Discover more from Stable Digest
STABLE DIGEST #10
Artist spotlight - DiffuseTogether Winners & Full Videos Exclusive!
THIS WEEK’S ISSUE
Hi, we missed you! But we are back with our double digit edition of Stable Digest!
SDXL 0.9 has been released and we are excited to tell you all the details!
Plus, we have introduced a new section! Tools & Tuts will highlight some of our favourite community contributions from the past few weeks! We love to see it!
We also have video exclusives and spotlight interviews with the winners of the #DiffuseTogether Challenge, where we get insights into their background and future plans.
And of course, no Digest is complete without a showcase of art from our community!
🖼️ This weeks Cover Feature: Stas (stiv406) from the SDXL pantheon in Discord.
Never miss an issue of Stable Digest by subscribing for free!
Have suggestions for artists spotlights or community features? We wanna hear from YOU. Email us firstname.lastname@example.org
FOR THE COMMUNITY
Put on your digital berets, AI artists! We're rolling out SDXL 0.9 – a turbocharged text-to-image virtuoso that's leaps and bounds beyond the beta, and just a stone’s throw from the mighty SDXL 1.0!
But here’s the best part – this model doesn’t require a supercomputer! With a decent GPU and some thirst for invention, you're all set. Oh, and it’s more than just text-to-image – it will have all the fun features such as inpainting, outpainting and img2img!
For now, the open source is an exclusive treat for selected researchers, but stay alert for the general release coming soon! But just because the open source isn’t out for everyone doesn’t mean you can’t play!
First you can explore our Discord Bot and unleash your creativity! This is your playground to experiment with our beta SDXL models and see the magic unfold.
Here are just a few of the creme de la creme of images coming out of our Pantheon Channel - the epitome of brilliance produced by those using bots right now!
📎You can also find SDXL 0.9 on ClipDrop
Craving specifics? Dive into our blog.
Want to be part of our early access research group? Send us an email at email@example.com.
BY THE COMMUNITY
SDXL 0.9 vs. MJ Comparison
We know, we know, we just told you about this! But what is the community saying? Well, take a look at this well-crafted youtube video by MattVidPro AI, who takes a deep-dive into the pros and cons of these two image generation behemoths!
Plus, it’s a great overview if you don’t want to navigate the scholarly jungle above!
MODELS & EMBEDDINGS
It’s no secret that anime played a pivotal role in training some of the earliest beloved fine-tunes. Now, those elements have been mixed into many models swirling around the community. So, if you've been spotting Asian and anime themes sneaking into your generated images when you weren’t aiming for them, you are not alone! But hold your paintbrushes, because Zovya is here to save the day with the ingenious 'Style Less Asian' embedding!
This savvy negative embedding dials down the Asian styling so your prompts can more effectively embody what you had in mind, free from the heavy Asian imagery influence.
There are three magic files to download - a nifty negative prompt, and two positive prompts for those extra oomph moments. Be sure to test the all!
TOOLS & TUTS
Imagine holding a magic wand that grants you the power of ultra-precision in composing with SD models - well, folks, that's what Map Bashing feels like! This ingenious technique shared by Olivio Sarkaris on his YouTube channel gets your hands dirty with maps, unlocking a treasure trove of finesse that regular photo bashing could only dream of.
ControlNet maps are as pure as freshly fallen snow, free of color and lighting, meaning your creative canvas is limitless, as you can conjure endless variations from the same composition.
So, don your wizarding hat and let this tutorial sweep you into uncharted realms of AI art creation!
We love discovering the latest goodies created by the brilliant AI community. Our latest find? The open-source Zeroscope, a text2video generator from cerspense that gives Runway's Gen2 a run for its money!
text2video tests with zeroscope_v2_XL by @pharmapsychotic
text2video tests with zeroscope_v2_XL by @amli_art
Zeroscope_v2_XL is an Auto1111 extension we've happily haunted all week. And it is ever delightful ride!
Not sure how to raise this imaginative generator? @DigThatData's got a handy YouTube tutorial. Once you're set up, prepare for the phantasmic fun of animating your ideas into a parade of the uncanny and surreal.
WITH THE COMMUNITY
STABLE SOCIETY DEEP DIVE - DiffuseTogether Winners!
The contest #DiffuseTogether Contest with Peter Gabriel was so much fun we decided to keep the celebration going a bit longer by spotlighting out four winners!
Join us for a peek into their stories - where they began and the futures they're building. These pioneering artists wowed us with their passion for creativity and community. They showed what’s possible by fusing imagination, music, and AI.
