When will we get AI edits?

[adamapalmer34]

Lightroom Version

12

Operating System

1. Windows 11

Would love some AI based auto editing in Lightroom. Imagen AI has been great but the workflow is clunky and it's way too expensive.

 

[Califdan]

AI assist in selection of content (e.g., select subject, select sky, etc.) is fine (IMHO), and Content Aware tools are also OK, but I'm not looking forward to AI Gnerated Content at all. I'm looking at it as a photographer and AI Generated content is not photography and is not even photo editing. However, I those who look at it as Digital Art rather than Photography may not be as scared of Generative AI as I am.

 

[Johan Elzenga]

I don’t think the OP is talking about AI generated content, but about automatic edits done by AI (like what Imagen offers). A kind of “Auto Tone on Steroids”.

 

[Zenon]

What does Imagen do that missing in LrC? Between Auto, the new masking and Denoise AI I find LrC adequate. PS is getting Firefly so that would be something you could use via LrC.

 

[BobT]

I've just read a review of a similar thing called Neurapix. It comes as a LrC plugin. We may see more of these.

 

[johnrellis]

Please add your constructive opinion and vote to this feature request in the Adobe forums:
https://community.adobe.com/t5/ligh...applies-it-to-bulk-photo-edits/idi-p/13927431

 

[Zenon]

I have been an Adobe customer for 18 years. Even when LrC went subscription I had no concerns for the last 6 years. There has been a bit of chatter out there about Firefly having an extra charge. All this advanced AI is terrific and I can see a benefit for pro editors needing to speed up developing. Even for no pros some just like to work with this stuff.

As a hobbyist that just wants to do some basic editing of a few bird shots I have no interest in Firefly or that type of advanced AI. I am very grateful for the new masking and Denoise AI. I hope to see more.

Around 2017 Adobe said that the will keep the costs down as long as membership is high. I fully realize no business can stick to any statements made. It is always changing with the times.Adobe is doing quite well but for any corporation you will never hear the statement " we are doing good enough". I'm kinda wondering to where this will wind up price-wise. These are not a few brush stroke improvements. These stuff will be more than game changing.

Sorry for going off topic. I was going to wait until the fall releases but it has been in my thoughts.

 

[PhilBurton]

I've just read a review of a similar thing called Neurapix. It comes as a LrC plugin. We may see more of these.

No doubt. Also, I would expect that there will a lot of plug-ins for Photoshop.

From what little I can glean from the Imagen and Neurapix. their initial target market is wedding photographers, and are priced accordingly. I don't do weddings, except very occasionally as a favor for a friend to supplement the work of the hired professional.

As much as the idea seems appealing on paper, I'm not sure that people who shoot a variety of photo subject matter can really benefit, until the AI gets more advanced.

At the risk of opening a big can of worms, and personalizing, I'll just say that I do "general," "family," and travel-related photography and also a lot of railroad photography (hobby). With the very wide range in lighting conditions and the variety of rolling stock paint schemes in railroad subjects, I'm not sure how much an AI can be "trained." (NO PUN!) Also, my photo library in LrC consists of a mix of my Nikon D3, and six iPhones in current use, my own plus contributed photos from my wife, kids, and their spouses taken with their iPhones. How does the AI cope with that???

And since we have all upgraded our iPhones at some point, the real iPhone model count may be around 12.

So I'm skeptical, at least for the current generation of AI-trained LrC plug-ins.

 

[johnbeardy]

I've just read a review of a similar thing called Neurapix. It comes as a LrC plugin. We may see more of these.

It's funny how their demo targets LrC, not Cloudy, and emphasizes "no more uploading into cloud services" and "your images on your computer" as selling points.

By the way, if you're going to say "read in on the internet", do give a link so others can read it too

 

[PhilBurton]

Please add your constructive opinion and vote to this feature request in the Adobe forums:
https://community.adobe.com/t5/ligh...applies-it-to-bulk-photo-edits/idi-p/13927431

Just posted a reply. https://community.adobe.com/t5/ligh...-it-to-bulk-photo-edits/idc-p/13979458#M19546

 

[BobT]

It's funny how their demo targets LrC, not Cloudy, and emphasizes "no more uploading into cloud services" and "your images on your computer" as selling points.

