1

I am looking to make an arm where each joint can be controlled given an input. For example, I would receive a location each millisecond and I would need each joint to move to that new location. This is going to be used to simulate motion capture. Are there any good tutorials about how to do this. I am programming this on an atom board with limited support for openGL. I tried using C++ G3D but it seems to be too complicated for the graphics card. There were some glitches when rendering. Would a more basic OpenGL only approach be better? Are there models for arms that have pivot points built into them?

slimbo
  • 2,699
  • 4
  • 25
  • 36

1 Answers1

1

G3D itself uses OpenGL as a rasterizer backend, so I doubt that would make any differency. After all OpenGL is just a drawing API and doesn't maintain a scene. You send it commands to draw points, lines and triangles and it executes them. It's really nothing more. You need to implement scene management to do anything usefull with OpenGL, and controlling a figure is already a quite complex task.

datenwolf
  • 159,371
  • 13
  • 185
  • 298
  • I'm worried that if I just use OpenGL I can only access basic functions, whereas G3D is accessing more complex functions from OpenGL that my not be supported by my low-end graphics support – slimbo Mar 02 '12 at 10:53
  • Afterall, I don't need high end ray tracing or shadowing. I just need a really basic 3D model that can display motion – slimbo Mar 02 '12 at 10:54
  • @macneil: Well, did you try out one of the simplemost OpenGL test applications, for how well they work? Oh, BTW: Atom GPUs are PowerVRs and Linux driver support for them is very, very poor ATM. I struggled over 3 weeks with one and then gave up. So if you're on Linux if the model is not too complex, software rendering with Mesa might be the way to go. – datenwolf Mar 02 '12 at 12:11