Hi folks - I’m pretty new to Blender, and I’ve DL’d it with a view to making Machinima movies - like CGI but rendered in a real-time engine. I’d be interested in any links/tuts/advice on how to do this using Blender. Has anyone out there gone this route, rather than using the standard renderer?
As for specific questions -
What’s the recommended polycount for a character model?
What are the biggest textures you can use?
Does the engine do both skeletal and vertex animation? So if I had an armatured guy walking around, could I also morph the face to have the guy speaking at the same time?
Can I do a standard animated scene, and then just render it in the game engine?
Looking forward to hearing from anybody who’s tried this stuff in Blender!
(I’m on a Pentium 4 @ 2.66 Ghz, 512 megs ram, Radeon 9000 card)
No, the game-engine’s quality of images usually sucks compared to the renderer’s.
What’s the recommended polycount for a character model?
1000 tris
What are the biggest textures you can use?
I would generally reccomend 512x512 max size. Some gfx cards can’t handle sizes above that.
Does the engine do both skeletal and vertex animation? So if I had an armatured guy walking around, could I also morph the face to have the guy speaking at the same time?
Skeletal? Yes. Vertex? No. Unfortunately, the game engine is very limited. You would have to use armatures for the face, also.
Can I do a standard animated scene, and then just render it in the game engine?
No, you have to use logicbricks.
(I’m on a Pentium 4 @ 2.66 Ghz, 512 megs ram, Radeon 9000 card)
Radeon eh? Hmm…I’m not sure how well tha works with blender. Some people have experienced extreme slowness problems, or something…
machinima isn’t about rendering in realtime, it’s more about doing the animation in realtime.
so it would be cool if you could control characters and the camera in a realtime environment with blender’s game engine and then use blender’s software renderer to output the recorded movie. if you don’t use raytracing blender’s renderer is quite fast and the quality is better than what you can achieve even with the most sophisticated of the current game engines.
so would this be possible currently? i heard that blender’s game engine is quite flawed at the moment.
For simple visualization, I think the game engine is nice. Not sure how you would render (record) in realtime. %| I assume it would require Python or some kind of screen capture software. Interesting idea though.
You would have to use armatures for the face, also.
Not sure how you would render (record) in realtime. %| I assume it would require Python or some kind of screen capture software. Interesting idea though. 
