Jump to content

DALL-E and other AI image generation software


Klinger
 Share

Recommended Posts

Has anyone tried it yet?

 

The best answer to the quality of the photos generated is "Not ready for prime time". ūüėÄ They are 1024 PNG images. I don't know what the other do, but that's DALL-E. I have uploaded an illustration and created variations and that I find to be a better result. Uploading a photo, just gives some strange and bizarre distortions.

 

On the other side, the text tags and definitions and telling DALL-E what to create, is pretty amusing. Again the quality is more like a sketch, not a real photo, in most examples. It does a good job on some and terrible messes on others.

 

I'm not in fear that AI (machine learning) is going to replace real photos very soon, but the way technology and development goes, maybe in a few years, there will be some serious competition.

 

Just for fun:  I asked for - triple cheese burger with top and bottom bun only tomatoe onion lettuce pickels on a plain plate with rustic wood background - I got this.

 

cheeseburger-example.jpg

Link to comment
Share on other sites

Hi Pete,
I hope you're not still upset about my little joke. But on topic:

I am sick of these AI programs. It's incredible the pace at which they have developed and improved. Fortunately they are still very far from "real" photos when viewed in high resolution. But I'm sure that in 1 to 2 years at the latest they will no longer be distinguishable from real images. This will mean big changes for the MS artists.
I've been playing around a bit with Midjourney and Dall-e, a fascinating way to waste one's time....Really fascinating how an image is created from a few words hacked into the keyboard.
And I'm already using these images too, but only as blurry reworked backgrounds onto which I place my own images.
Soon, "normal" stock photos will only be made with AI, I think. Maybe this is the chance for agencies like Alamy, which have a strong foothold in news and editorials, to compete against the big three.
We'll see, the world is turning. 

  • Love 1
Link to comment
Share on other sites

I recently received an email from my ex-agency (G) stating that they will no longer be accepting¬†submissions created using AI generative models (e.g., Stable Diffusion, Dall‚ÄĎE 2, MidJourney, etc.). Prior submissions are to be removed.

Apparently, there are questions with respect to the copyright of outputs from these models and there are issues with respect to the underlying imagery and metadata used to train these models.

I don't know why I am still on G's email list.

 

Link to comment
Share on other sites

3 hours ago, Rico said:

they will no longer be accepting submissions created using AI generative models

 

Yes, these are attempts to nip the devilish seed in the bud. However, failure is already foreseeable in the medium term.
AI images can already be placed with any MS agency with a little post-processing, if you keep your imagination in check with the prompts and don't exaggerate.
And soon it will no longer be recognizable whether a picture was taken by the AI or the camera. We should get used to it early on. The agencies will certainly get used to it if it makes them money.
I think the copyrigth argument is pretextual. At least the results I've gotten so far with AI are so far removed from the trainig images that it should hardly count.
 

Link to comment
Share on other sites

It's pretty amazing once you get the hang of it but disturbing as well. I got some very realistic images that I don't feel comfortable using when I went for a photo look. Then I played with some fantasy illustrations for my grandson which felt more like my own creations which I plan to further refine so it's really my own.

 

I'm interested in a class during AdobeMAX on Wednesday called Leveraging AI to Extend Your Creative Toolkit  which says it will discuss the ethical implications of AI as well ... the works IMHO seem to be more transformative so I'm not sure that copyright violation is an issue - but the use of our photos and our keywords to "train" the AI engines without our permission or payment certainly doesn't seem right. As does pretending you drew something from scratch when a machine started it - paint by numbers anyone? 

 

Like sharks we need to keep learning ... moving forward ... or we die. I've been more into illustration lately, learning Adobe Fresco and so I hope if AI is the future I can make it work for me. I haven't spent enough time yet to figure out how to start with my own photo and go from there but I plan to learn. As long as it is just another tool, it's great but between Dall-E, midjourney and 3D modeling, the future of creativity looks very different.

 

Then again, digital photography is very different than what I learned in a high school darkroom back in 1975 - but that grounding in the darkroom still informs my work. And when I took a class in Illustrator as an adult, the instructor urged me to take life drawing and other classes to supplement what I was learning. I eventually gravitated toward photography over illustration since it dovetailed with my freelance writing, but hopefully all these skills can come into play and work together rather than just relying on a machine to do all the work. Though painters thought that's what cameras did. 

 

We'll just have to see ... it's certainly addictive...and I got some fun stuff my 4-year-old grandson loved. I think at Christmas I might have him sit with me and collaborate. It could be very entertaining. 

 

 

Edited by Marianne
Added Link
Link to comment
Share on other sites

A major MS player just announced an AI image generation initiative for both customers and contributors. 

 

Customers will submit "keywords" and get back an AI image.

 

Contributors can receive compensation when their image is used by the AI engine to generate an image.

 

https://www.shutterstock.com/press/20435

 

Stock photographer's future = feed the AI machine?

 

 

Edited by Phil
Link to comment
Share on other sites

3 hours ago, Phil said:

A major MS player just announced an AI image generation initiative for both customers and contributors. 

 

Customers will submit "keywords" and get back an AI image.

 

Contributors can receive compensation when their image is used by the AI engine to generate an image.

 

https://www.shutterstock.com/press/20435

 

Stock photographer's future = feed the AI machine?

 

 

They also bought some e-mailing list and have been spamming an offer of ten or so free images for signing up as a customer.  I asked them twice in two days to remove me and they sent a questionnaire asking what I did so that they could serve me better.   "Retiree," "Photographer," or "Expat in Nicaragua" weren't among the options.  

 

 

Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
 Share

×
×
  • Create New...

Important Information

We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue.