With ~2000 entries, I was surprised by the patterns that emerged in submissions--themes that told a story of things we're all collectively experiencing. I've broken these into categories. Hope to keep that context for gallery exhibitions.
1) A reminder that Transformative Use has been mainstream for 60 years, by @mg_deangelis 2) A case for emotive, realistic hands by @d0vetaiI 3) A question of art in any medium, by @giokarloart 4) A physical, non-AI piece depicting the start of a prompt, by @Dorkholm
"AI is theft"
"Forward," by @EclecticMethod. A Transformative Use remix of split-second film clips, his own music, & frame by frame AI animation.
Like all art, its shaped by its influences, deepened by their subtext, & transformed to express something entirely new.
Forward.
"AI Art doesn't say anything"
These say how it feels to be an AI artist in this moment, before what comes.
From the pulse of that feeling, @womancalledwild used ChatGPT & text-to-image to create a manifesto for AI art--in the style of Martin Luther's 95 Theses. She then made it an AR filter to be virtually tacked at the door of museums & galleries. Guerilla art. Innovation.
Forward.
"You just typed a few words."
Speaking of innovation, @nptacek made a cityscape with AI, asked ChatGPT to write a script to port it to a VR headset as an environment, iterated until it worked, & stepped inside. They've now made many more.
As they says: "Words create Worlds."
"I can always tell when it's AI."
Maybe, for now. As tools rapidly evolve, artists are creating increasingly realistic work. Post-photography is new, but before long, will be indistinguishable from photographs.
This is by @scizors_eth. They dance to dissonant words+music stage center, the sides a remix of my art (by @EclecticMethod), transformed by AI that imitates knitted wool. It's bizarre, fascinating, nouveau dada.
@kreaturecastle's video offers a glimpse into their process. There are as many processes as there are artists. All personal. All valid. All none of your business.
To any hate-reading, please know: this phrase is a self report.
On types of skill:
1) A hand-made woodblock print combined with AI @sonyamova 2) Data visualization combined with AI @DataVelvet 3) Multilingual poetry combined with AI @marta_gazzola 4) plaster cast sculpture combined with AI @NoortjeStorteld
One pattern in submissions was using AI to navigate grief and recreate lost memories.
@redactedpride lost all photos but one of her late grandmother in a house fire. She combined it with one of her mother, adjusted from memory, and made this portrait.
1) @vulcanstudionft remembers a loved one lost during the pandemic. 2) @Dvyntwoman creates her late grandmothers favorite colors/flowers. 3) @elnaztaassob's memory of her mother's grave, through the lens of her love for her. 4) @PhoenixFawkes25 recalls schoolmates lost to time.
Disabled artists (including visually!) using AI to create & connect again. The overworked discovering time to express themselves. Victims of injustice finding righteous anger. These were not the only ones.
All entries--funny, angry, forlorn, loving, etc--felt personal. VERY personal. Personal in the way you speak to yourself in your head, but don't share w/o sanitizing.
The pattern: speaking from the heart to something that listens, & reflects you back to you. (@1DigitalCrypto)
Winners were impossible to choose, but I had to. Finalists will also be exhibited, and those on my website will have options to exhibit if they like.
There was more worthy work than spots for finalists. I made a gallery on my website for more highlights from Contest 3--each picture, when clicked, links back to your post on Twitter. I'm still adding more, will be up in a few days.
Let me know if you'd prefer yours isn't shown.
Thank you to everyone for submitting, and for reading. I hope you've gained something, like I have, from it. The website will be finished within a few days, will post it then. (Closing piece by @LimnDigital)
See you next time.
• • •
Missing some Tweet in this thread? You can try to
force a refresh
) & say: "But Claire, my video doesn't look like your promo! Well dear reader, that's b/c the model used for mine is one you don't have yet! It's Analog Diffusion,trained on photography, and can be found below
🌟Lets go!🌟
First tut thread was science, this one is art. Stick with me--Let's have some fun!
Click CKPT DOWNLOAD. Once done, open the google drive you're using for this tut, then open "AI," then "models," & drag AnalogDiffusion CKPT to the models folder.
🌟Got 30 minutes to make your first AI animation? This thread is for you!🌟
STEP BY STEP AI ANIMATION TUTORIAL: THREAD ONE
(link in the last tweet for thread 2, which is a deep dive of tips and tricks!) #AIArt
This thread looks complicated, but is written for anyone to be able to do. If you follow this, you'll end up with an animation--today! Hang in there, it's not hard, just wordy, and I promise: you'll be glad you did.
You've got this!
Okay, Ready? Let's go!🌟
Step 1) Create a fresh gmail. This is important, as when you explore more tools in the future, you want to be sure your primary google drive is protected if any are untrustworthy! 🕵️♀️
NYE resolution to train AI? This tutorial thread is for you.
This is the "easy" version, which requires more VRAM--I'm aware there are alternatives.
If you don't want to train your own models, just make AI art with other peoples models (much easier & free), skip to step 3.
1) Figure out how much VRAM you have--& make sure its Nvidia. You can find this on windows by going to device manager, display adapters, then googling the name of your GPU to see the VRAM number.
2) Visit paperspace.com (core), or lambdalabs.com (gpu cloud) & get a virtual machine with a GPU of (imo) >24gb of VRAM. This starts from ~.80 cents an hour, so 3 hours a day costs a medium coffee. The more VRAM, the faster the images & model training.
In the interest of pushing AI art's aesthetic by increasing range, here's a visual thread of tips, tricks, and expanded possibilities for several AI-Collaborative tools.👇🌟
1) Dalle2. Dalle has exceptional range, and is great at creating spaces and filling them with details. Currenrly less good at faces.
Tips: (both using a Midjourney piece) 1) You can upload your own images to create variations. 2-3) You can use inpainting to create new sections.
2) StabilityAI. Stable Diffusion is great at faces, figures, photos. Very high fidelity. You can adjust weights, aspect ratios, sampler models, steps, etc in your prompt. You can't yet upload your own images.
1 & 2 are the same prompt w/ the exception of stylization vs realism.
🌟AI-Art Contest🌟
* Prize: 3 ETH for the winner, bids for others I love
* Time-frame: 8/9 - 8/15 (PST)
* Theme: Use AI (alone or in mixed process) to make a piece that looks unique. Mix your favorite influences to make a style that feels like you. Push your vision with the tool.
If you don't have the funds to mint, you dont have to--you can post the image to the final thread on the 15th & only mint if it's chosen. That said, feel free to mint if you like! Use any or all AI models, though I do own a lot of pure Midjourney, and am looking for other models.
This is an extremely transparent, extremely vulnerable thread about my personal current struggles with scarcity. It’s long (18 tweets). Please read it anyway. I’m not the only artist (or collector) who is affected by these concepts, & we don’t talk about them.
I recognize not all artists will resonate with this thread. I’m very privileged to have these worries, but I know there are many artists and collectors who also struggle with them, and many more who will, given time. This is my current experience, & I welcome hearing yours.
Several months ago I was offered a handful of opportunities. They were incredible, and had a time-span that wasn’t yet fully defined, so I took them. When you’re an artist trying to grow–for yourself and for the benefit of those that collect your work–that’s what you do.