AI (Artificial Intelligence) Creators Club!

[-Asdfgh-]

Pink Floyd weren't called agency users because that would be silly.
No one who uses an agency calls themselves a creator or artist and they don't take credit for the art. The artist is credited separately. Not so with AI prompters, who present the images as their creations and take credit for them and, like in this thread, call themselves creators.

My point is that some seem to be denigrated for using AI in situations where others used agencies, and often for just part of their output. So someone might have written a song but used an arranger and were not considered to be less creative for it but people who use AI for that function are seen as somehow less creative for it now. It's a double standard.

 

[-Asdfgh-]

In order to make your point, you conveniently cut out the part of my comment where I said art can also be a product. Quite deceiving editing. You would make a great journalist.

I did not intend in any way to deceive, I just missed that part of your comment. It was incompetence not malice.

 

[JRA]

amazing except for the fingers, bass, strings, headstock, strap, devices in the back

well then, you're really gonna freak when you see some of the stuff --- nothing like 'reality' at all --- not even close.

 

[Low Down Brown]

per the AI generators: what is it with the AI generated hands? they generally seem mangled, with extra or deformed digits --- i'm not complaining i think it's funny (part of the fun!) --- anyone's guess?

I see you're using NightCafe as I am. Negative prompts! That is helpful for cleaning up all the crazy extra fingers and limbs. There are several tutorials on YaTube about how to use them, but essentially, you just type in another prompt using words to describe what you DON'T want (e.g. extra limbs, blurred, extra hands, etc.) Then you have to adjust the weight of the prompt to the negative. -3 is a good starting point. Also, using multiple positive prompts can trip it up, but I've been using a series of 3 positives (all weighted differently in descending order of importance) and one negative to clean up the artifacts. It seems to work really well.

 

[-Asdfgh-]

@asdfgh, of all my philosophical queries, leanings, and hills-to-die-on, i'm afraid that i can only apologize for my uncluttered, lightweight position on the subject of AI, in general, but especially and specifically as it relates to this club/thread --- i just wanna see what others are doing (can do) with AI images, and i want to see how it's applied to some TB member-generated music --- the good stuff!

Yes, perhaps I am getting too much into the semantics. They are tools which are immature but may still have some level of usefulness. My experiences so far have been very hit-and-miss. The advantage for those who don't have a lot of time or funds is that it is quick and cheap, especially for things that are outside their area of expertise.

Where I would worry is me using them as a crutch for not nurturing my own creativity. That's probably stifled by having to spend time unclogging the washing machine, fix bass necks, mow the lawn, etc., during which I find I can't hum a tune and later remember it and use it to write a symphony. The other is not prioritising time I do have to listen to music that might inspire me. I think AI might be useful for suggesting things to listen to, but digesting and absorbing things needs immersion that can be difficult to achieve. So then AI looks attractive as an aid to create seeds of ideas but perhaps I'd be better with a robotic lawn mower, digital assistants to make sure other tasks have done, etc? Or AI aids to allow me to quickly hum ideas? Some AI tool that could read my handwriting and turn into text files would also be good, but no AI is currently strong enough to do that so it's hard to create a searchable library of ideas. Even I can have issues reading my handwriting.

 

[-Asdfgh-]

I did not intend in any way to deceive, I just missed that part of your comment. It was incompetence not malice.

Ironically, AI might have done a better job of reading the whole post I originally replied to.

 

[-Asdfgh-]

Yes, perhaps I am getting too much into the semantics. They are tools which are immature but may still have some level of usefulness. My experiences so far have been very hit-and-miss. The advantage for those who don't have a lot of time or funds is that it is quick and cheap, especially for things that are outside their area of expertise.

Where I would worry is me using them as a crutch for not nurturing my own creativity. That's probably stifled by having to spend time unclogging the washing machine, fix bass necks, mow the lawn, etc., during which I find I can't hum a tune and later remember it and use it to write a symphony. The other is not prioritising time I do have to listen to music that might inspire me. I think AI might be useful for suggesting things to listen to, but digesting and absorbing things needs immersion that can be difficult to achieve. So then AI looks attractive as an aid to create seeds of ideas but perhaps I'd be better with a robotic lawn mower, digital assistants to make sure other tasks have done, etc? Or AI aids to allow me to quickly hum ideas? Some AI tool that could read my handwriting and turn into text files would also be good, but no AI is currently strong enough to do that so it's hard to create a searchable library of ideas. Even I can have issues reading my handwriting.

P.S. so far my forays into AI tools for harmony, drumming along with things or generation of things from prompts have been UNinspiring to poor. There was one tool that I now cannot remember that was OK at creating some atmospheric soundtrack stuff but I think that was more randomisation based on pre-recorded loops than AI. But after going on my semantic tangents I should shut up, see what people are using, give them a go, and report back here in a constructive manner.

 

[CryingBass]

That’s a great point. AI is a time sink.

Put a value to your time my lads. Life is too short.

will AI make it shorter? Only time will tell…

 

[JRA]

I see you're using NightCafe as I am. Negative prompts! That is helpful for cleaning up all the crazy extra fingers and limbs. There are several tutorials on YaTube about how to use them, but essentially, you just type in another prompt using words to describe what you DON'T want (e.g. extra limbs, blurred, extra hands, etc.) Then you have to adjust the weight of the prompt to the negative. -3 is a good starting point. Also, using multiple positive prompts can trip it up, but I've been using a series of 3 positives (all weighted differently in descending order of importance) and one negative to clean up the artifacts. It seems to work really well.

my one and only 'production' is full of trippy, deforming, colorful, moving shapes. and i still like it. a bass player with 7 fingers just doesn't bother me like it might some.

i have yet to learn/use that technique ("negative prompts"), you're way ahead of me. you can bet i'll be calling for your advice! thanks!

