STABLE DIGEST #6
Artist spotlight - PurzBeats, DeepFloyd, Stable Stage with AppliedML Team and more!
THIS WEEK’S ISSUE
We're back with another jam-packed edition that's going to knock your socks off!
Join us as we sit down with the multi-talented musician and AI animator extraordinaire, PurzBeats, for an exclusive spotlight interview you won't want to miss!
But wait, there's more! We've got some colossal announcements to share, including the game-changing DeepFloyd IF image generator, the new StableVicuna chatbot and more!
Plus, the discord bot is back, but this time in SDXL! So be sure to join us today for our Stable Stage with the Applied Team where we talk everything SDXL.
Hold onto your hats, y’all, because this issue is going to be a wild ride!
BY THE COMMUNITY
MODS MADE THIS
May the Fourth be with you!
Our Jedi mod team has channeled the Force through the SDXL bot to craft awe-inspiring art pieces that are truly out of this world. Admire the interstellar beauty of their most recent artworks, inspired by a galaxy far, far away, to commemorate the iconic May the Fourth!
Interested in joining our Stable squad and becoming part of the volunteer Discord mod team? We’re always looking for wonderful members of the community to join us!
Please fill in this application form if you’re up for joining the crew!
WINNERS OF THE WEEK
Get ready for another exciting presentation of the week's top selections, chosen by you and celebrated by all!
As we continue to celebrate our impressive Picture of the Week tradition, let’s not forget we recently introduced the best new thing to do on a Saturday, the Challenge of the Weekend!
Get ready for a double dose of artistic brilliance in this edition of our POW picks! First up, prepare to be Encapsulated in a whimsical,artistic vision by Zekrow that will leave you daydreaming of your next jungle escape.
But wait, there's more! Zekrow has done it again, scoring a second victory with a delightful Pokémon vacation scene set in a Pixel Infinity pool that will have you swooning with nostalgia and wanderlust. Clearly, Zekrow is the ultimate champion of vacation-inspired art!
Moo’ving on to our COW creations, Strix takes us on an artful battle of Chaos & Order that is devilishly delightful to behold, and Zekrow's precious Un-mythical Creature has our hearts orbiting with delight!
Head on over to the Community Events Centre on the Discord and join in the fun!
MODELS AND EMBEDDINGS
As always we love to highlight some of the best trainings to come out of our Models and Embeddings forum!
Craving a touch of darkness with a dash of pizzazz? Look no further TheAlly's Mix III, with Noise-Offset!
This captivating model helps paint a world of brooding portraits and shadowy scenes, rich in inky blacks and dramatic depth. Excelling in img2img and controlnet use, it's just the ticket for adding a dash of daring edginess to your artistic endeavors! Embrace the shadows and let TheAlly's Mix III transform your work into a moody masterpiece!
Check out this brilliant creation and more on the Models and Embeddings forum!
WITH THE COMMUNITY
STABLE SOCIETY DEEP DIVE - PurzBeats
“Give away the knowledge, charge for convenience.”
Firstly, welcome! We're delighted to have you join us today, and we're eager to explore the amazing realm of your artistic creations!
To begin, would you mind sharing some insights into your personal journey and experience as an artist and creative in multiple mediums?
Hey, thanks for having me! I got my first computer in 1985, a Commodore 128. To say it had a huge impact on me is an understatement. I was the first kid in the family to figure out how to make it do anything. Since then I’ve pretty much been glued to a computer. I spent most of my early days and teens playing C64, Nintendo, PC, SNES, and Arcade games. Pixels on CRT monitors evokes feelings from my youth and the wonder and joy of just breaking stuff until it works or looks cool. I got into music, piano and drums from an early age and have been a drummer and band leader for over 20 years. I caught the production bug just before high school and I’ve steadily been making music in various groups and solo projects ever since.
