AI Art - Good prompts?

[ReaderPaul]

At this point I have not used AI tp generate any pictures.  I probably will not be experimenting with that in the near future, due to limited time.

Edited December 4, 2023 by ReaderPaul
Correct wrongly spelled word.

[Krista]

On 12/2/2023 at 9:15 AM, LJCC said:

That's the problem with AI. As you can see in the second picture, they look very similar.

The coding has a hard time identifying different faces and creating hands and feet.

So you're bound to have incestuous-looking same-sex models.

Although the AI mapping did create some very realistic images for me--which is on the uncanny valley side of things.

Yes, that is the problem with 'this' IA. I've seen other IA's manage a lot better. This one is severely limited, but it is a free use AI that is highly restricted, for obvious reasons. I haven't really attempted to word things specifically like, "Ordinary looking young male..." to see if I can get away from the Airbrushed model look. Most AI generated images try for clarity over flaws, like scars, stretchmarks, etc unless you ask for them.

The first picture I put.. "Average.." the one below it.. "Ordinary.." there's little difference between the two. 

-- 

[Krista]

And this is when I put in "Ugly young adult male.." I imagine his photographer telling him to make an ugly face and that's as far as he could manage.. lol. 

[Myr]

Severus Snape as Ebenezer Scrooge

[Myr]

Darth Severus

[FrenchCanadian]

You guys got me curious, which AI generator are you using? if not bing?

Edited December 30, 2023 by FrenchCanadian

[raven1]

2 hours ago, FrenchCanadian said:

You guys got me curious, which AI generator are you using? if not bing?

Good question!  I would also like the answer, please.

[LJCC]

2 hours ago, raven1 said:

Good question!  I would also like the answer, please.

seart.ai

It's limited to 20 photos per account I think.

But it's very very good with lots of options. The AI is still dumb like the rest of them, but it does create more specific choices.

 

[LJCC]

On 12/7/2023 at 11:37 AM, Krista said:

And this is when I put in "Ugly young adult male.." I imagine his photographer telling him to make an ugly face and that's as far as he could manage.. lol. 

Type: Ugly young adult male with 7 fingers per hand. 

It might generate different results.

[Krista]

20 hours ago, LJCC said:

Type: Ugly young adult male with 7 fingers per hand. 

It might generate different results.

You might get the correct appendages everywhere then. 

[Ron]

There’s an article in the Boston Globe today about a startup news service called Channel One using avatars to give personalized news to viewers. The avatars are based in digital scans of real people. I’m guessing that they only scanned faces for appearances and expressions, because the picture of the avatar in the newspaper had a thumb but only three fingers on each hand.

Edit to add: There actually was two pictures, a male and female avatar. The male’s picture was large and it was easy to see the finger configuration but when I pinched open the picture of the female to expand it the fingers were slimmer but they had the same configuration.

Edited January 3 by Ron

[raven1]

21 minutes ago, Ron said:

There’s an article in the Boston Globe today about a startup news service called Channel One using avatars to give personalized news to viewers. The avatars are based in digital scans of real people. I’m guessing that they only scanned faces for appearances and expressions, because the picture of the avatar in the newspaper had a thumb but only three fingers on each hand.

I think at least one or two of the pictures I have seen in msn.com news stories in the past few weeks were AI-generated.  Has anyone else noticed this?

[Backwoods Boy]

I need to find an organization called Image Creators Anonymous   

However, I got to playing around with sepia-tone photos from the 1930s as an offshoot of earlier gas-station-attendant examples.  I was playing around with city tenements and possible socialization therein when I got this one, which had little to do with the prompt but has turned out to be my favorite.  It might even be worthy of a story   

Edited January 5 by Backwoods Boy

[Backwoods Boy]

Probably not a candidate for a story prompt, but another experiment was the world of Canadian Northwest-Coast First Nations art.  Sue Coleman, who does paintings of animals with their native equivalents, does things like this.  Here's the prompt:

"A Raven perched on a branch of a Cedar tree looking at its Northwest-Coast First-Nations totem-pole equivalent carved into the bole of the Cedar tree, in the style of Sue Coleman. Oil painting."

This particular example didn't follow the prompt exactly, but again it turned out to be the best of the several I produced.

[raven1]

Both of these pictures are beautiful, @Backwoods Boy!  I think the sepia picture would make an excellent prompt.  Of course, coming from the same neck of the woods as you, I can greatly appreciate the second picture.  I obviously also like that it includes my avatar.

[Backwoods Boy]

9 minutes ago, raven1 said:

Both of these pictures are beautiful, @Backwoods Boy!  I think the sepia picture would make an excellent prompt.  Of course, coming from the same neck of the woods as you, I can greatly appreciate the second picture.  I obviously also like that it includes my avatar.

