Is it wrong to use AI generated art for Bandcamp projects?

[Fez909]

I will do. The one thing I think Midjourney is missing is themes. If I have a character in a plot I want them to look the same in different situations or it doesn't make sense. Which limits narratives somewhat.

You can feed midjouney images to do a sort of mashup. I've not really played with it, but you might be able to give it a character along with the scenario to get a more consistent look.

I think stable diffusion can do this, too. Again, never tried though!

 

[Magnus McGinty]

You can feed midjouney images to do a sort of mashup. I've not really played with it, but you might be able to give it a character along with the scenario to get a more consistent look.

I think stable diffusion can do this, too. Again, never tried though!

I've got some credits in Stable Diffusion but mostly because other software has opted to use that for paint tasks. I haven't used it as a standalone thing. I've got a sub with Midjourney and a sub with Runway. I'm pretty much bleeding money . Maybe I need to take stock and research what's best nowadays but I assumed it was Midjourney for images.

 

[Fez909]

I've got some credits in Stable Diffusion but mostly because other software has opted to use that for paint tasks. I haven't used it as a standalone thing. I've got a sub with Midjourney and a sub with Runway. I'm pretty much bleeding money . Maybe I need to take stock and research what's best nowadays but I assumed it was Midjourney for images.

I think this is the current state of things:

Midjourney is/was the best. Paid.
Dall-E 3 has just been released and is apparently amazing - maybe better than Midjourney? It's free via Bing's AI, or Paid via ChatGPT subscription (and not yet available to everyone, but will be in the next two weeks).
StableDiffusion - There's lots of different flavours of this, but it's mostly free. But you'll need to faff around a little bit to get good results. It's definitely the more technical one of the three, and requires a beefy computer to get fast results. If you have a slow one, it'll work just the same, but take a long time.

If you're on a Mac, there's Diffusion Bee which makes running SD much simpler. Just download an app and run it, though you'll need to tweak some parameters to get the best from it: DiffusionBee - Stable Diffusion App for AI Art

If on Windows, I think the closest equivalent to DiffusionBee is AUTOMATIC1111: GitHub - AUTOMATIC1111/stable-diffusion-webui: Stable Diffusion web UI

It's quite technical to setup, but will probably give the best results overall (after tweaks, of course).

DiffusionBee is apparently coming to Windows so might be worth keeping an eye on that if you want something that 'Just Works' without any hassle.

 

[Karl Masks]

It's legal, but as far as I know you can't claim copyright on the images. Which might be problematic down the line.

eta ChatGPT: what the law says about who owns the copyright of AI-generated content

'Copyright law is based around a general principle that only content created by human beings can be protected.'

Given that I've been using royalty free pics so far it's not really a big deal. Of course, in the round, it could be. But anyone could have used any of the pics i've used thus far for anything. I don't own them

 

[Karl Masks]

Just checking..

This js what I am seeing.. is this correct?

View attachment 393743


Eta..I've had a listen and have an idea. Sent you a pm.

Great music!!

Thank you

I've got some credits in Stable Diffusion but mostly because other software has opted to use that for paint tasks. I haven't used it as a standalone thing. I've got a sub with Midjourney and a sub with Runway. I'm pretty much bleeding money . Maybe I need to take stock and research what's best nowadays but I assumed it was Midjourney for images.

I've been using stable diffustion for free thus far. It comes up with good results. Real art by humans would be better, when i can affords it

 

[pug]

You could take a photo then manipulate it with photoshop, something like that anyway.

 

[Karl Masks]

You could take a photo then manipulate it with photoshop, something like that anyway.

i don't think so. The only camera I have is my phone and it's a shitty camera. I don't want to do that

 

[Aladdin]

Are you creating art when you get an artist to make it for you?
If the artist is a digital artist does this change things?

I'd say creating a prompt is not the same as making art.

The art bit if the art comes from the model the ai was trained on.

A machine is fundamental different from a personal but if they take the the exact same role do you need to consider them an equivalent?

I had thought @Karlmasks was actually making his own AI art.

What I am doing is painting a picture ...and if @Karlmasks wants to put it through an AI programme after that I am perfectly ok with it.

 

[Karl Masks]

i was referring to using one of these ai systems to generate art through a prompt.

 

[Magnus McGinty]

i was referring to using one of these ai systems to generate art through a prompt.

You can feed an image into midjourney whereupon it gives you prompts it thinks matches the image.

 

[Shippou-Sensei]

I had thought @Karlmasks was actually making his own AI art.

What I am doing is painting a picture ...and if @Karlmasks wants to put it through an AI programme after that I am perfectly ok with it.

I was posing those as more reflective questions about the nature of AI art rather than asking about anyone's specifics.

 

[Aladdin]

I was posing those as more reflective questions about the nature of AI art rather than asking about anyone's specifics.

Oh ok...

 

[Karl Masks]

You can feed an image into midjourney whereupon it gives you prompts it thinks matches the image.

how very meta!

 

[Magnus McGinty]

how very meta!

Not saying it will give you the results you want but it’s a starting point if you don’t know what prompts to use.

 

[Karl Masks]

Mega thanks to Aladdin for his sterling work improving muh bandcamp page.