Quick recap

The team discussed potential changes to the Codec Identifier for clarity and relevance, and the performance of Baselight, with a focus on understanding the channels used in image processing. They also explored the possibility of integrating a different library, such as Kakadu, for better performance, and the potential benefits of enabling a reversible color transform for certain images. Lastly, they discussed the need for more information about the setup and content used for testing, and the upcoming presentation on color management of openexr files.

Next steps

• Cary to update the pull request with the plan to change the codec identifier to "HTJ2K" when merging into the main branch.

• Cary to follow up with Jay Schulte at ILM for feedback on testing the codec with their content.

• Peter to test the codec using images from the exrmetrics tool testing.

• Pierre-Anthony to investigate Bill Barrett's feedback on Baselight performance if more details about the experimental setup and content are provided.

• Cary to wait for a couple more things from Kimball before making a new patch release.

• Cary and Peter to provide internal results before merging the codec into the main branch.

• Cary to attend the color interop meeting on Monday.

• Doug to present the proposal for color management of OpenEXR files at the color interop meeting on Monday.

Summary

Updating Codec Identifier for Clarity

Summary points: Cary proposed changing "ht" to "htJ2K" in the Codec Identifier for clarity and relevance. Pierre-Anthony agreed and suggested entering this change in a pull request. Cary will update the pull request with the proposed change. The team is currently evaluating the new feature, and any changes will be implemented before its release.

Baselight Performance Review and Testing

Cary and Pierre-Anthony discussed the performance of Baselight, with Pierre-Anthony suggesting that more details about the setup and content used were needed to fully understand the results. Cary mentioned that Jay Schulte from Ilm was planning to test it on some content he was working with. Peter also mentioned that he had prepared some images to test the exilemetrics tool, but noted that the results didn't reveal anything new.

Understanding Image Processing Channels and Libraries

Pierre-Anthony discussed the importance of understanding the channels used in image processing, as the current encoder only detects RGB channels. He expressed reluctance to change the algorithm without more information about the non-RGB files. Peter suggested asking Phil to post the Exr header of the files for more clarity. Pierre-Anthony also mentioned the need for better understanding of the exact setup for optimization. Michael and Peter discussed the possibility of using the Kakadu library instead of the open source Jtk library for faster playback. Michael mentioned the interoperability between the two libraries and the existence of a third codec implementation, comparmato. Cary confirmed that they don't have a souped-up internal version.

Performance and Customization in Library

The team discussed potential improvements to the library, specifically regarding performance and customization. Pierre-Anthony suggested the possibility of integrating a different library, such as Kakadu, for better performance, but noted that this would require significant changes and maintenance. Peter agreed that such modifications would be heavy and not just a simple flag change. The team also discussed the potential benefits of enabling a reversible color transform for certain images, but noted that this would depend on various factors such as image content and noise levels. Cary mentioned that they were waiting for a couple more things from Kimball before making a new patch release, after which they could merge the current changes into the main branch. Doug announced a presentation on color management of openexr files at a meeting on Monday.