Making visuals to go with my music has been a lifelong passion, it started with lots of tweaking old Winamp visualizers and performing sets for friends. I’ve always been interested in mixing every kind of media and software I could get my hands on to make stuff that might be impossible to achieve in one medium/process alone.
An interactive video/music installation created by PurzBeats where the community was invited to explore hitting sensory-triggered pads to create their own multi-media sound and light show
You have openly discussed living with aphantasia, the inability to create mental imagery.
Can you elaborate on how this condition has influenced your creative process, both positively and negatively, and whether working with AI has impacted your approach in any way?
Sure, my aphantasia is the full-blown total darkness version. I’ve always been laughably bad at drawing and painting, so the fact that I’m a 3D artist now is pretty wild. Being able to do things on the computer has been really nice because it acts as the mind’s eye I’m missing.
Being able to generate images from just text prompts though, that’s a whole other kind of mind’s eye. When this technology dropped, I felt like I had grown a limb back or something. Something I’d never had was now at my fingertips, I created around 6k images on Midjourney in my first wave of testing. Being able to walk around latent space, it’s like being able to actually wander around inside a dream. For a person with no way to conjure up these images in their own head, this technology is truly something special.
My process with creating anything has always been very generative and procedural, permutations and mutations. Being able to collaborate with the computer is a very natural extension of that kind of creative relationship. Not to mention how much LLMs have helped me with all the things that always terrified me about programming. I’ve created dozens of P5.js, Three.js, and python projects using ChatGPT. We’re just scratching the surface of what’s possible.
You come from a heavy background in music and live project installations. What led you to venture into the world of NFTs, and how? If applicable, how have NFTs impacted your artistic practice in terms of creating, sharing, and monetizing your work?
During the pandemic I had a lot of down time, all the gigs went virtual so I spent the first year and a half diving into 3D design software such as Blender and Houdini. This has opened up so many new doors and possibilities in procedural and generative art.
NFTs came into my life around early 2021, I spent a few months researching how other artists were moving in the space and starting to poke my head into various glitch and 3D NFT communities. I was really lucky to meet some amazing people who helped onboard me to Tezos and Ethereum as a way to present my art to a brand new audience.
You are well-known for your dedication to providing educational materials on your processes using AI Tools. What drives you to be an educator in this space, and how do you think that influences the way you navigate the AI art and NFT landscape? Also, where can people find your educational platforms?
I had an incredible drum teacher when I was in high school who really believed in me and lit a lifelong passion for teaching and education. I really believe in paying that forward and helping to demystify some of the scary-looking stuff that prevents artists from making cool stuff.
I mostly live stream on Twitch, and I have some stuff available on my Youtube. I primarily help people one-on-one in DM, or in a group setting on my Discord. On Wednesdays and Saturdays I host a hangout for anyone who wants to hang out while I work and tinker or ask questions about Blender, AI, and Cables.gl.
Your artwork often features elements reminiscent of retro scifi artwork, as showcased in your Deforum animations such as “Resonance Wave” and “The Last Ferocious Whimper Of Society”. Could you share some insights into the sources of inspiration for these distinct aesthetics, as well as any personal connections or experiences that contribute to your affinity for this particular style?
I have a real love for 70s sci-fi and speculative fiction book covers and illustrations. Retrofuturism and the 70s view of the way technology might look in the future has been a massive influence on my work. Naturally that’s the area of latent space I love to explore the most. I’m really interested in the link between organic life, technology, and the metaphysical planes of existence accessed by psychedelics.
I’ve always loved artists and directors like H.R. Giger, David Cronenberg, and John Carpenter for the way they fearlessly explore the connection between humanity and technology. My visual inspirations are mostly based on 70s and 80s cinema and film and the intersection of the practical effects and computer graphics of the time.
In your interview with AI Art Weekly, you mentioned that your journey into generative art began with creating projections for your music. Your generative art, such as the recent "cycling 1985" collection, often features colors and patterns reminiscent of the analog world. What inspired you to combine these elements with your music-related projects?
