STABLE DIGEST #13

Artist spotlight - Infiniteyay, plus Stable Beluga's, birthdays and more!

Stable Foundation

and

Amli

Aug 10, 2023

THIS WEEK’S ISSUE

Get ready for a spectacular edition of Stable Digest!

Stable Beluga 1 & 2 are making waves, Stable Diffusion is about to turn have a birthday celebration, and StableCode and our Alpha Japanese LM!

This month's Spotlight unveils the wonders of life with Infiniteyay.

Our Model highlight takes a quirky turn with Voxel XL – we're talking pixels galore!

We've amped our Tools & Tuts section with features on InvokeAI, FullJourney, KaiberAI’s latest motion innovation, and the ever-convenient Pinokio!

And let’s not forget the heart of our Digest – the jaw-dropping works from our gifted community members.

Have suggestions for artists spotlights or community features? We wanna hear from YOU. Email us community@stability.ai

FOR THE COMMUNITY

Stable Beluga 1 & 2

Get ready to meet Stable Beluga1 and Stable Beluga2! Freshly hatched by Stability AI and CarperAI lab, these Large Language Models (LLMs) are powerhouses of deep reasoning, acing a variety of benchmarks.

Stable Beluga 1, built upon the robust foundations of LLaMA models, was fine-tuned to precision with a cutting-edge synthetic dataset. Not to be outdone, Stable Beluga 2 harnesses the strength of LLaMA 2 70B foundation model, delivering unrivalled performance.

While we've worked hard to ensure our Belugas swim smoothly and politely in the sea of AI, we're always eager to hear your feedback to further refine them. Drop us some feedback in Discord

For info on what is going on under the hood check out our blog!

Our debut Alpha Japanese Language Model just dropped!

Fun Fact: Did you know that the Japanese StableLM, a language model with a whopping 7 billion parameters, has taken the spotlight as the ultimate Japanese language guru? It's like having a super-smart virtual language buddy in your pocket!

Getting Started: Ready to dive into the world of Japanese StableLM? There are two awesome versions to choose from:

Japanese StableLM Base Alpha 7B: This powerhouse is all set to rock the commercial stage with its incredible capabilities. It's like getting a front-row seat to the language revolution. And guess what? It's released under the super-friendly Apache License 2.0.
Japanese StableLM Instruct Alpha 7B: This one's a research superstar! It's tailor-made for those curious minds exploring the depths of language wizardry. Released exclusively for research purposes, this model is like a treasure chest for language enthusiasts.

For all the nitty-gritty details, head over to our Stability AI blog!

"We're taking a giant leap into the Japanese generative AI world, creating models that capture the essence of Japanese culture, language, and beauty."
🚀🇯🇵 - Meng Lee, the Project Lead of Japanese StableLM

Stable Diffusion’s one-year Anniversary!

Can you believe it's been a year already? Because we can’t!

As we celebrate Stable Diffusion's first anniversary, we're in awe of how we've blossomed into the world's top open-source generative media collective—all thanks to YOU! Our community's spirit and collaboration have made this journey unforgettable. Stay tuned for a whirlwind of fun events, mega stages, giveaways, and surprises from August 22nd to September 22nd.

Keep a close eye on our Discord and socials for all the festivities. Let's celebrate the magic we've created together!

StableCode

StableCode released this week fresh off the press from our Stability AI website and boy, did people line up to grab it! A premiere LLM generative AI tool using three different models to help developers with their coding, StableCode stands out with its capability to review or modify the equivalent of up to five Python files, thanks to an extensive context window. It’s not just code completion; it's a game changer for both seasoned devs and those just getting started.

Take a look at the Base Model and Instruction Model or dive deeper into its features in our latest blog post!

BY THE COMMUNITY

Here at Stability, we're all about celebrating the vibrant AI community around us. So, we've created this little corner to share some of the fab things folks are creating out there! Just a heads-up though, just because we are showcasing these goodies does not mean they are all affiliated with us.

JUST FOR FUN!

Julian Bilcke has whipped up a neat little game called VideoQuest. By blending AI tools like SDXL 1.0, Grounded Segment Anything and LLama2 , they've turned startup prompts into an interactive visual narrative. Still growing, VideoQuest is a delightful demonstration of AI's creative potential. Why not give it a whirl?

MODELS & EMBEDDINGS

Be sure to check out the delightful Voxel XL lora by Fictiverse and imagine your favorite 8-bit memories, but with a modern, cubed twist. Whether you're looking to recreate retro game aesthetics or just love that blocky charm, this lora’s got your back. Dive in and let the voxel voyage begin!

Head on over to Hugging Face or Civitai to start you voxel voyage today!

TOOLS & TUTS

Tutorial Spotlight!

