Hi there,
I`m testing some simple workflow with PNG or fixed to float conversation in general.
How do i move from one to another - seem to be simple task,

I’m not sure I understand precisely the question you’re trying to ask, but you should be able to convert to and from fixed and float pixel formats the format parameter of most non-null TOPs (for example, a channelmixTOP).

To my knowledge your only really losing data if you’re reducing your bit-depth or number of channels (unless your pixel packing). If you’re referring to the Display Pixel Values readout, from my observation it shows RGBA in integers if you’re using Fixed and always shows float values regardless. For 8-bit fixed it’s between 0-255, 16-bit it’s between 0-65535, etc.

As you can see from the attached image, there seems to be an observable difference in pixel values when down sampling bit depth but not when converting from fixed to float at the same bit depth.

Converting 16-bit fixed point to 16-bit float will lose data since 16-bit fixed has 16-bits of data for values between 0-1, while 16-bit float only has 10-bits for values between 0.1 and 1. You need to use 32-bit float to represent 16-bit fixed data fully.