mju9's midjourney ai art thread

[psychosomaticon]

Agreed Traveler. Pandoras box has been opened there is no going back now.

And I asked on chatroom, maybe an AI category, with AI science/art/writing subcategories?

 

[acacian]

Many of these apps require a subscription fee though don't they? If these companies are making money off an algorithm that references people's copyrighted work - and doing so without the required consent to use their work, then I think that is probably illegal. Much like people get paid a small royalty when their music is played on the radio, or if somebody's work is used in a film or advertisement on TV. Companies pay good money to be able to use artist's content for financial gain.. its why people go to such pains in disclaiming that their use of someone's work is not for monetary gain. This isn't something new.. but these companies seem to be somehow evading that process. And they should be called out for it so that it can become fairer.

Surely a similar royalty system could be integrated into these apps.. even if a few dollars for every thousand uses of someone's work. There are copyright laws with this stuff.. it seems cut and dry to me that if someone doesn't consent to their work being used by a for profit app then a law is being broken. Its not that different to Spotify uploading a person's song onto the app without asking their permission/paying them for streams. Guarantee pretty much every app that wants to have such and such sound bite or copyrighted image every time the user opens it up.. they have to pay for it. Unless they are using royalty free stuff

I dunno its all very new territory. And yes it is clearly not going away.. but it can and should be tweaked in a way that is fair to those who helped make it what it is..

Its certainly healthy that it is being discussed.. even if a step far with all the name calling.

But yeah.. as there is a section for most other topics on the forums.. I think it makes sense that AI art have its own section under Music/Art/Literature rather than constantly popping up in open discussion or what have you

 

[fractals4life]

There is one means of generating AI images that requires no subscription, and also lets those fascinated by parameter tweaking to access a lot more of the variables involved rather than some "safe" presets (that ensure subscriptions keep coming in) that I daresay the commercial offerings have.

This is the "AI horde" an open source distributed network of volunteers who run machines with GPUs that take in image request jobs from anyone anywhere for free, then return the result, using web based front end. No subscription required.

More info here: AI Horde

Web interface here: ArtBot - Create images with Stable Diffusion, utilizing the AI Horde

Make of it what you will, have a play if you want, with no investment other than time.

Full disclosure: I occasionally run a worker node and, as far as I can tell, it mostly (by a huge margin) gets used to make pics of anime, or furry, characters with unfeasibly large sexual characteristics, which is kinda disappointing, but not surprising I suppose.

All I can add regarding the whole subject, is that it's not as easy as you'd think to get anything decent out of the system. There is an art to describing things properly to realise a vision of an image that you might have. It's a fun "surprising image machine" though, if you just feed in nonsense or surreal descriptions.

I wouldn't pay for the service or be able to make a profit if I did, judging by my efforts so far!
But it has allowed me, one who can't draw or sculpt, but has the urge to create, to make some cracking custom xmas cards for my friends this year.

The tendency to a lack of conformation to physical reality in the images ( because the system knows nothing about composition, or the world ) and general weird distortion of perspective that usually turns up really does lend itself to psychedelic imagery though!

P.S. if anyone likes those "happy creature made of eyeballs, smoking" images above, they could do worse than having a look at the work of Pete Loveday: Art Shop | Welcome to the Cheaper than Art online market stall buy some of his stuff if you want, the comix are great fun and he's clearly no stranger to the 'shroom

 

[psychosomaticon]

There is one means of generating AI images that requires no subscription, and also lets those fascinated by parameter tweaking to access a lot more of the variables involved rather than some "safe" presets (that ensure subscriptions keep coming in) that I daresay the commercial offerings have.

This is the "AI horde" an open source distributed network of volunteers who run machines with GPUs that take in image request jobs from anyone anywhere for free, then return the result, using web based front end. No subscription required.

More info here: AI Horde

Web interface here: ArtBot - Create images with Stable Diffusion, utilizing the AI Horde

Make of it what you will, have a play if you want, with no investment other than time.

