Did you hear the news?

There’s this new fantastic AI tool in Photoshop that is going to… Blow. Your. Mind. Or is it?

Every top creator has been saying roughly the same thing about Photoshop’s new AI technology, which includes a Midjourney-like app directly in Photoshop they call Firefly.

They’re saying things like OMG, this is the best thing ever, blah, blah, blah.

Well, I’m not so impressed. I think it’s the WORST Photoshop update for photographers EVER.

Yes, it’s just Beta, and it will get better in time. But, the art of photography is doomed.

However, even if the results were perfect one hundred percent of the time, I don’t see myself using it.

Why? Well, I’ll let you know why at the end.

Adobe Didn’t Tell You This…

You need to know something critical about Photoshop’s Generative Fill that no one talks about.

And that is, Adobe is stealing your photos. Well, maybe not stealing, but they are analyzing and using your images to improve Generative Fill.

You probably didn’t know that, did you?

Here’s a “help” article that Adobe quietly put out that basically says… that Adobe performs content analysis on content processed or stored on Adobe’s servers. In other words, to use Generative Fill, you must be connected to the internet so Adobe can see and analyze your image to give you your desired results and improve their technology.

That last part, “or stored on Adobe’s servers…” is quite interesting, isn’t it? Does that mean if you’re using their Creative Cloud photo storage, they’re analyzing those images too? That’s how I read it. How ’bout you?

Adobe didn’t ask for your permission to analyze your photos. You’ve already been automatically enrolled!

You didn’t know that? Well, check this out…

…go into your Adobe account, navigate to Account and Security, click on Privacy and personal data, and under Content analysis, it says… “Adobe may analyze your content using techniques such as machine learning to develop and improve our products.” Wow, so they are using our images without our initial consent!

To disable consent, click on the switch.

Adobe content analysis

Adobe content analysis approval

For transparency, Adobe has said the following…

“Adobe takes your privacy seriously. Your Privacy is maintained during content analysis for product improvement and development. None of your content is included in our products or services unless you make them public.”

Ok, fine, but I still don’t wish to have images of my kids analyzed. Or my paying portrait and wedding clients since they haven’t given me permission for this type of use via the contracts they’ve already signed.

Generative Fill Low Res Output?

The next reason I’m not a big fan of Generative Fill is that, at this time, results are limited to a resolution of 1024×1024, meaning the output is only suitable for online use. And this is understandable when you factor in what it probably costs Adobe to run this AI technology on their servers and the fact that it’s not costing us anything extra… yet.

Does anyone honestly believe that Adobe will eat the costs of running the servers once this becomes universally available? I doubt it since their costs will probably increase once they remove the resolution limits and more photographers have easy access to it. How much is Generative Fill going to cost? Another ten dollars per month, twenty, fifty? Or will there be a tier system where you can process “x” images per month?

Also, I’ve read that the content they’re sourcing from is through their Adobe Stock agency. And they plan to compensate photographers for having their images used with this AI technology to ensure copyrights are not infringed upon. Enjoy it while it’s free!

Generative Fill Produces Horrible Results!

The last reason I’m not a big fan of Generative Fill is because it doesn’t suit my current needs.

Plus, I guess you could say I’m an old-school photographer. And I have no interest in being a digital artist. I elaborate a little with images from some of the top Photoshop instructors via the video tutorial below.

Russel Brown from the Adobe Photoshop channel

I’ll admit, this composite from Russel Brown that he created with Generative Fill is pretty awesome.

Aaron from Phlearn

Aaron from Phlearn also showed off several images and how easy it is to add content. Although the results are somewhat impressive…

…to me, once you add a photo or two or more and place them into another photo, it’s a composite. Not a photograph!

Generative Fill is probably a dream come true if you’re into creating composites or digital art. I have no desire to do these types of composites.

Change Background

Aaron also showed how easy it is to replace the background. Again, to me, not that impressive.

If I was commissioned to create a portrait with a dark, moody feel in a forest, guess what…

… I’m going to shoot that on location. I find no joy in compositing another image to achieve what is possible while shooting.

Remove a Person

For this image, Aaron showed how easy it is to remove someone from a photo.

This is a little more practical. Although, if I were doing street photography, a commissioned portrait, or a wedding shot, I would do my best to get it right in camera…

  • by either asking the person in the back to move
  • or reposition my subject
  • or change my position and/or perspective… to name a few

Changing Clothes

What about changing someone’s attire?

I saw Unmesh over at Piximperfect do a demo of changing outfits, and although the results were decent, I think he chose a safe image to demo Generative Fill.

The demo by Colin Smith at photoshopCafe… is a little more complex, and the final results are awful.

Canvas Fill

The one feature I could benefit from in Generative Fill is using it to expand a Canvas.

Sometimes, when shooting wildlife, I’m presented with a situation of capturing the moment or skipping it.

This was the case when photographing this juvenile Burrowing Owl that was trying to take off for the first time.

This was in a field next to some homes, and there was a PVC pipe sticking out of the ground with a warning sign about not disturbing the owls.

I’ll probably use Generative Fill on less than one percent of my images.

Original image prior to Generative Fill. PVC pipe is clearly visible.

Original image before Generative Fill. PVC pipe is clearly visible.

Canvas expanded and filled with Generative Fill. Not bad!

Canvas expanded and filled with Generative Fill. Not bad!

Generative fill is not perfect.

In the end, I guess I’m old school and like to get it right, as much as possible, in camera, which I’ve done for the past thirty-plus years.
For those needing this technology more than myself…

… you’ll still have to master Color Grading, dodging and burning, and more to get the best results.

At least based on my testing and what I’ve seen from other creators.

I’m a dinosaur

Back in 2001, I embraced digital photography. Twenty-two years later, I’m now a dinosaur. The “Art of Photography” is becoming a thing of the past with each passing year, and with the advancement of AI technology, it’s, well, who knows what photography will look like in five or ten years.

About the Author

Chris Parker is a photographer and educator based in Hollywood, CA. You can see more of his work, and learn quite a bit on his website here. This article was also published here and shared with permission.

Source link

Leave a Reply

Your email address will not be published. Required fields are marked *