Lam Son Tied 1st Place
Lam Son is a freelance artist in Vietnam with a passion for storytelling and AI. Although he began his career as an engineer, he recently discovered stable diffusion and became hooked on creating art with AI tools.
Lam Son struggled with stress and lack of fulfilment in his engineering jobs. He attempted a career switch to marketing but still felt uninspired. Then, in March 2023, he found stable diffusion. "For the first time, I found myself engaging in intense learning and practice without feeling the same level of stress as before," he says. A friend soon asked him to help with AI work for a short film, launching him into the world of AI art.
When Lam Son heard about #DiffuseTogether just 5 days before the deadline, he was dealing with career, financial, and relationship difficulties. Still, his passion for AI art and the inspiration from judge Sagans' work motivated him to create the winning "Panopticom" music video on time. "It all begins with inspiration," Lam Son explains. "I immerse myself in a song...allowing the melodies and lyrics to evoke vivid imagery, stories, and emotions in my mind." He storyboards each shot, assigns keywords, and employs tools like Warpfusion and Blender to achieve his vision. Editing in Davinci Resolve, he ensures the final product elicits emotion.
For "Panopticom," the song resonated deeply with Lam Son's feelings of being constantly watched on social media. "Within the first hour...half the story came to me effortlessly," he says. "The other half emerged as I worked...coinciding with personal challenges in my relationships. It reinforced my belief in the power of love to overcome obstacles." Lam Son felt his entries, though imperfect, conveyed real emotion - which resonated with the judges.
Winning left Lam Son "paralyzed" and "overwhelmed." Comparing himself to other entries, he felt his knowledge of art and cinematography was lacking. But he's found his passion. "I feel fortunate...my strong emotions and stories...resonated," he says. "I've found the job I truly love."
Lam Son feels "profound gratitude" for the experience and support. "Witnessing the remarkable journeys of other participants has been an immense joy," he says. "Many of your works have truly blown my mind." Motivated to explore AI and art, he hopes all will "continue producing masterpieces" and "inspire and uplift one another" in this "new AI-driven world."
Looking ahead, Lam Son aims to create emotive AI animations and stories. He's learning art, colour theory, cinematography, and storytelling while honing AI skills; he plans to study 3D animation. Currently freelancing for a small studio, he believes in fusing tech and art. "There are many people out there, just like me, who find happiness by blending technology and art," he says. "AI art...will find a place in people's hearts as a new form of art." Lam Son hopes to contribute to making that happen.
Driven by inspiration and passion, Lam Son's Diffuse Together win is just the start of his artistic journey to spread creativity using AI. With hard work and heart, his future as a pioneering artist in Vietnam is bright.
Junie Lau Tied 1st Place
Junie Lau, a Shanghai and London-based multidisciplinary artist, creates work that inspires and challenges. Her art has been featured in the prestigious Royal Academy of Arts in the UK and British Vogue. As a director and production designer, she crafts cutting-edge visual narratives for fashion films, music videos, and advertising productions. Her work passionately explores social stereotypes and gender issues while drawing on her multicultural experiences.
Junie is driven by a thirst for innovation. She delves into virtual identity and digital humanity, critically reflecting on impermanence in the digital age. Harnessing AI, her artistry transcends conventional storytelling.
For the #DiffuseTogether competition, Junie used AI tools like Stable Diffusion, DreamStudio, and Kaiber to craft an immersive music video for Peter Gabriel's "The Court." This alternate world breaks free of physical limitations. As Junie says, "The power of AI and stable diffusion allows us to invite and transport individuals anywhere, dressed in anything, experiencing realities beyond the real." She plans to continue improving the project using Warpfusion, NeRF, and other AI models.
Winning the competition was deeply moving for Junie. She was overwhelmed by heartfelt messages from Peter Gabriel's fans, humbled that her work could resonate so profoundly. "I am deeply grateful for winning this competition and being appreciated by so many," she says. "The recognition I have received is immensely meaningful. I never imagined my art could have such power and impact so many who believe I deserve this honour."
She extended her deepest thanks to Peter Gabriel, Stability AI, the Stable Foundation, and the #DiffuseTogether community.
Looking ahead, Junie sees huge potential for AI art, especially in China and Asia. She believes more Asian artists will engage with and contribute to Stable Diffusion. The response to her win in China's AI industry "has been remarkable, with profound implications for world peace and integration."
