[Mac Boy]

I have been trying to find a way get OpenGL to create a graphics context with 64-bit color. Why so accurate? The data is not actually being displayed on screen, but being used to model real-life objects, to capture light intensity, doppler information, etc. so I high level of accuracy is necessary. I can fully understand a graphics card not being accelerated for this level of accuracy, but shouldn't the software renderer be dynamic enough to allow for it? For both renderers it is listed as "--not available--" in the OpenGL Info program.

Any ideas?
Thanks in advance.

- Ross

[hayesk]

If your data model has all the information, can you just downsample to 32 bits when passing it off to OpenGL?

That way you can just retain the 64-bit version outside of the "OpenGL space"

[Mac Boy]

Originally posted by hayesk:
If your data model has all the information, can you just downsample to 32 bits when passing it off to OpenGL?

Actually we are using OpenGL to perform the calculations. Set the vertices to be particular colors (which is really a representation for other various attributes, like velocity), render the object, and OpenGL wonderfully does the necessary interpolation for every single pixel on the screen, which is exactly what we need. We just need more precision than 8-bits (or 16?) for each color channel.

This is why we need 64-bits. So can it be done?

- Ross

[calumr]

Originally posted by Mac Boy:

This is why we need 64-bits. So can it be done?

- Ross

John Carmack started pushing for 64-bit colour in his .plan files just after Quake3 was released (IIRC). Until it is supported in HW by ATI/NV I wouldn't have expected Apple to support it in SW either.

I was going to suggest the open source Mesa3D OpenGL library, but it doesn't look like they go above 32bpp either.