ID 228821 - Nuke's performance is noticeably worse when there are large amounts of metadata in a script's Read nodes

Problem summary
Reading in EXR files with a large amount of metadata causes Nuke to be laggy and sometimes unresponsive for up to several minutes when many Read nodes are used.

 

Customer reported version
Nuke 8.0v6, 10.0v3

 

Linux CentOS 6.2

 

 

 

1) Open the attached Nuke script and change the file paths of the Read nodes to point to where the attached EXRs are located. This can be done using the following Python code:

 

selectedNodes = nuke.allNodes("Read")for node in selectedNodes:  node['file'].setValue('filepath/manyattrs.%04d.exr')

2) Nuke will become unresponsive while reading in all the files, and again when trying to change the frame.

 

The example EXRs are generated with a Python script that adds an attribute with a large string in it to the EXR header. 

 

Additional Reproduction:

1) Open the attached Nuke script.

2) Select a Read node and add a Grade node, observing performance.

 

 

 

Nuke 15.0v4 - Windows 11

Nuke 14.1v4 - Windows 10, CentOS 7

Nuke 12.0v1 - Windows 10

Nuke 10.0v3 - Windows 7
Nuke 9.0v8 - Windows 7

 

Expected behavior
Nuke should be able to handle reading in the EXRs faster and not become unresponsive for several minutes.

 

Actual behavior
Performance is very slow when panning or adding nodes in the Node Graph as well as changing frames in the Viewer.