version control system integration perhaps? (something like bazaar). Also i will like a mind-maping module

Thing about it... Some features from pencil software integrated ... and more on the text editing capabilities. The ideea is to have blender for all art related / development stuff. Like from starting a project: design documents (text or ming maping), 2d story board (animated).... hope you get the ideea... my post is messy. I could make a design document for the "mind-maping" viewport/feature/tool.
This tool is a BASIC tool that we don't have in blender and we need it for making games
- Soft / Hard normals. A way to break the normals between 2 faces without creating more geometry.
my 2 cents
No. What he means is smooth groups:
http://www.3dtutorials.sk/index.php?id=104
target weld would also be nice or just redoing the snap tool so it's possible to use proportional edit with it.
Fog,
see the edge split modifier.
For the 3 views - I meant set up the background image so that it stays behind the object for three views - ie rotate the object and the background, top and side images would stay relative to the object top side and front.
LetterRip
Letterrip
The edgesplit creates more geometry which is not good for game models. Less geomoetry more FPS.
Last edited by Mariolink on Sun Dec 16, 2007 11:36 am; edited 1 time in total