By the way, if you're going to say "read in on the internet", do give a link so others can read it too

I wasn't sure it was appropriate to refer to a link that may need membership to access. Here it is https://www.dpreview.com/reviews/ne...lassic-image-editing-without-lifting-a-finger

 

[Jimmsp]

I am waiting for the AI generated Keywords, where I can prompt some like "cold" & "hot" and it always gets "rose" correct.
I don't want or need AI to completely redo a photo, but I would use some smart AI local adjustments via a mask & brush that i might drive with text/voice, such as "increase the contrast, saturation and sharpening" where I could later readjust the slider values if I wanted to. I can do the single adjustments pretty well right now with the tools we have available. AI would also help if I gave it the same voice adjustment commands on an object , like "On the shoes increase ... "

 

[johnbeardy]

I am waiting for the AI generated Keywords, where I can prompt some like "cold" & "hot" and it always gets "rose" correct.

In fact that's already being done by Adobe and is used in LrWeb, Cloudy, and Mobile, though I've plenty of doubts about the quality of the tagging. Unfortunately (I'm being polite) they decided not to bring this information down into Classic Lightroom in some way that might be useful.

For a lot of my photos I feel I do have a repeatable recipe which could be learnt by a machine:

1. Hit Auto
2. Reduce its Shadows slider (Auto's is too high and artificial)
3. Reduce the Highlights
4. Zero the Vibrance and Saturation (I live in the UK, not the West Coast)
5. Find subjects and add masks to lift shadows and add a bit of clarity
6. Find sky and pull back highlights
7. Add a little vignetting

After then, things become more specific to the image.

But essentially the bulk of the editing task is already being covered by the AI/ML that's already in LR.

As a curiosity, I'll post a screenshot of a plugin that I've never released. This bit allows me to apply Auto, then adjust its results by percentages or to absolute numbers, and apply some AI-based presets.



So I wonder if these AI or Machine Learning -based plugins are proofs of concept and might not be around for long.

Anyway for me the AI I'd want most:

• Along the line of LrWeb's Best Photos, detect the subjects and key details (eg eyes) in the selected photos and tell me which are sharpest and which are soft
• Identifying and fixing dust spots in skies, still water, and other low-detail areas

 

[RobOK]

Luminar has a number of AI effects, some are well done and some are tacky (Luminar has many challenges as a company and product). All of them have a strength slider. They have for example sky replacement or object elimination. All of these things are on a spectrum now that each person has to find their comfort zone... from one side, film photography, fully manual shooting, darkroom development to the other end of AI generated imagery. Once you use a DSLR, your are introduction AI in the form of Auto Focus or Auto Exposure. The developing from RAW to Jpeg applies all kinds of mathematical wizardry. If an AI tool allows the common photographer to do what Madison Avenue has done for decades, now is that "too much". I appreciate some that want to not use such tools and I have my own personal compass of what I think appropriate for my images. Having said that, I hope Adobe continues to implement them (such as the AI masking is super useful for me) and give us choices.

My ideal AI tool -- take 15 images of a group of 30 people (my family reunion) and composite the best faces for everyone, fix the harsh lighting I had to shoot under, and make us all look 10% skinnier!

Next tool would be to hyper curate my Lightroom library. Thousands of images that are bad or where I have duplicates. And clean up where I have changed my culling approach from 5 stars to flags and back again!

 

[PhilBurton]

I am waiting for the AI generated Keywords, where I can prompt some like "cold" & "hot" and it always gets "rose" correct.
I don't want or need AI to completely redo a photo, but I would use some smart AI local adjustments via a mask & brush that i might drive with text/voice, such as "increase the contrast, saturation and sharpening" where I could later readjust the slider values if I wanted to. I can do the single adjustments pretty well right now with the tools we have available. AI would also help if I gave it the same voice adjustment commands on an object , like "On the shoes increase ... "

Reading the review of Neuropix, I liked the "after" photos more than the originals. One thing I liked was the vertical slider. I hope that Adobe incorporates a slider into their future DEVELOP AI features. Going further, I would like to see a vertical slider added to many of DEVELOP's image adjustments.

 

[PhilBurton]

Reading the review of Neuropix, I liked the "after" photos more than the originals. One thing I liked was the vertical slider. I hope that Adobe incorporates a slider into their future DEVELOP AI features. Going further, I would like to see a vertical slider added to many of DEVELOP's image adjustments.

So I realized that I should post a suggestion on Adobe's Community Forum (since they apparently ignore posts in this forum).

https://community.adobe.com/t5/ligh...evelop-image-adjustments/m-p/13989558#M331277

Please upvote it if you agree.

 

[PhilBurton]

In fact that's already being done by Adobe and is used in LrWeb, Cloudy, and Mobile, though I've plenty of doubts about the quality of the tagging. Unfortunately (I'm being polite) they decided not to bring this information down into Classic Lightroom in some way that might be useful.

