Talk:Choosing Hardware for Fusion

From VFXPedia

"On some cards floating point images are uploaded as 8bit clamping the image's pixels to the [0.0, 1.0] range for performance reasons. This causes problems with hardware LUTs since the LUT gets applied after the color data has been crushed to the [0.0, 1.0] range. To work around you will have to use software LUTs which are slower. " Is there a list of what does or doesn't work like this? It's very important to the work we do. --Chad

All non-Nvidia cards are converted to int8. This default setting may be modified using the Texture Depth Preference. --Stephen

"Smooth resize in the view. Smooth should work for all integer images. Depending on the graphics card it will also work for float images."

Can anyone please give a list of graphics cards that support float images smoothing in Fusion? --Gringo

Well, I found out that GeForce 7900 GTX supports smooth resize only for 16 bit float, while GeForce 8800 GTS supports it for both 16 and 32 bits float. --Gringo