Artist behind “Life In West America”, "Reworld" and "Smile" NFT collections. WME represented. Designer, creator, photographer, screenwriter, endless learner.
A few words on 'VACATION', my upcoming collection for
@verse_works
, my followup to LIWA and Reworld.
Oxford Dictionary says:
vacation
the action of leaving something one previously occupied
And in some way, yes, it is about that too.
"Escape Perspective"
🧵
@paularambles
European here.
I went walking in Albuquerque towards a shop - and the pedestrian walkway just ended. Without any good reason to do so. Dead end.
The city gave up.
I am probably the last person to do this tutorial - everyone did this a while ago - but no matter;
Consistent characters in MidJourney and SD
It's quite useful! 🧵
Film Camera
Trained a new model for 'realistic photos' - whatever that ever means. It works pretty well.
Imagine where the quality will be in a year.
#stablediffusion
#portrait
Sooner or later this feature will be integrated into your favourite editor/generator.
Have a good image but not happy with the exact face & expression -> swap it out!
Here using invoke AI - works even with large images.
Apologies for the “point iPhone to screen” video.
Dreambooth training for Stable Diffusion is extremely powerful. You can train a new token to the "person" class to create very convincing looking images of them. I've posted some examples in the past days.
But it's not the coolest thing you can do...
WinAmp skins made with
@midjourney
- worlds collide.
If Winamp3 would have happened at the same time as Midjourney, MP3’s would have never gone away.
Few understand this.
AGENT STORY
Fully text-to-video generated.
No init images, no init videos.
If you have two minutes of time to watch it, please check it out.
#zeroscope
#texttovideo
World Camera
Took my Dreambooth model for 'realistic photos' from yesterday - ran it with the same prompt, but with a tiny addition: each prompt now explicitly mentions a (random) nationality.
Suddenly the world is not just white Western people.
#stablediffusion
#ai
#photo
Don't know quite what to say except thank you
@ClaireSilver12
!
One of my favourite pieces - abundance of math and technology that tries to express something that's the opposite - analog and human.
Finetuned the new SD1.5 model with some realistic portrait photos, ran a quick test of quality.
No GFPGAN/CodeFormer used, no inpainting - there's details I could Photoshop here and there, but ... Not bad. Quickly looking, one might be fooled.
#stablediffusion
#portrait
SDXL is the real deal. Preparing images for my next AI gen short film. I prefer these over what I can do with MidJourney.
Better depth of field. More variance in layouts. Better control. Not so... glossy. Run here without the refiner, I think.
#sdxl
Ok here we go!
SMILE
- An artistic PFP collection of 5000 1/1 AI art pieces
- My first solo collection
- On SOL
- Minting on Jan 30th
- 0.2 SOL per piece
Check out for all the details
Read the thread 🧵
Artificial Life
Fully AI-generated, no initial images.
This is what happens when you find a stranger in the Alps and decide to try to max out the VRAM of the GTX3090 to squeeze out as much resolution as possible.
#AI
#generated
#illustration
#NFT
#diffusion
This thing is absolutely mind-boggling to me.
Are we entering some post-satire age? Are people's minds so far gone that they don't recognize a good piss-take? Doesn't the very first sentence already make it obvious?
Go and read the 3000+ quote tweets.
Using AI, I was able to take some old painting and make it better. First the painting. Notice how you can barely see what's even going on. Where even is this? Who are the people? Huh? You're too far away to really see the setup. Whole left side blank. Nothing here to grab onto.
THE METAMORPHOSIS
Modern adaptation of "The Metamorphosis" by Franz Kafka.
Created using generative AI tools: RunwayML Gen-2, ZeroScope, SDXL, MidJourney, Bark, ElevenLabs, Coqui, wav2lip.
Some thoughts and musings 🧵
Super fun project - got the opportunity to create some "AI visual ecstasy" for the latest
@dannapaola
music video "XT4S!S", directed by the one and only
@edgarestevess
Bullish on SOL!
My first art collection of 2024:
- On SOL
- Biggest collection I've ever done (by far!)
- Super affordable
- Extensive whitelists for all my current collectors (any art, any current chain)
- I'm thrilled about the art & theme
All details revealed next week!
Been beta testing a new
@KaiberAI
feature - here some test results.