Being able to make visuals for the music I make and vice versa is integral to my process. Before the pandemic I’d developed an interactive, reactive VJ system to use for live performances. By bridging Resolume Arena with a TouchOSC interface, I created a beat-synced mixture of generative and pre-rendered visuals I’d created using videos processed with After Effects and some other fun tools. Since I can’t VJ while I play the drums, I have a bunch of controls wired up through TouchOSC and a friend in the crowd can control our visuals with an iPad from the audience’s perspective.
The right combination of music and visuals can create a transcendent multi sensory experience, I strive to bring that to my performances and work. I think well-suited concert visuals are a beautiful art form in themselves. Curating the right visuals to match the music or vice versa is an endlessly fun puzzle to solve. I like to create tons of visual loops that I can then remix and use as samples for live performances, ai-animation source input, glitch source input, or music videos.
You have extensive experience using and testing numerous AI tools. Could you share some of your favorite tools that have emerged from these experiences and have had the most significant impact on your current creative processes?
I do most of my tinkering and exploration with Stable Diffusion using the Automatic1111 local version. Deforum really broke my brain when it came out, Disco Diffusion was always a bit too scary, but once I dove into Deforum there was no looking back. After that I got into Stable WarpFusion which is a patreon supported notebook for doing really wild AI animations with vector flow maps and all kinds of powerful tools for controlling how your animation interacts with the input video. My latest obsession is Runway’s Gen-2, the txt2video stuff that’s coming out right now is a whole new type of experimental filmmaking.
As a multimodal artist, you are known for incorporating various elements and tools in your work, including music, generative art, and AI. Could you share some of your favorite multimodal processes for creating your pieces? Additionally, are there any tools or technologies you wish existed that have yet to be developed?
A lot of my process involves mixing tools, so I’ll run Blender animations through Deforum, or filmed video through Stable WarpFusion, or Gen-2 into Deforum, then pixelsorted with a python script I made with ChatGPT, then sent through WarpFusion. A running thread through my work has always been glitching and forcing tools to do things they weren’t exactly designed for.
The next technology I’m looking forward to is finding ways to integrate all these tools I currently use in more exciting and human-interactable ways. Using LLMs to help build interfaces that bridge technologies and platforms to really create some new artistic paths.
As your artistic journey continues to evolve, we're excited to learn more about your future plans.
What are your upcoming projects or goals as an artist, and how do you see the evolving landscape of AI art and NFTs playing a role in your future creative endeavors?
Right now I’m working on a few confidential projects that will be announced soon, as well as creating and minting pieces for my existing and future Tezos and Ethereum collections. I have an FXHash piece in the works that I’ve been working on in collaboration with ChatGPT. I’m really excited about the process of having a “programming buddy” to help conceptualize and execute my crazy ideas.
There are new emerging AI technologies every single day that disrupt and improve workflow to the point where I can’t even imagine what things will be like in 2 months, let alone 2 years. I will continue to adapt new tools and workflows into my creative journey.
Thank you for taking the time to speak with us! Would you like to share any additional thoughts with us and our community of readers, or give shoutouts to anyone who has been an inspiration to you along the way?
Don’t be afraid to press the buttons you’re not supposed to press and see what happens. It’s pretty rare that you can break something so badly it can’t be fixed, so explore! I also think that stepping back and looking at the landscape of AI/ML tools can be extremely overwhelming, it helps to pick a task or simple goal to accomplish and break things down into smaller digestible pieces. Before you know it you’ll be able to tie everything together to make really amazing things.
There are so many people I’d like to shout out for being extremely helpful and nice, Chris Allen (@zippy731) was the first person to really show me what was possible with Disco Diffusion and to not be afraid of digging into Google Colab notebooks. The people at Harmonai have been super helpful in learning how to understand and navigate the AI audio world. Stability, Midjourney, OpenAI, and RunwayML for allowing me into all the betas of their generation tools. There’s not enough room here to shout out all the incredible individuals who inspire me to help others, but I retweet them every day.