 

[ThinCrappyTone2]

I see you're using NightCafe as I am. Negative prompts! That is helpful for cleaning up all the crazy extra fingers and limbs. There are several tutorials on YaTube about how to use them, but essentially, you just type in another prompt using words to describe what you DON'T want (e.g. extra limbs, blurred, extra hands, etc.) Then you have to adjust the weight of the prompt to the negative. -3 is a good starting point. Also, using multiple positive prompts can trip it up, but I've been using a series of 3 positives (all weighted differently in descending order of importance) and one negative to clean up the artifacts. It seems to work really well.

Great tip!

 

[JRA]

i need a tip re: nightcafe which does not take my use of numbers seriously. e.g., i want an image that shows three players, but whether i use the word "three" or the numeral "3" or both, and whether or not i use any qualifers "only three" or "exactly three" or "no more than three," the images all show up with at least four (usually five or more) figures.

any tips for getting the 'right count/amount' of figures (i.e., three/3) in nightcafe? TIA

FWIW: i've had good results getting one figure/player, but only after qualifiers such as "one lone..."

edit: when i used the term "exactly two..." the image was correct, first time....is there something about "three?"

 

[Barticus]

Like this?
Using your prompt:

Three handsome young man wearing a huge red sombrero and a black tuxedo while playing a tenor saxophone in the desert, Hyperrealistic, splash art, concept art, 3/4 shot, intricately detailed, color depth, dramatic, side light, colorful background
​

 

[Low Down Brown]

i need a tip re: nightcafe which does not take my use of numbers seriously. e.g., i want an image that shows three players, but whether i use the word "three" or the numeral "3" or both, and whether or not i use any qualifers "only three" or "exactly three" or "no more than three," the images all show up with at least four (usually five or more) figures.

any tips for getting the 'right count/amount' of figures (i.e., three/3) in nightcafe? TIA

FWIW: i've had good results getting one figure/player, but only after qualifiers such as "one lone..."

edit: when i used the term "exactly two..." the image was correct, first time....is there something about "three?"

Multiple positive prompts (even in the same prompt box) can cause crazy stuff to go down. You could try separating it into 2 separate prompts and adjust the weight.

E.g. Three handsome young man wearing a huge red sombrero and a black tuxedo while playing a tenor saxophone in the desert - first (main) prompt weighted at 1

Hyperrealistic, splash art, concept art, 3/4 shot, intricately detailed, color depth, dramatic, side light, colorful background - second prompt weighted at .5 to .7

Then you can mess around with the overall prompt value (which defaults at 50%) and raise or lower it to see how it responds. I usually set mine at 60% and adjust the prompts down to keep it closer to what I'm after. Seems to work ok.

Also, you could try the negative prompt using phrases like 4 or more people, 1 or 2 people, disfigured, blurry, out of focus, extra limbs, etc. Then you would want to weight it at around -3 to -5 and see how it responds.

 

[JRA]

thanks, @Low Down Brown. because of your advice/guidance, i'm just now seeing the 'add another prompt' feature. between old eyes, old brain, and 'hurry up and get me a pic' i missed that all along. so a huge thanks for being instructive and patient.

i'll try it! thanks again.

 

[JRA]

@Barticus --- yeah, man, that's it!

so what other settings/prompts (i assume nightcafe?) did you use?

(also, i downloaded that pic. i won't use it if you'd rather i not.)


edit: there are so many variables involved that my questions end up being too general, i.e., i just have to experiment! sorry for the use of bandwidth on this thread just to teach me what i need to know.

 

[FRoss6788]

My point is that some seem to be denigrated for using AI in situations where others used agencies, and often for just part of their output. So someone might have written a song but used an arranger and were not considered to be less creative for it but people who use AI for that function are seen as somehow less creative for it now. It's a double standard.

The arranger is generally credited in the liner notes of the recordings.

 

[Barticus]

@Barticus --- yeah, man, that's it!

so what other settings/prompts (i assume nightcafe?) did you use?

(also, i downloaded that pic. i won't use it if you'd rather i not.)

I used Bing Image Creator (powered by Dall-E) to create this one using your prompt verbatim… it gave me 4 versions, it’s yours to use or just recreate.

 

[-Asdfgh-]

The arranger is generally credited in the liner notes of the recordings.

Indeed. No one seems to grumble about people using them. How do you credit AI?

 

[FRoss6788]

Indeed. No one seems to grumble about people using them. How do you credit AI?

That has kinda been the point, AI being used but not "credited" somehow, and the "creator" taking credit for the "art".

No one grumbles about artists using arrangers because they are giving the arrangers the credit. Seems fairly straightforward to me...

 

[Barticus]

@Barticus --- yeah, man, that's it!

so what other settings/prompts (i assume nightcafe?) did you use?
(also, i downloaded that pic. i won't use it if you'd rather i not.)

edit: there are so many variables involved that my questions end up being too general, i.e., i just have to experiment! sorry for the use bandwidth on this thread.

Okay, played for another 10min with your prompt verbatim in Adobe Gen AI which uses their stock image library as the source images... all very abstract of course

then some selective prompts to adjust faces on this one, makes it a little more interesting.