ApplyRGB should honor Numpy internal matrix storage order

[doug-walker]

Numpy matrices may be stored in either row-major or column-major order. The user is typically not aware of the order being used, especially since some Numpy operators swap the order if it is more efficient for them. For example, transpose simply swaps the ordering flag rather than moving numeric values.

The applyRGB function in OCIO essentially just flattens the input and expects channel-interleaved ordering. Ideally, applyRGB would check the storage order and transpose values, if needed. At a minimum, it should issue a warning if the storage order is not as expected.

[nick-shaw]

The flags property of a NumPy array shows the storage order.

  C_CONTIGUOUS : True
  F_CONTIGUOUS : False
  OWNDATA : False
  WRITEABLE : True
  ALIGNED : True
  WRITEBACKIFCOPY : False

Currently applyRGB() will only work correctly if C_CONTIGUOUS is True.

[KelSolaar]

Easiest is wrapping the array with np.ascontiguousarray before using ApplyRGB. I modified Colour for now with this.

[Asheikhm]

Hi, could I work on this for ASWF Dev days?

[doug-walker]

Thank you @Asheikhm , I've assigned this one to you!

[Allanodora]

Hi there! I'm a Python engineer and I'm interested in helping resolve this. I noticed there hasn't been an update here in a little while—is this issue still available for me to work on? I'd love to contribute."

[doug-walker]

@Allanodora , welcome, it would be great to have your help!

Unfortunately, this issue was fixed already, although it did require C++ coding capability, so I'm not sure it would have been a good fit?

Could I recommend that you look at these issues. Those would all be pure Python coding.

[doug-walker]

Closing this issue as fixed by PR #2206.