Hello, to all friends
I tried to convert this shader
in TouchDesigner, it works fine in version 2021 but the same code doesn’t work in version 2023
There are several posts about this
and I read them all carefully and made several changes to the code but unfortunately it still doesn’t work in version 2023.
Thank you for your help
Converted version in 2021 and 2023:
After all the efforts, I think the problem is in BufferB because it has NAN output
But I don’t know exactly which part of the code is wrong
Thank you for helping me 
Hey, thanks for the sample. It’s hard to tell example with a more complex example such as this, but the most likely reason is because intBitsToFloat() and floatBitsToInt() is not as stable of an operation as one would hope.
In particular, if you are generating an integer ID that you want to encode as a float to be saved out into a floating point buffer, you need to make sure it doesn’t create a value that would create undefined values with inBitsToFloat().
In particular, you can’t have a set of bits that would create NaN or Inf
If the encoding of a NaN is passed in *x* , it will not signal and the resulting value will be undefined. If the encoding of a floating point infinity is passed in parameter *x* , the resulting floating-point value is the corresponding (positive or negative) floating point infinity.
Additionally, denormalized floats can be flushed to 0 at any time
Any denormalized value input into a shader or potentially generated by any operation in a shader can be flushed to 0.
This means an integer 0x1, which is a denormalized float when interpreted as a float, may get turned into 0x0 (but there are many such values)
Can your algorithm encode IDs directly as floats, rather than using intBitsToFloat() and back?
Thank you for your explanation
I did these methods, but unfortunately I did not get an answer
This method is a little close, but it still doesn’t work
Bufferb output in version 2021
You shouldn’t be using 255, since you aren’t working in 8-bit anywhere. The big question is what are the ranges of values that the integers can be in this algorithm, and can those be expressed uniquely in a float.
Yes it worked with this change
Thank you 
Hello friends
I encountered another problem in another shader
It works fine in version 2021 and in version 2023 I either encounter this error or it does not work properly
It is a simple shader but I can’t find the problem with it
It is not returned as NaN or noise, this time it is a new problem I think
Thank you for your help
The graphics card driver is also up to date.
