STABLE DIGEST #1
Exclusive artist spotlight with Remi Molettee
THIS ISSUE: A DEBUT EXCLUSIVE INTERVIEW WITH REMI MOLETTEE
Welcome everyone to the first issue of Stable Digest!
Featuring community spotlights, interviews, all the latest developments and more, this newsletter is the place to keep up to date on everything from across our Stable community and beyond!
Ever since we began this exciting journey together with the release of Stable Diffusion, your creativity, collaboration and initiative has made us what we are today: AI by the community, with the community, for the community!
BY THE COMMUNITY
MODS MADE THIS
Our Discord moderator team put out some awesome artwork, and each issue we’ll showcase some of their latest and greatest creations in this segment. Here are just a few examples of what the team has been working on lately!
PICTURE OF THE WEEK
POW! Head on over to the Community Events Centre on the Discord and get involved in our Picture of the Week challenge! Check out the weekly theme, show us your amazing AI art skills and vote for your favourite submissions!
Don’t forget to opt in to our Challengers role to keep up to date on all the latest!
We rounded out 2022 with a celebration of Stable Diffusion’s first year and the POW: Best of SD event!
The new year brought a fresh start and the beginning of a new chapter, and what better way to kick off the new year than with POW: Beginnings!
From new beginnings, our events evolved and transformed in POW: Metamorph!
Check out the winners of our featured POW events right here!
MODELS AND EMBEDDINGS
Our Model and Embedding share forum is a great place to find the latest and greatest community-trained models and embeddings!
Each issue, we’ll showcase one of the most popular picks from the forum.
Here we have Stillie Willem’s FloralMarble for SD 2.X! This awesome embedding can be used to achieve spectacular marble statues with beautiful floral flourishes.
Take a look at these wonderful examples, and head to the forum to discover even more!
DREAMER COMMUNITIES
Our Dreamer Community forum galleries over on the Discord are a great place to browse, share and discuss a range of themed and categorised SD art. Head over there and check out what the community are creating right now!
Each issue, we’ll showcase a popular community and your favourite picks!
This time, we’re showcasing an awesome range of Vehicles!
WITH THE COMMUNITY
STABLE SOCIETY DEEP DIVE: REMI MOLETTEE
“Art is personal, share your story, your vision. Do things for yourself first, not for others.”
Hey, Rémi! Tell us about yourself and your introduction to AI!
Rémi: I discovered AI in 2020 while developing a film project. Honestly, I didn’t know what I was getting into. I wanted to make a film based on a personal reflection of the place technology has taken in our lives since the 2000s. I’m in my forties, and I’ve seen technology creep into our lives more and more. I sometimes feel that technology has lost its status as a tool. People are looking for their daily dose of dopamine through technology.
I wanted to talk about the hold of technology on human beings under the prism of an artist who would use technological innovations for creative purposes. But, I didn’t know what medium this artist could use to illustrate this story. 3D? Something else? I read a lot about AI during that time, and I found that this medium is the symbol that represents our time well.I started to write a structure, but I kept coming back to the same problem. I couldn’t quite see how AI could be used for creative purposes. To continue writing this project, I needed some credibility. If I show an artist who works with an AI, I needed people to believe in it, and for that – there was only way: put myself in the shoes of an AI artist.
I had to live in an immersive experience, so I created an Instagram account under the pseudo of “Rémi”-- my main character. I had some time, and I thought that the works I had created could perhaps be used in film.
We notice a trend between two different types of hobbyists in this space. Extreme mode and light mode-- no in between! Did AI start off as a casual hobby for you, or have more of a rabbit-hole obsessionist effect?
Rémi: It was a complicated start, because in this world everything runs on code. My knowledge of code was zero. I was frustrated at first that I couldn’t do anything.
I discovered an interesting technique called Neural Style Transfer, which allows you to transform an image by transposing the style of another image. It’s a very different outcome from the prompt, but just as creative.
I started on an online site called Deep Dream Generator, which allows you to use the style of a transfer technique without any coding skills. That’s when I started making images. I spent a bit of time on it everyday. It was a lot of fun to use! One day, I mixed an image of a bison with anemone hair and fish lodged in the eyes.
Even though it was very nice to do, I had no authorship over these images. I wanted to do something more personal– to use my own images! So, I contacted two dancer friends, Riscc and Mulunesh, who have been with me since the beginning of this adventure. That’s when things got more serious. I didn’t want to lose sight of my job as a director. I wanted to use the movements of their dances as a reference and then style them.
I was impatient to see the first animations, so it was quite a laborious method. I stylized the images one by one, that’s why my animations were very short at the beginning. It moved in all directions, but I had the vision to advance!
I discovered a whole community that used different codes than I, like VQ-GAN+CLIP, Pity, Disco… I absolutely had to try them. And that's when, I think, it became obsessive.
