AI generated art

So I have found a new favorite technique. If you turn on remix mode for the prompts you can edit the prompt between generations of the AI renders. By completely rewriting the prompt but ask for a “subtle” variation, the AI will try and shoehorn new information and context but keep the composition intact in terms of color and value.

Example:

let’s start with:

prompt: look at the size of my hat, you filthy peasants!

Then I ask for a subtle variation but edit the prompt to a snippet of Alice in Wonderland.

prompt: “It is the same thing with you,” said the Hatter, and he poured a little hot tea upon its nose. The Dormouse shook its head impatiently and said, without opening its eyes, "Of course, of course; just what I was going to remark myself."

Which brings back some of that old school surrealism from earlier versions of AI that mid journey usually tried to downplay in hopes of becoming a universal stock image generator.

2 Likes

:wave:This was the first time I used ai for the cover art for one of the watt radio broadcasts, I also used photoshops ai to make the upper part of the visual longer. Also did some color modifications in photoshop
Ps. The logo isn’t ai, it’s our logo i made a few years ago

https://wearetimetravelers.com/we-are-time-travelers-watt-10062023-playlist/

6 Likes

Love it!
Why are the last three characters distorted?

2 Likes

to visualize the possible time travel that may occur when listening to our broadcasts :joy: like traveling close to the speed of light, starting to experience weird bending and warping effects :alien:
Thank you :heart:
Ps. The logo isn’t ai, something I made a few years ago

1 Like

Some hallucinated album art for the fanz:

4 Likes

no disrespect to the ai prompt creators, i appreciate some of you guys put in some extra graft to add your own spin: but as the technology improves and they get more detailed and refined it just gets more boring and soulless to my eyes. preferred the abstract inaccuracies of earlier models.
probably cynical and jealous i don’t have a mj account to test out my own random promptage.
stuck with free dall-e credits waft.

5 Likes

I feel you.

Despite being impressed by the latest developments I feel like we didn’t spend enough time in the dreamy aesthetic territory of VQGAN and the like.

2 Likes

It’s funny how retro has become a few years ago as things speed up :smiley:

2 Likes

A few months ago even!

More on instagram #fuzzy_ocean

10 Likes

It’s like you are daring uss to post our weirdest midjourney AI pics


Honestly it’s a constant barrage of the uncanny. It’s especially funny when you discover something that midjourney does not know how to render, for example eye patches.



Dall-E is still pretty awesome though.

4 Likes

vintage MJ july 1st 22 darkness. i like it when it goes completely screwed up. haven’t seen as much of this kinda deal lately.

logged back in and found the prompt, for research purposes if anyone fancies seeing what MJ does over a year later i’d be interested to find out! (if you want to correct the spelling of extravagantly please do, not sure how much impact on the result tho…)

Nose hair eyebrow hair clumps clotted notted head nits infestation tesselate extravegently facial features mutation rotting beef infusion cyborg

4 Likes

very true, here is the same prompt ver 5.2

and then for fun ran the prompt again with the --v 1 tag to use the older model and got some ver different results:

conclusion: I should try out the older versions more.

6 Likes

yawn.

yes. with some upscaling that’s the vibe i love. disturbed smashing of prompt. you can see the infusion element, rather than just focussing on cyborg as the be all and end all.

thanks for that. as mentioned, it’s got too boring and pandering to generic expectations.

2 Likes

Used a mix of AI stuff with Blender & After Effects for Nathan Fake’s tour visuals:

https://twitter.com/Infinite__Vibes/status/1706634074603864562

3 Likes

New music vid that I used 15 different AI techniques on just dropped:

11 Likes

Spill the beans, what are the secret techniques!? :slightly_smiling_face:

1 Like

SDXL, SD-CN, Gen-2, AnimateDiff, SadTalker, ModelScope text-to-video, Warp Fusion, Deforum, Photoshop Generative Fill, Infinite Zoom A1111, Runwayml automatic background removal, FILM interpolation and Topaz upscaling.
:upside_down_face:

2 Likes

Sounds like I need to branch out from midjourney to try more things!

1 Like