Hi!

This is probably a stupid questions, but I try to read back a staging texture on cpu, the format is 16bit float, however to my limited knowledge there is no standard 16bit float in C++, how can I read the values in a meaningful way?

The mapping procedure work, I tried with 32bit float and I read them fine.

Thank you very much!

Jacques

You can use the header directxmath_packedvector.h, part of directxmath to do the conversions.

Sixteen bits float on gpu are well documented, you can also write your own conversion. It is one bit sign, 5 exponent and 10 mantissa.