There is such a profusion of technologies, that if you don’t spend a lot of time on it, you quickly find yourself completely lost. With AI, everything can become obsolete in a few weeks or days. You have no choice if you don’t want to lose track.
Speaking of extreme mode...Your video with The Chainsmokers was absolutely incredible! Can you give us a bit of some "Behind the scenes" of how you got this opportunity, and some of your workflow behind this project?
Rémi: They contacted me because they liked one of my videos “Anemon girl" on IG. They wanted a one minute visualizer for their next track with Gilles Cheyenne. It's always stressful working with AI, because there's a randomness factor that you can't control, and the dancer's movements were super fast. The deadline was very short, only 10 days. We had to make artistic and aesthetic choices very quickly.
At their request, they had written a list of keywords representing their universe (rocket, fire, explosion...). These words were very different from the style I am experimenting with. That's what I'm looking for too.
To renew yourself, you have to get out of your comfort zone and try new stuff, new styles.
How do you get in the zone? Tell us about your setup! (music you listen to, environment around you, inspiration)
Rémi: My office is in a former cabinet-maker’s workshop. It’s a place I invested in a few years ago. It’s a place that can serve several purposes, including that of a film studio. The decoration of this place is rather “low-tech” and furnished with recycled objects.
In terms of equipment, as I sometimes work in VFX, I am rather well-equipped, but AI requires a lot of resources. I work locally, and had to invest in a better graphics card. I’d really like to be better-equipped, but it can be expensive.
I don’t really have a routine or precise schedule, since I don’t do this full time. When I work, I like to keep a certain spontaneity in the work. Movie-making has allowed me to learn different skills such as post-production, editing, framing, colour-grading, and storyboarding. All these skills are really useful and allow me to be independent.
Dare we ask...is AI art real art? What constitutes art?
Rémi: Yes, no doubt AI art can be art. Like all the different forms of art before it, it represents the time in which we live. Each artistic movement had a particular resonance according to the historical context. Artists have always used the technology of their time to create their works: paintings, photography, videos…Art is, for me, a reflection of the time and the artist very often uses the techniques which are contemporary to him.
Of course, not everything is art in AI. Anyone can create images, and soon enough videos– simply by writing a prompt. It must be used creatively. You have to understand the tool and be aware of what it represents. In the more long term vision, we will be able to distinguish the “artist” from the simple “AI user”.
There have been many reservations as to the ethics of AI art. We're at a stage now where we can easily recreate the styles of our favorite artists in the fraction of a time it would take them to create the work we're drawing inspiration from. What are your thoughts on the idea that AI art devalues existing artists' work?
Rémi: I have very mixed feelings. I understand and empathize with all the artists who are afraid of having their art, knowledge, income, and jobs stolen. Some companies would rather generate AI art in a few seconds than pay artists. I myself wonder if everything I have learned will not become obsolete soon, and if cinema will be able to adapt.
It’s still a young technology. It’s very easy with AI to imitate famous artists in a new seconds. It’s fun, yes. But, then…what’s the point? It only lasts for a while. The goal for an artist is to have his own style. His own vision.
Whether you’re for it or against it, it won’t change much. AI is here now. It won’t go away. I’d rather own it and use it than be subjected to it in a few years. Of course, AI will make jobs disappear and create new ones– that’s obvious, but I think that traditional artists will continue to exist. The different artistic techniques have always coexisted.
There needs to be a dialogue between artists and the companies that develop AI. For me, a frontal fight is useless. Anger has never made anyone grow. On the contrary, it is necessary to find a ground of appeasement. I hope that solution will be found, so that no one feels wronged. If you don’t want AI technology to get out of control, traditional artists must contribute to its development by bringing their ideas and visions to make AI evolve in the right way– whether it is from an ethical, legal, or artistic point of view.
We can now describe an image in conceptual terms and see it realized before us in an instant. Previously, a great deal of technical ability and knowledge was required to visualize some of these concepts using traditional methods. How do you see this lowering on the barrier to creativity influencing the wider world of art in the near future?
Rémi: This lowering of creation time overwhelms us, especially on social media with millions of images. Today, everyone is able to create images, which raises the question of the value of an image. With AI, it will become difficult to differentiate oneself from others. This will de-motivate many artists to create new things, by saying to themselves: “What’s the point? If I spend time doing something new, it will be stolen from me in a few seconds”. Some artists may even no longer find meaning in their own art.
I have children and I’m sometimes afraid that they won’t want to create, to learn music, to draw…because they’ll think: “What’s the point? The AI will always do it better than me”. I learned a trade, and by testing the AI, I realized that the creation of film will completely change. One day, not so far away, a single person will be able to make a film which will call into question many more things.
But, you can’t let yourself be defeated. You have to keep learning, studying the past, and getting inspired. Having a cultural background is important. You have to have things to say, and have stories to tell to be able to use AI creatively. Ideas are the basis. I was often frustrated because I had ideas that I could never materialize. But now, thanks to AI, we will all be able to unload our thoughts on a screen just by describing them. That is fantastic.
