Machine learning in Nuke | 02 Using CopyCat to Deblur a More Complex Frame With

Delve deeper into deblurring with CopyCat in this second video by working on more complex frames. This second tutorial expands your knowledge of CopyCat in Nuke, making you a deblurring expert.

Follow along by downloading the assets on our course page: https://learn.foundry.com/course/7115...

For more Nuke tutorials, check out https://learn.foundry.com/nuke
For product information, go to: https://www.foundry.com/products/nuke...

Chapters:

00:00 - Introduction
00:33 - Create Ground Truth Frames to Manually Sharpen Out-of-Focus Frames
01:19 - Paint Out the Highlights to Restore the Original Shading of the Blurred Image
02:39 - Match the Position With the Vector Generator to Put Details Back Into the Blurred Image
08:07 - Define the Regions CopyCat Can Learn From to Save Processing Time
09:09 - Create More In-Focus Training Frames to Help Copycat Produce a Better Result
19:44 - Specify Out-of-Focus Input Images for CopyCat to Deblur
24:11 - Render the Ground Truth and Input Sequences to Speed up Training
25:05 - Create an Identity Frame for Reference to Help CopyCat During Training
27:14 - Connect the Inference Node to Apply the Trained Model to Out-Of-Focus Frames