this must have to do with the value range of the second input. When inspecting it via the TOP viewer’s Normalized Split feature, I see that each color channel’s values have a range of -0.25 to 1.xxxx. Now I guess the Cross TOP uses a function where you end up with some remaining values that are being subtracted from the first input.
I’ll post this question internally if the Cross TOP can not somehow take care of this but currently an additional Level TOP just before the Cross TOP with the Clamp Input parameter being set to “Clamp [0-1]” will get you around this.
That is kind of what I thought was happening, but I wasn’t sure. Looked like values outside of a normal range were slipping through.
Thank you for asking this internally. My brain says the Cross should handle this but maybe it is by design, and I am not aware of the possible applications? Would not be the first time I missed seeing the whole picture and/or the advantages of certain ways TD does things until they are pointed out to me.
Thanks for the advice for using the Clamp Input. Nice quick fix, much appreciated!
Hey,
The issue here is that pow(x, y) is undefined for x <0, and so your gamma operation on the Level TOP is creating NaN values due to < 0 values on input. I’ve clamped the value in the Level TOP in the next 2023.10000 build we post, but in general its good to be careful with negative values and doing compositing operations.