4 ms·
Slight hijacking: I have a DSLR and have always liked photography (the taking photos part), but I've never been interested enough to get into post-processing o
by croon 6y ago
Slight hijacking:
I have a DSLR and have always liked photography (the taking photos part), but I've never been interested enough to get into post-processing of images.
> Colorlab's neural network can even import a reference image, analyze and train on it, and then apply its inferred style to footage.
This is pretty much what I want, but doesn't have to be industry quality. Is there any CLI/API batchable tool with sane defaults that preferably costs less than $49/month?
- eurekin 6y agoA large portion of post processing in photography seem close to impossible to automate. Typically best studio quality pictures are retouched, where a lot of eyeballing is done - touched up until no further "issues" are left. Example of an issue resolution would be dodge and burning. It's a locally applied lightening or darkening of picture. It is used to tone map an image and apply make up, among others. Tone mapping itself seems like a good candidate to automate, cause most often it's used to squish down high dynamic range scene into a screen quality image. It can recover details that would otherwise be overblown or underdeveloped. Still, retoucher can selectively apply dodge and burn to guide viewers attention to main subject and away from distractions. That's near impossible to automate,the decision if it should or shouldnt be done for a particular image. Most often than not, a good post processing (or retouch) is those actual decisions, where to apply and what kind of change. Everything else can easily be automated
- croon 6y agoI see what you're saying, and am not qualified enough to disagree, but the quote in my OP at least indicates that something can be done. I realize this won't yield the best results everywhere, but might yield better results than nothing in most cases, which is basically what I want. But regardless, I appreciate the thorough response.