CTL
CTL copied to clipboard
ampas
New CTL programs
I've come across a few scenarios where having programs capable of tasks not currently supported by the built-in tools in CTL would be highly advantageous.
The first scenario involves the desire to input a scalar, vector, or array of numbers directly into CTL and observe the numerical output. At present, we utilize CTL through Matlab by creating a temporary image, processing it with ctlrender, and then importing the results back into Matlab. I believe @KelSolaar employs a similar method with colour-science. Ideally, a program that accepts numerical data directly might be more efficient, although I acknowledge there might be intricate details that need to be addressed.
The second scenario is the need to execute a CTL program without feeding an image into it. This functionality would be particularly beneficial for conducting unit tests, such as the ones I have outlined here: https://github.com/ampas/aces-unittest-test. Currently, my approach involves using ctlrender, even though the image data itself isn't being utilized, which seems like an inelegant solution.
I'd be interested in hearing your opinions. Thanks.
Passing a scalar or vector and getting it transformed on stdout would be great yeah!
@aforsythe @KelSolaar do you imagine this functionality should be part of a separate program independent of ctlrender or as alternate input/output path added to ctlrender, or something else ?
@michaeldsmith Sorry the repo was private. Public now.
re: stdout ... I'm open. I can see arguments both ways.
@aforsythe @KelSolaar do you imagine this functionality should be part of a separate program independent of ctlrender or as alternate input/output path added to ctlrender, or something else ?
Oh sorry, I was not clear, yes on existing ctlrender. I would keep things simple, it is already huge progress to see someone stepping up to update this repo so I would not want to add the extra burden of having to maintain yet another program.
I tend to agree ... thinking about it, the command line interface we'd want would probably be nearly identical to ctlrender, so why not just build on top of that.