For a lot of my photos I feel I do have a repeatable recipe which could be learnt by a machine:

1. Hit Auto
2. Reduce its Shadows slider (Auto's is too high and artificial)
3. Reduce the Highlights
4. Zero the Vibrance and Saturation (I live in the UK, not the West Coast)
5. Find subjects and add masks to lift shadows and add a bit of clarity
6. Find sky and pull back highlights
7. Add a little vignetting

Nice summary for overall DEVELOP work.

After then, things become more specific to the image.

But essentially the bulk of the editing task is already being covered by the AI/ML that's already in LR.

As a curiosity, I'll post a screenshot of a plugin that I've never released. This bit allows me to apply Auto, then adjust its results by percentages or to absolute numbers, and apply some AI-based presets.

View attachment 21425

This plugin sound like it could be very useful. Do you have plans to release it in the future?

So I wonder if these AI or Machine Learning -based plugins are proofs of concept and might not be around for long.

No doubt. AI and AI training is still early days. The training aspect may require significant GPU upgrades to users's systems.

Think about spreadsheets as an analogy. First we had paper. With the Apple II, we had VisiCalc, which I remember using (and hating). Then with CP/M systems we got Lotus 1-2-3 and SuperCalc (anyone remember that). Then we got Excel, which required its own version of Windows 286. Now we have an (overfeatured) Excel that is part of Microsoft Office.

Anyway for me the AI I'd want most:

• Along the line of LrWeb's Best Photos, detect the subjects and key details (eg eyes) in the selected photos and tell me which are sharpest and which are soft

Check

• Identifying and fixing dust spots in skies, still water, and other low-detail areas

Yes, for dirty sensors and especially for scanned slides.

 

[BobT]

So I realized that I should post a suggestion on Adobe's Community Forum (since they apparently ignore posts in this forum).

https://community.adobe.com/t5/ligh...evelop-image-adjustments/m-p/13989558#M331277

Please upvote it if you agree.

This forum effectively provides free customer research. If Adobe ignores it, they do so at their peril. I suspect their competitors don't.

 

[PhilBurton]

This forum effectively provides free customer research.

I've said this several times over the years. Once I even comlied some lists that could be provided to Adobe. But I don't know the outcome.

If Adobe ignores it, they do so at their peril.

True enough.

I suspect their competitors don't.

In my experience doing product management as an employee and then as a consultant competitive analysis is way down on the priorities list. I would hope that Adobe competitors, for their own good, are surveying their customers.

 

[Johan Elzenga]

A highly successful company like Adobe will probably know how to do customer research and do that frequently. The main problem is that every individual customer thinks they represent the majority of customers, so if Adobe does not prioritize exactly what they would prioritize, then it is obvious that Adobe is not doing that research properly.

 

[johnbeardy]

Anyway for me the AI I'd want most:

• Along the line of LrWeb's Best Photos, detect the subjects and key details (eg eyes) in the selected photos and tell me which are sharpest and which are soft
• Identifying and fixing dust spots in skies, still water, and other low-detail areas

I'd obviously not been keeping my eye on CaptureOne as much as I used to, because in their May update:

• "Auto Dust Removal is the quick and easy way to take care of pesky dust spots in loads of photos at once. In seconds, you can automatically remove dust spots and dramatically speed up your editing process. You can even remove dust spots on your photos directly during import by enabling it under Auto Adjust."
• "Let Face Focus instantly reveal facial sharpness and expression. In the Import and Cull viewers, our new Face Focus gives you an instant close-up of your subject’s face. This allows you to select your sharpest images from the start even faster without needing to manually zoom in on every photo."

As their headline put it: "Let AI do the dirty work".

 

[Zenon]

I'd obviously not been keeping my eye on CaptureOne as much as I used to, because in their May update:

• "Auto Dust Removal is the quick and easy way to take care of pesky dust spots in loads of photos at once. In seconds, you can automatically remove dust spots and dramatically speed up your editing process. You can even remove dust spots on your photos directly during import by enabling it under Auto Adjust."
• "Let Face Focus instantly reveal facial sharpness and expression. In the Import and Cull viewers, our new Face Focus gives you an instant close-up of your subject’s face. This allows you to select your sharpest images from the start even faster without needing to manually zoom in on every photo."

As their headline put it: "Let AI do the dirty work".

That is what Auto does for me using LrC. Between it and adaptive ISO presets I edited about 500 files in just over an hour for a charity shoot. Now with Adobe Denoise AI I have the best of both worlds. Since Denoise AI I don't even apply my Adaptive ISO presets at import anymore unless it is a mass edit. I apply it per file but these days 90% of my hobby files go through Adobe Denoise AI. Even low ISO files. Why not and I have the time.