A big concern in this space is differentiating synthetic material from reality. What if we eventually cannot tell the two apart?
Rémi: It’s not really a problem for me. In art, it is the imaginary that takes precedence. When I create a video of a dancer who is half woman and half robot, everyone knows that this is not real. On the other hand, when it concerns the news, political personalities, or documentaries supposed to inform the population about what is really happening on Earth– it is very worrying. The multiplication of fake news is frightening because it weakens democracies. With video-editing, we have always been able to make images say what we wanted, but with AI and Deep Fakes we have passed a new stage. Even the most seasoned eye will soon be unable to tell if a video is real or not. The only way now is to trust the journalist or the channel that broadcasts it. (The drifts can be fast)
What was the first image you ever made using text-to-image? What are your thoughts on this starting point of your journey into AI, and now reflecting back?
Rémi:Look at my first insta post: A dinosaur generated by Deep dream generator! Before history, there was prehistory! My wall is my personal history with AI. I think we're still in the early days of AI and can't wait to see what we can create in 2023!
Is there anything else you'd like to share with us and tell the community, artists to mention, or shoutouts to friends? Give us the first sentence that pops into your mind for today's inspiration!
Rémi: Of course. I would like to thank the dancers Rissc, Mulunesh, Kine-ama-selase, and the composer Rim Laurens– who has followed me and helped me since the beginning. Without them, I would never have been able to find a direction in my work.
The videos I’ve created with them allowed me to make many collaborations. I think of Irina Bashuk, Oleh Ibrahimov, Soyuz, Dylan Mayoral, Francesco Misceo, Eric Lefaure… each in their own way trusted me and allowed me to renew myself– to make new experiences.
Thank you also to the coding communities that produce innovative tools and are building the AI of tomorrow. I give a shout out to @devdef who developed Warp Diffusion, which became my favorite tool for video. Shoutout as well to the team at Deforum, DiscoDiffusion, Pytti… I surely forget others.
I would also like to thank Superchief Gallery and all the groups that broadcast and share my work, all the people who have encouraged me through messages, and of course the whole team at Stability. These thanks could be very long because many people inspire me daily with their work.
When we create– we are not alone. We inspire each other, we share, we exchange…
My sentence of the day…not easy. Some of you may have noticed by my Twitter account that I’m not very talkative, and I don’t answer much (sorry). But, I could give this advice to those who want to create with AI:
Art is personal. Share your story. Share your vision. Do things for yourself first, not for others.
FOR THE COMMUNITY
STABLE FOUNDATION
Ever since we first got our hands on Stable Diffusion back in August, you’ve grown our community into a thriving hub of creativity where we’ve seen you all push AI technology towards new heights.
Just as we’ve all adapted and grown alongside this rapidly emerging technology, the Stable Diffusion server has entered an exciting new chapter!
Stable Foundation: The Stable Diffusion-focussed server you all know and love, reimagined and updated with a whole host of improvements and new features! Reinforced by all the latest news and updates from across our Stable communities, including Harmonai, DeepFloyd, CarperAI, EleutherAi, OpenBioML, LAION and Stability Japan!
All your favourites are here to stay, backed up by the latest AI news and announcements all in one place. Expect all sorts of new events, features and a few exciting surprises as we begin this new chapter together!
DEEPFLOYD
Speaking of DeepFloyd, you may have seen teasers of their amazing new image generation model, blowing us all away with coherent text outputs and some truly outstanding interpretations of prompts!
We’ll have more to share very soon, along with a few surprises! Here are just a few examples of amazing outputs from the closed alpha in the mean time.
TWITCH
We launched the Stability Twitch channel back in December with a huge 7-hour launch stream! The first of many streams to come, this all-star event featured guests from throughout Stability and beyond.
We gained exclusive insights into the Stability.ai collective, and what the teams at LAION, OpenBioML, CarperAI and Harmonai had created during a year of rapid-paced innovation, and what’s still to come on the horizon!
Following showcases from the DreamStudio and API teams, we joined community artist pioneers LablabAI, Deforum, Braindrops, The New Renaissance with Black Label Art cult & Amli and Claire Silver, where we explored the new frontier of AI art!
We rounded things out with a community showcase reflecting back on a year of chaos and creativity, and a celebration of what makes us special: YOU!
Take a trip back through time and re-live the highlights all over again in this slick video edit!
Here we are at the end of our first issue! We hope you’ve enjoyed this insight into just a few of the exciting happenings in the AI world right now! Come and join us and our wonderful community over on the Discord server for exciting events and more!
We’ll see you all soon for the second issue of the Digest; Stay Stable everyone!
First!
Thank you for bringing the first digest of many! and for starting this with Remi! I've always found his work so inspiring and was waiting to hear what he has to say