Full disclosure: I occasionally run a worker node and, as far as I can tell, it mostly (by a huge margin) gets used to make pics of anime, or furry, characters with unfeasibly large sexual characteristics, which is kinda disappointing, but not surprising I suppose.

All I can add regarding the whole subject, is that it's not as easy as you'd think to get anything decent out of the system. There is an art to describing things properly to realise a vision of an image that you might have. It's a fun "surprising image machine" though, if you just feed in nonsense or surreal descriptions.

I wouldn't pay for the service or be able to make a profit if I did, judging by my efforts so far!
But it has allowed me, one who can't draw or sculpt, but has the urge to create, to make some cracking custom xmas cards for my friends this year.

The tendency to a lack of conformation to physical reality in the images ( because the system knows nothing about composition, or the world ) and general weird distortion of perspective that usually turns up really does lend itself to psychedelic imagery though!

P.S. if anyone likes those "happy creature made of eyeballs, smoking" images above, they could do worse than having a look at the work of Pete Loveday: Art Shop | Welcome to the Cheaper than Art online market stall buy some of his stuff if you want, the comix are great fun and he's clearly no stranger to the 'shroom

Very cool, I recommend currently Dalle3 on OpenAI and Midjourney.com, both have good monthly plans. Dalle3 is highly accurate with prompt generation as well but they both have 1080p quality output. OpenAI is 20$USD/month and Midjourney (MJ) $30 for pro plan which has unlimited mode.

 

[psychosomaticon]

More dalle3 stuff mainly

 

[The Traveler]

Beautiful!

Now I'm also very curious to the original art that was used to train this ability.


Kind regards,

The Traveler

 

[dragonrider]

Beautiful!

Now I'm also very curious to the original art that was used to train this ability.


Kind regards,

The Traveler

Yes.

Apparently people can check if their work has been used to train these machines, though those databases are probably way too large to offer any real insights.

I think it would be helpfull if people who're posting these pictures would also tell what words they'v been using to generate these images. That could maybe shed some more light on the origins of the imagery and visual styles we're seeing.

 

[CosmicLion]

I think it would be helpfull if people who're posting these pictures would also tell what words they'v been using to generate these images. That could maybe shed some more light on the origins of the imagery and visual styles we're seeing.

A persons ability to write prompts is the only skill and competitive advantage they have in this world!

I feel like it's the kind of thing that many "AI Artists" would keep to themselves so they can maintain superiority and clout for their amazing proprietary portfolios of their generated art.

Although psychosomaticon has been known to share some prompts on occasion

 

[downwardsfromzero]

[youtube]

 

[Nydex]

A persons ability to write prompts is the only skill and competitive advantage they have in this world!

That, too, is temporary. Some people are already using LLMs (GPT4, Bard, etc) to construct prompts with a pretty satisfactory result. It's only a matter of time, and not much at that, before the human is almost completely removed from the process.

 

[psychosomaticon]

A persons ability to write prompts is the only skill and competitive advantage they have in this world!

I feel like it's the kind of thing that many "AI Artists" would keep to themselves so they can maintain superiority and clout for their amazing proprietary portfolios of their generated art.

Although psychosomaticon has been known to share some prompts on occasion

Yeah if anyone wants the prompt for any of these I can go dig it up I just have giant stacks of them, wish I could figure out a way to monetize it somehow

 

[psychosomaticon]

That, too, is temporary. Some people are already using LLMs (GPT4, Bard, etc) to construct prompts with a pretty satisfactory result. It's only a matter of time, and not much at that, before the human is almost completely removed from the process.

idk if the human can ever be removed because a smaller fractal cant ever fully encapsulate the larger fractal ? idk

 

[psychosomaticon]

[youtube]

Yep I heard about that, they're trying to poison the models now, I think its kind of mean spirited =/ but then again the models scraped a lot of data without permission and they are charging users for it, and many of them are kinda stuckup

 

[psychosomaticon]

 

[basil47]

looks kind of like the biblical angels. atom without the rib removed.

 

[psychosomaticon]

 

[psychosomaticon]

frog