Last month, InvokeAI (a local UI that runs on as little as 4 GB of RAM) rolled out their dazzling 3.0 update, and oh boy, did they knock it out of the park! Olivio Sarikas steps up to the plate with an insightful tutorial covering all the fresh updates. From Nodes and Loras to ControlNet and dynamic prompting - he's got you covered. Plus, he's sprinkling in valuable tips on how to make the most of these shiny new features.

FullJourney!

Ever wished there was a one-stop shop for all your AI creativity needs? Well, enter FullJourney. They’ve got music, sound effects, speech, lipsync, image and video generation all right in their Discord. And get this: they've got a fantastic “/movie” feature where you can whip up a 12-second video with scenes and background tunes! Their video generation also boasts camera control commands!

So be sure to check out all their other features in their Server Commands channels and start creating today!

KaiberAI - Motion

KaiberAI is back with their innovative Motion feature! Just released this week and promising to deliver smoother, more fluid animations, it retains the user-friendly interface and budget-friendly pricing you know and love. So don't wait, animate!

Pinokio

Has the mere thought of AI tool installations made you break out in a cold sweat? Well don't fret, because Pinokio has got you covered! Essentially a one-stop browser, Pinokio transforms the process of installing, running, and automating AI applications and models into a stress-free experience.

Say goodbye to terminal trials, 'git clone', 'conda install', and entanglements with execution environments. With Pinokio, all these hurdles are effortlessly leapfrogged with just a single click.

WITH THE COMMUNITY

STABLE SOCIETY DEEP DIVE - Infiniteyay

“Life is magical!” - Infiniteyay

We're thrilled to have you here, Infiniteyay! Could you kick things off by sharing the pivotal twists and turns that carved the path from your childhood doodles to the fantastical, avant-garde AI works that we can't get enough of?

Thanks for having me! I’m thrilled to be here too!
That’s quite a journey to summarise but I’ll give it a go! My dad sold art supplies, so my garage growing up was literally like an art supply store. I would spend a lot of my childhood drawing as well as making little zines and comic strips. I also really got into making movies on my family's VHS camcorder. I’d film magic shows, make spooky movies with my friends, and do stop motion animations with my toys and clay. I was always creating in some way and often got in trouble in school for doodling. Growing up I watched tons of movies and had a love for the fantastical and more surreal types. I eventually got into acting and did theatre throughout school. I played in a band for about 6 years and did all of our art as well as art for other bands. This eventually led to a career in Design and Creative Direction. Having a job that is at the intersection of art and technology led me to the discovery of AI image generators a little over a year ago. I was in the discord for Stable Diffusion in the beginning and have just kept exploring AI tools for the past year. It’s been a wonderful journey and I’m so grateful for how AI has allowed me to explore my imagination.

Building The World You Want To See - Infiniteyay

Your pseudonym 'Infiniteyay' radiates positivity, it's infectious! Could you delve a little into how this name embodies your perspective towards life, and how that philosophy seeps into your art and your online presence?

Thanks! I’ve always been a more optimistic person and have long been in awe of the abundance of joy and possibilities in life. I started using the name Infiniteyay on Instagram originally when I was starting to curate art I liked. I wanted a name that reflected my outlook on life and embodied my excitement about all the incredible art I was discovering at the time.
I know there’s a lot in the world to be down about but I try to focus on the infinite amazing and interesting things going on. The universe is literally bursting with endless beauty and opportunities if you just look for them. Even though my art gets a little strange and sometimes dark I try to keep a thread of magic, mystery and joy throughout the bulk of it. On Twitter (X) I also make lists of emerging artists using AI with under 1000 followers that I think are creating exciting work. It’s important to uplift each other and I love shining light on artists that inspire me.

Although originally from the US, you’ve lived the better part of a decade in Asia. How do you think living across these different cultures, especially your time in Asia, has influenced the themes and narratives you present in your art?

Living and traveling throughout Asia has opened my eyes in so many ways to new perspectives and cultures that I had no idea about. I’ve learned over the years that I grow as an artist by constantly being inspired and having new experiences. I get an abundance of both of those things by traveling and living in new places. I often try to make my art a reflection of my imagination and all the things I love. So naturally it’s highly influenced by my time spent here. Plus, compared to where I grew up in the U.S. , Japan is insanely magical and dripping with beauty!

Your artwork often pairs familiar scenes with fantastical elements - like the ‘NEIGHBORS’ series, where suburban life gets an uncanny twist. Can you elaborate on the desire to merge the mundane and the magical in your creative process?

I grew up in the suburbs in the 90’s. It was pretty blah culturally and aesthetically. Somehow I managed to develop a wild imagination. I always had crazy vivid dreams too! I pretty quickly realized there was more magic to life than what I was experiencing in the suburbs. Spent a lot of time daydreaming and drawing crazy creatures. That contrast of normal everyday situations paired with something fantastical probably is a reflection of my childhood in many ways. I think many of us find ourselves in monotonous or mundane situations from time to time. Life can be so magical though! I sometimes try to mix the two in my art in hopes that it inspires others to dream and tap into the magical side of life more often.

