• Welcome to the Lightroom Queen Forums! We're a friendly bunch, so please feel free to register and join in the conversation. If you're not familiar with forums, you'll find step by step instructions on how to post your first thread under Help at the bottom of the page. You're also welcome to download our free Lightroom Quick Start eBooks and explore our other FAQ resources.
  • Stop struggling with Lightroom! There's no need to spend hours hunting for the answers to your Lightroom Classic questions. All the information you need is in Adobe Lightroom Classic - The Missing FAQ!

    To help you get started, there's a series of easy tutorials to guide you through a simple workflow. As you grow in confidence, the book switches to a conversational FAQ format, so you can quickly find answers to advanced questions. And better still, the eBooks are updated for every release, so it's always up to date.
  • Dark mode now has a single preference for the whole site! It's a simple toggle switch in the bottom right-hand corner of any page. As it uses a cookie to store your preference, you may need to dismiss the cookie banner before you can see it. Any problems, please let us know!

Face Recognition - Negatives - Sync regions?

Status
Not open for further replies.

Linwood Ferguson

Linwood Ferguson
Lightroom Guru
Joined
Jan 18, 2009
Messages
2,587
Location
Fort Myers, FL
Lightroom Experience
Advanced
Lightroom Version
Classic
Lightroom Version Number
9.2.1
Operating System
  1. Windows 10
I'm digitizing old images and trying to do facial recognition as a quick way to tag. I'm also doing a lot of negatives, and I like saving the original raw inverted, but that defeats facial recognition.

I'm tempted to export and import a JPG so I can do face recognition, but keeping stacks updated is just too painful -- they only accept mass changes if expanded, so if you are otherwise changing metadata, it's WAY too easy to not get all items in a stack.

So...

Can I export a JPG and take it back into a stack (I know how to do that), run face recognition on the JPG's, and then somehow move the face regions into the corresponding raw stack member?

There are two issues with that -- one is just getting it to sync the region, I'm not aware of anywhere you can do that at all? The other is if I could (let's say I write some SQL), is this going to screw up face recognition (having regions that are inverted)? I'm guessing yes?

So all I can do with these is find a way to sync metadata on the other stack memory to pick up the keyword from face recognition, but not the region. Is that right?

Do any of the plugins floating around do that? I think I vaguely recall one like that but some quick searching did not find it.

Obviously the other alternative is to discard the raw and just keep the inverted export. I'm not quite there yet, but issues like this are pushing me in that direction.
 
Not sure if this can do it, but the plug-in that comes to mind is the Syncomatic plug-in from John Beardsworth (I'm sure John can tell us!)
 
Partly is the answer. Syncomatic will happily sync the keywords, but I am pretty sure that the regions aren't available to plugins. A plugin could conceivably read that information directly from the files' xmp, then do a Syncomatic-style updating of other files, but I haven't heard of any doing that kind of job.
 
That's the tool I thought I remembered, but...

Any idea, if it was done, whether it will screw up recognition? I.e. if I create a region with unintelligble data (because it's inverted), will that screw up the database of what faces look like that LR builds internally, making it less able to recognize them?

I think all that stuff is in the SQL tables, though not sure if regions are explicit or messily embedded in xml. But there's a fair chance I could just write SQL to copy them.

I just realized there's another issue -- I have to do the export without cropping (in LR) the original and without changing the pixel dimensions, or the region is not going to actually be in the same place. Oh crap... I've also got to worry about rotation, as the original will be rotated (might be) and the export will have been created unrotated.

This is sounding more and more like a bad idea.
 
I don't think it will affect recognition. LR applies keywords which are visible in the UI and to plugins, it records regions which are hidden, and it must have a table matching the regions to keywords. Rotation will be an issue for the kind of reasons you suggest, and handling portrait orientation is particularly awkward. Sure, it's just SQL underneath, but it'll be messy, very messy. Why do you think I learnt Lua in the first place?
 
I don't think it will affect recognition. LR applies keywords which are visible in the UI and to plugins, it records regions which are hidden, and it must have a table matching the regions to keywords. Rotation will be an issue for the kind of reasons you suggest, and handling portrait orientation is particularly awkward. Sure, it's just SQL underneath, but it'll be messy, very messy. Why do you think I learnt Lua in the first place?
Any idea where LR does store its accumulated data on what face X looks like?

As to Lua, I just assumed you did it so you could sit with the cool kids at the Adobe table at lunch. ;)

But i'm coming around to the idea I need to export/import these into positives. There's just too many issues.
 
Any idea where LR does store its accumulated data on what face X looks like?

As to Lua, I just assumed you did it so you could sit with the cool kids at the Adobe table at lunch. ;)

But i'm coming around to the idea I need to export/import these into positives. There's just too many issues.
Possibly really dumb question: Can you "force" Lightroom into having a facial recognition region in a negative? I'm really inclinded to keep the original inverted scanned original, just as I always keep RAW files.
 
Possibly really dumb question: Can you "force" Lightroom into having a facial recognition region in a negative? I'm really inclinded to keep the original inverted scanned original, just as I always keep RAW files.
Yes. You can draw a region and type in a name, but the recognition aspect doesn't work.

You can also just draw the region (not type in a name) and it will come up in the unnamed people section, but (I see now) it does so in the original raw. By that I don't just mean not inverted, but it is showing cropped off areas as well.

So yes, it is possible to do all this manually.

And as I worried above, I still worry if this is polluting the recognition algorithm with invalid data.
 
Status
Not open for further replies.
Back
Top