Suggestion: Generic Body in SDK

Yeah, I think it's a great idea.

I absolutely second that notion! Would be optimal if it was a standard prefab, full IK such as the new 'Final IK' … would make starting a true virtual game project much smoother for all...

The problem that I see with using a headless body (also from personal experience) is that if you use a real-time reflection or shadows, it will show.. a headless body.

Now, if you make it so that the head is not rendered by the Rift camera, that would be a better way to do it.

I think a very primitive example would be great, but it should stay at a primitive level unless OVR is planning to dedicate assets to building examples for this sort of thing. IMO its up to the developer, but I recognize that OVR has an interest in solving many issues and providing a great foundation for developers, so while its their choice of course I wouldn't personally expect this from them.

But on the other side of things.. It's not difficult to figure out how to setup the camera to be mounted on a head, and there are a few different (totally different, and not fully explored yet) ways to handle it as a whole so having a simple example would be good to orientate new developers but honestly this is something that will be different for almost every game, especially if you've already started development and are building on a character prefab already.

So I guess for me it just seems like Developers should be handling this, but i suppose OVR could benefit from doing the groundwork and exposing the correct methods for developers.

True; I suppose I was looking at having a production-ready player avatar, and not just a sample body :)

There is no reason not to render the head as long as it is positioned correctly with respect to the eyes. As a rule I think it is best to avoid abstractions like that in VR, and just represent things as they are in the real world.

This is an area where the SDK is pretty weak though. Nowhere is there a transform that defines the center position of the Rift that you could line up with a player model bone. Best I could do without modification is to average the positions of the two cameras.

"owenwp" wrote:
This is an area where the SDK is pretty weak though. Nowhere is there a transform that defines the center position of the Rift that you could line up with a player model bone. Best I could do without modification is to average the positions of the two cameras.

The sample code in the SDK uses the orientation combined with simple model of where the Rift is relative to the point of rotation.

It treats the viewpoint as being 15 centimeters up from, and 9 centimeters forward of the rotation point at the base of the neck. So given that, if you have a head model and have (or can calculate) the origin as being on the neck axis and at eye level, you can scale it so the distance Z distance between the eyes and the axis is 9 cm and then translate it backwards from the view position.

If any of you are going to GDC you might want to stop by the Sixense booth. We will have something to show along these lines.

Don't think this is realistic. An SDK should focus on exposing the hardware controls and continuously improving on that.

I think providing a standard body to work with is up to the game-engines. Different engines will have different ways storing, importing, animating skeletal models. It's not realistic to expect oculus to provide one for every game engine possible.

I know that unity and epic game engines come standard with a default skeletal mesh and animations so I don't see this as really a problem.

I agree that Oculus does not have to do this. On the other hand I see many demos without bodies (with or without heads ;-))

I am the author of the VR Body Movements Unity asset http://u3d.as/6v8 and I am willing to make a free but limited 'Rift Edition' of my asset, including two bodies (with and without head, based on the MakeHuman default characters). You will be able to use these freely in your projects.

Would this fulfil your needs?