So I’ve been tasked with setting up an image processing server to reduce repetitive Photoshop tasks. For this task, I have a 12-core 2013 Mac Pro of all things (not going to use macOS though). Part of me wants to take the path of least resistance and just set up some hot folders and processing images via imagemagick calls in bash or python or whatever.
But the better half of me wants to take the opportunity to learn kubernetes and deploy some stateless imagemagick containers that perform the conversions…
The opportunity here is that I’m reasonably confident that no one here knows:
What imagemagick is
What containers are
How long it should take me to complete this task
So I think I’d like to take advantage of this latitude and take a bit longer to get them something functional while being paid to learn new things. Additionally, of course, if it ever becomes necessary to scale up later, that’ll be a walk in the park.
Anyway, this is what I had in mind:
How possible is it to perform OpenCL workloads in a container? imagemagick supports it for some operations including resizing.
Yeah, that’s what I’m planning on, plus a macOS vm that will be the test client.
Ah, “image processing” in this case is fairly complex and will involve scraping and auditing metadata and doing various prepress color management things, and will eventually send assets to a DAM via API. The imagemagick operations will only be part of it.
Also, I have had problems in the past with automator not triggering on watched folders…
In other news, I get to do this soon. Came home to down network