“Give away the knowledge, charge for convenience.”
FOR THE COMMUNITY
DeepFloyd IF 🤘
Feeling blue because AI graphics and text just won't play nice? Well, cheer up!
Get ready for a palette-pleasing adventure with our multimodal AI lab, DeepFloydAI, and their public release of DeepFloyd IF – the cutting-edge text-to-image model that's here to revolutionise your artistic horizons!
Explore a treasure trove of mind-blowing features with DeepFloyd IF, including deep prompt understanding, seamless text integration, picture-perfect photorealism, and aspect ratio versatility. And that's not all! Modify style, patterns, and details while preserving the essence of the source image, all without breaking a sweat on fine-tuning!
Lyric video using DeepFloyd IF
Eager to learn more? Head over to our blog for all the brainy details.
Or go straight to giving it a test drive on HuggingFace.
Ready to take the plunge? Access the model card and code here.
Dive in, and unleash your inner artist!
StableVicuna
It's a bird, it's a plane... No, it's StableVicuna, the first large-scale open source chatbot trained through reinforced learning from human feedback (RLHF)!
What's the secret sauce, you ask? StableVicuna is a souped-up version of Vicuna 1.0 13B, which itself is an instruction fine-tuned LLaMA 13b model.
Its robust performance is achieved by tapping into a treasure trove of datasets and harnessing the power of Proximal Policy Optimization (PPO) reinforcement learning. Want to know about what’s going on under the hood? Check out our blog for details.
But wait, there's more! We have an upcoming chat interface, currently in the final stages of development, that promises to create an interactive and user-friendly experience that'll leave you saying, "AI, Captain!"
Remember, this is just the beginning for our lovely StableVicuna. We're all about improving, so we'll be tinkering and refining it in the coming weeks. Feel free to give it a spin on HuggingFace and share your valuable feedback with us.
API Upscaler
Hold on to your pixels, we are jazzing things up with the release of our Image Upscaling API! This snazzy tool enlarges your images while keeping them sharp and detailed!
What's the secret sauce behind this marvel? Two open-source models: the lightning-fast Real-ESRGAN and the detail-oriented latent Stable Diffusion 4x Upscaler. Choose your upscaling adventure based on your needs, and bask in the glory of pristine, upscaled images.
Learn more on our blog, or get started with this API here.
For all you Stability for Photoshop and Stability for Blender fans, we've got you covered – our latest add-ons have the API integrated for your convenience!
Happy 200k May-arthon!
As we sprint into summer/winter wherever you are, we've got a marathon of goodies for you ahead of our 200k Discord member celebration! From spicy releases to Nitro & credit giveaways, Stable Stage events, and more surprises!
What's on the horray-zon?
The bot is back, and it's super-sized! We've launched the SDXL bot for a limited "SD-Xperiment" with a select group of beta testers. Stay tuned as we onboard more users and refine the bot!
Vote for victory! Everyone can cast their votes for their favorite SDXL images on the Showdown channel, with top choices displayed in the Pantheon channel. Your votes train our model, so let's push the limits of our tech together!
Stable Stages are storming back! Don't miss our lineup of thrilling behind-the-scenes stages this month, talking to the minds behind SDXL and AppliedML team DeepFloyd and more! Keep an eye on our discord for all the juicy details!
Animation SDK
Been wishing you could harness the power of Stable Diffusion’s cutting-edge models to create breathtaking animations? Stay tuned for more news next week!
Events, events, events! Don’t forget about the #DiffuseTogether Challenge in Discord, culminating on a full moon live on Twitch where Peter Gabriel himself will announce the unforgettable winner! Official Announcement on extension and prizes tomorrow Friday 5th!
Let the May-arthon begin!
Wow, the interview with PurzBeats made me realize I have aphantasia and have my whole life! I’m almost 30 and never knew. So thanks for highlighting that in the interview and linking to an article about it!!!