Thanks, Terry.  "Raven" has long been the voice that speaks to me in the woods.  I'm sure, were I Native American, Raven would be my spirit guide.  I have a Sue Coleman "Raven" picture in every room of my house.  I can put words together sometimes, but have absolutely no artistic talent, so a tool to feed text to and get a painting out is marvelous.  I'm afraid I've been spending far too much time playing with it for the last month or so - a good way to get through the season of greed and gluttony.  Time to get back to other things  

Edited January 5 by Backwoods Boy

[raven1]

1 hour ago, Backwoods Boy said:

Thanks, Terry.  "Raven" has long been the voice that speaks to me in the woods.  I'm sure, were I Native American, Raven would be my spirit guide.  I have a Sue Coleman "Raven" picture in every room of my house.  I can put words together sometimes, but have absolutely no artistic talent, so a tool to feed text to and get a painting out is marvelous.  I'm afraid I've been spending far too much time playing with it for the last month or so - a good way to get through the season of greed and gluttony.  Time to get back to other things  

I do hope those other things are a bunch of new stories! 

[Backwoods Boy]

There's a story here for sure, since it has already been written.

But there's always room for another sequel.

The prompt was the lyrics to Jefferson Airplane's, White Rabbit

As  the Cheshire Cat would say, 😁

Edited January 11 by Backwoods Boy

[drown]

After writing 30,000 words in three weeks while working, I needed a break and dabbled in some generative AI art... Hmm.

I tried making an image resembling my story Oregon in the Fall and this is the result.

Several disclaimers:

• I stand by my stance that generative AI should not replace human creation, so this is a mockup, not something I would ever use "officially" for my projects.
• It was created using DALL-E 3, with roughly 200 different prompts and iterations.
• I used an ML model to upscale the image, because the DALL-E images are only 1024x1024.
• I used several outputs from DALL-E 3 and then old-school composited them together in Affinity Photo.
• I graded the images in Affinity Photo.
• I retouched several parts, especially the wolf's face, because it looked bad.
• The text artwork and banner, moon to the right of the face, wavy sun rays are mine and thus "human generated" lol.

It's good as inspiration. I am currently working on "real" artwork that doesn't use any AI. I really don't want to sound presumptious, but I don't want AI in my text, so I don't want AI in my images either. But using it as a guide, a mockup, something to illustrate officially for myself. Sure, why not? It's fun.

And final note: All the steps above wouldn't have been necessary I guess. I just wanted something different than what the AI gave me even after extensive tries.

Edited January 18 by drown
typo

[raven1]

3 hours ago, drown said:

After writing 30,000 words in three weeks while working, I needed a break and dabbled in some generative AI art... Hmm.

I tried making an image resembling my story Oregon in the Fall and this is the result.

Several disclaimers:

• I stand by my stance that generative AI should not replace human creation, so this is a mockup, not something I would ever use "officially" for my projects.
• It was created using DALL-E 3, with roughly 200 different prompts and iterations.
• I used an ML model to upscale the image, because the DALL-E images are only 1024x1024.
• I used several outputs from DALL-E 3 and then old-school composited them together in Affinity Photo.
• I graded the images in Affinity Photo.
• I retouched several parts, especially the wolf's face, because it looked bad.
• The text artwork and banner, moon to the right of the face, wavy sun rays are mine and thus "human generated" lol.

It's good as inspiration. I am currently working on "real" artwork that doesn't use any AI. I really don't want to sound presumptious, but I don't want AI in my text, so I don't want AI in my images either. But using it as a guide, a mockup, something to illustrate officially for myself. Sure, why not? It's fun.

And final note: All the steps above wouldn't have been necessary I guess. I just wanted something different than what the AI gave me even after extensive tries.

I'm impressed by the effort you put into doing this.  Like you, I don't think AI-generated pictures are art, but they are a fun activity when needing a break. 

[Myr]

Well.. this didn't go well...

Harry looks fine... Hogwarts in the background is a nice touch.... but holding hands... hmmm....

[Mancunian]

28 minutes ago, Myr said:

Well.. this didn't go well...

Harry looks fine... Hogwarts in the background is a nice touch.... but holding hands... hmmm....

I think Harry holding Draco's head is better, it looks good. Do you plan on writing a story around the picture?

[raven1]

With Myr's inspiration, I decided to try out my fantasy pair of matched Griffendors, Harry and Charlie.  I got this on the second try with this prompt. 

Harry Potter and Charlie Weasley holding hands and looking at each other with love. Charlie is slightly older and taller. Realistic picture

 

The hands still look messed up, but I've gotten used to that.  I think the AI does it deliberately since it doesn't have hands and is envious.  

Edited January 21 by raven1

[Myr]

6 hours ago, raven1 said:

With Myr's inspiration, I decided to try out my fantasy pair of matched Griffendors, Harry and Charlie.  I got this on the second try with this prompt. 

Harry Potter and Charlie Weasley holding hands and looking at each other with love. Charlie is slightly older and taller. Realistic picture

 

The hands still look messed up, but I've gotten used to that.  I think the AI does it deliberately since it doesn't have hands and is envious.  

Charlie also seems to have stolen a tie off a short first year...

[raven1]

1 minute ago, Myr said:

Charlie also seems to have stolen a tie off a short first year...

Probably one of Harry's old ties from his first year.