Previously you’ve discussed your interest in using AI as a magic portal to explore new worlds and craft narratives. This is particularly noticeable in your ‘IMAGINAIRIA’ collection, where every image feels like a snapshot from some never-before-seen realm. Do you outline narratives or stories before creating pieces, or do these narratives evolve as you create?

I do sometimes start with a specific idea or direction I want to go. More often though, I’ll lean into the collaborative nature of AI and explore ideas while creating. In the case of ‘IMAGINAIRIA’ for example, I created a prompt that was a mixture of many things I love. Aesthetics, places, feelings, lighting, styles, actions, etc. I mixed them together in sort of a narrative way to see where it would lead me. The idea was surely I’d end up somewhere interesting to me if I prompt all the things I love! Sure enough I was shown a realm that was totally captivating to me. It was like someone had gone deep into my imagination and took a bunch of photos. That’s one of the things I love most about AI. It’s ability to mix together an absurd amount of inputs and beautifully tie them together.

THE ARCHITECT from **IMAGINAIRIA - Infiniteyay**

With a toolkit including AI art generators, Photoshop, and your traditional illustrations, what's the nutshell version of your creation process? Are you all about prompts, or do you do iterative and post processes as well? What’s the journey from idea to finished piece?

I create in a variety of ways but the most common workflow is something like this: I’ll use a few of my own artworks as reference images in my prompt. Sometimes I’ll create a piece specifically to be used as a reference image and not to be shown. Then I’ll craft a prompt that gets me going in the right direction I’m aiming for. I’ll then create many variations, select ones I like, make edits, change parameters and make more variations. This iterative process feels very explorative as the direction of the piece can often end up somewhere I never expected. Once I have a piece that I really connect with I’ll bring it into photoshop to clean up and refine with their AI tools. Then I’ll use various upscalers on the piece and maybe refine once more in photoshop before being finished.

You've mentioned that you work best in random bursts of inspiration. Could you share more about these bursts? Are they driven by a particular mood, a time of the day, or a stimulating environment? How do you balance this with the demands of the NFT scene?

I think some of the best art comes from moments of real inspiration. Or when you have the space and drive to get into a flow state. I think you can get into that creative space with a routined approach however between work, family life, and many other goals I’ve struggled to create that routine space for creation. It more so happens in random bursts when free time pops up or in between moments. That’s one of the great things about AI tools, you can create and iterate so quickly that you can dive right in whenever or wherever the inspiration hits you. My favorite time to create though is early morning or just before bed.

There Goes the Neighborhood - Infiniteyay

It can surely be hard to keep up with the demands of the NFT scene and social media. I do my best to show up everyday and post something though I definitely miss days from time to time. Thankfully when I do get into a creative space I’ll create a lot of work. So I usually have something to pull from when it comes time to post.

Being one of the 73 NFT artists handpicked by TimesPieces to craft unique art for Deepak Chopra's iconic book's anniversary must've been thrilling! Can you spill the tea on how that opportunity came about, and if there were any standout teachings from Chopra's book that influenced your contributed work?

That truly was such an exciting project to be a part of! I’m very grateful to Claire Silver for choosing me for that opportunity. She reached out one day, told me about the project and said she was allowed to invite one person. I was shocked but so thrilled as I not only respect her greatly as an artist but have long appreciated the work of Deepak Chopra. To be able to speak with him on a video call was definitely a major highlight of the past year! Also to be onboarded as a TimePiece artist by Keith Grossman who was the President at the time was insane. Everyone involved in that project was so kind and made the experience so memorable.
My piece was inspired by several of my favorite quotes from his book. Essentially I aimed to illustrate the essence of someone manifesting their dreams which is a large focus of the book we were to create a cover for.

The Seven Spiritual Laws of Success Cover - Infiniteyay

In your chat with MakersPlace earlier this year, you talked about bringing interactivity into your pieces. And we're seeing it, your Gen2 animations are oodles of fun! How close are we to reaching your dream of more interactive narratives? Anything fun on the horizon?

I think we’re getting close! I’m definitely going to do lots of exploring with AI and video. I think the first step will be when I merge my web design skills and art. I have a lot of ideas for interactive sites that use my AI art to create immersive stories and experiences. Then I think once the AI workflows in the 3D space evolve a bit more I’ll create interactive game-like experiences. I also dream of doing physical experiences utilizing large screens and projectors. That may take a bit more time, though who knows, I still can’t believe how far I’ve come and how far everything has evolved in just one year. Definitely have some things in the works so stay tuned!

Infinite Loops - Infiniteyay

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?

Thank you so much for having me. I’m so grateful to Stable Diffusion and the whole community here for helping me to get started in this space last year. I learned so much during those early days on Discord. I also want to thank all my artist friends on X (Twitter), including everyone in AIIA , for constantly keeping me inspired!