>Psychtoolbox>PsychGLImageProcessing

rval = kPsychUse32BPCFloatAsap

Return a flag that you can pass to the ‘imagingmode’ parameter of
Screen(‘OpenWindow’) in order to request support for internal
framebuffers that store color information with 32 bits per color
component in floating point format if this is supported by the graphics
hardware without any restrictions.

If you hardware is able to support all its functions (especially
framebuffer blending) without restrictions then all framebuffers will be
configured to 32 bpc precision. This should be the case on ATI Radeon HD
2000 and NVidia Geforce 8000 hardware. On older hardware, which is limited
in its functionality when using 32 bpc framebuffers (e.g., the very
useful framebuffer blending doesn’t work), PTB will allocate a lower
precision 16 bpc float framebuffer as initial framebuffer for user code
drawing. All further processing of the content of that 16 bpc framebuffer
will be carried out using 32 bpc precision framebuffers.

This function is a workaround for old hardware. If you intend to do lots
of serious work with HDR imagery or high contrast precision displays, do
yourself a favor and buy a current generation graphics card.