A happy mix of everything (SD, ControlNet, ebsynth etc. I'm guessing from the look of it). Another tool into the creator's toolkit - not a universal solution solving all problems, but that would be too easy!
For your Colab, add a new cell at the end of your notebook:
from google.colab import runtime
runtime.unassign()
Disconnects the notebook when you’re done. Saving precious credits. You’re welcome!
#colab
Disco Diffusion 5.1 is out
Of note: video_init_seed_continuity option has been added for video source mode. i.e. stop changing the seed during the video.
Here's an example - 50% skip steps. Night and day. I'm going to have fun with this one.
Blink 182 - All The Small Things (mad anime version)
The song brings back memories. I played guitar in a couple bands while in university. We covered this song a few times - it's catchy af.
#stablediffusion
#blink182
#warpfusion
Super proud to announce:
I've been invited to participate in upcoming Christie's 3.0 NFT Auction: NEXT WAVE - THE NEW YORK EDIT ( starting at April 12-19 ) with my piece "Swan Dive"
@ChristiesInc
DRIVING - ControlNet test
ControlNet is very useful in cases like this, where you want to retain the overall visual shape while adding extra flair and detail to the footage. Warpfusion v0.10, sd15_canny.
See thread for examples.
Announcing first Smile allowlists!
"Smile" will act as the second allowlist (after "Rubber") for the wonderful "Modern Zombies" collection from
@wubbushi
, minting Feb 7th
and!
🧵
Sia Chandelier AnimateDiff ToonYou test -
It's really determined to draw a face no matter what, isn't it! Openpose too strong?
Would need to manually fix. But quite promising in relation to effect vs consistency. The endless testing continues.
I put some AI-generated images up on Adobe Stock for sale. Properly tagged as AI generations, yes. Is this a good idea? Will any of them sell? Does any of this make any sense? Who knows?
I will, eventually.
Something that bugs me to no end: photos of people charging electric cars, HOLDING IT LIKE THEY WOULD A GAS HOSE.
"Let me stand here for the next hours as I casually chat with you."
"It's not photography since it doesn't involve capturing of light with the camera."
For the past few weeks I've been photographing the local neighborhoods here in Helsinki where I live. Capturing how the autumn hits the concrete.
Then I trained an AI with these images. ->🧵
... AND it sold out.
The public phase got heavily botted and a bit overloaded!
Just a second while we thaw the collection so that trading can begin. It'll be available on Tensor and MagicEden, I'll share the links soon.
Thank you all!
Another AI video edit experiment. A man in the shower. It's a bit like a poor man's A Scanner Darkly - kind of interesting.
One could a short movie, or an entire music video like this.
#vqgan
#ai
#aiart
#animation
#aftereffects
REWORLD;
It's what I spent this year creating. Day and night. Every day. 50000+ generations. Ludicrous effort. Using every technique I know of. Pushing the art and craft as far as I possibly could... Expressing what happens to us, individuals in society.
Hope you like it.
@PatrickJBlum
Soon in the near future, you can deepfake the Barbie characters into the Oppenheimer movie, or vice versa, for maximum cinematic enjoyment
Rainisto’s ‘Auto Dream No. 3’ has been selected as a prestigious J.P. Morgan Curators' Highlight at Paris Photo 2023, one of only 20 artworks from thousands on display at the fair.
I'm not going to share you my trained model, but I'll tell you how to train your own. Use JoePenna's Dreambooth.
Pick training material that is photographic. Invent a token, pick "style" class, train enough. Try different training sets.
AnimateDiff prompt traversal test
Trying out this thing. Prompt traversal: New York cars -> traffic -> rain -> pedestrians etc.
I see the potential! Infinite length possible, and they're pretty fast to create.
"Life In West America"
@braindrops
sold out instantly. Overwhelming response, hard to put my feelings into words right now. For now, I'll let the images do the talking - here's all the 500 pieces on OpenSea secondary.
Thank you all.
"Modern Woman Template"
Part of the "REWORLD" collection for
@FellowshipAi
.
Mint pass: Thu March 16th 🗓️
Reworld auction: Thu April 6th 🗓️
A small 🧵 on what Reworld is all about!
Spent two days at Paris Photo presenting and discussing REWORLD with visitors. Talked to - what, around 100 different people.
Zero persons told me to go to hell with this AI. People were curious, intrigued.
The bubbles on Twitter don't really represent the average reality.