Junie aims to keep exploring identity and humanity's digital future through AI film and emerging media. She is working on an ambitious video project fusing dance, visuals, and cinematography. AI lets her realise fantastical ideas once limited by resources. Blending storytelling, AI, and technology, she spreads "love, meaning, and compassionate care."
Through Sensory Atelier, her studio, Junie will keep collaborating with creatives in all fields to unlock AI's creative power.
Steve Grzanowski (AKA Vnderworld) 2nd place
Steve Grzanowski has explored AI art for over a year since leaving a 10-year tech sales career to pursue art full-time. Starting with pen and ink drawing and rug tufting, Steve now focuses on "epic maximalism masterpieces" with Stable Diffusion InPainting that will be dropping on Foundation in the near future. “When I initially started this journey, I had no idea how long I'd last, if I would fail or succeed,” Steve says. "Over the past 10 months, I've transitioned from physical art to digital art and my new obsession is trying to keep up with the rapidly evolving AI space."
When the #DiffuseTogether competition rolled around, Steve created a spectacular piece for Peter Gabriel’s “Panopticom”. He envisioned a futuristic society akin to the Truman Show. The video is an array of symbolism, including sunflowers, representing devotion and faithfulness, and a woven tapestry of eyeballs and security cameras as a nod to the surveillance state. Steve shares that the theme of the video is “questioning your reality”.
The video is a poetic blend of the raw power of AI technology and Steve’s incredible imagination. He takes pride in having kept the visuals as authentic as possible. “This video captures the raw output capabilities of AI technology without the bells and whistles of post-production,” he asserts. The first four minutes were a single Deforum prompt. Steve believes this "shows what you can accomplish in 1 week with AI and only a few months of expertise."
When he won, Steve was beyond ecstatic. “I am honoured to be a part of this community and to know that we are the early pioneers blazing the trail for AI adoption and the use of AI in artistic expression and animation,” he expresses. His journey reflects a pure passion for the art and a commitment to pushing the boundaries of what’s possible with AI.
Steve’s creative furnace is always burning, while he is excited about the opportunities arising in AI and Web3, he’s keeping an open mind about the future. He’s looking forward to getting back to some physical art and is geared up for showcasing his tufted rug wall art at the Sevigny Studio.
At the intersection of technology, creativity, and community, Steve's future as an AI artist is a bright one.
Matt Burn (aka Oranguerillatan) 3rd Place
Matt Burn is a pioneer in AI art with 20 years of experience as a CG Supervisor in VFX and 3D animation. Currently CG Supervising the Dark Ages drama "The Winter King," Matt's true passion is generating imagery with AI tools like Deforum and Warpfusion. "I became interested in AI last summer and have been addictively generating imagery using tools like Deforum and WarpFusion ever since," he says. Matt aims to merge VFX and AI. "I want to work with others embracing AI as a tool, not shying away from it," he says.
For the #DiffuseTogether contest, Matt found inspiration not only in Peter Gabriel's song but also in Tim Shaw's accompanying sculpture. "I tried capturing its essence without directly copying it, using prompts related to visuals and aesthetics rather than artist names," Matt explains. "The video aims to capture the duality of human experience, gritty and hopeful." He hopes his video "stands alongside Peter's music and Tim's fantastic art, instilling discomfort and empowering viewers." His goal was capturing "Peter's vision of a just, peaceful world where embattled people unite for a better future."
Meticulous planning was key. Matt spent four days analysing the audio and drafting 60 text prompts to generate imagery. He used a tool by Chigozie to generate keyframes based on the audio, controlling strength, noise, and 3D camera motion to match the song. After planning, generating the video took three days. Matt went with his first results due to the tight deadline. He upscaled the video to 4K with Topaz VideoEnhance AI and added audio in Adobe Premiere Pro. Though he heard of the extended deadline, he stuck with his initial challenge.
Post-win, Matt felt "humbled and proud." He says, "I have a huge amount of respect for everyone who entered and would like to thank the AI community for all their kind words." Eager to collaborate, he says, "Reach out if you'd like to work together on interesting projects."
Matt is also excited to collaborate with Kodiak Arcade. "I was stunned by their aesthetically beautiful AI video” for 'Grow Together’, he says. "Absolutely gorgeous work."
Matt has more DMC work ahead and hopes for other collaborations. "I'm up for hearing from any musicians or artists who would like to work together," he says. Offers are coming in, and he considers each one.
A trailblazer in AI art, Matt Burns is generating new forms of creative expression. Keep an eye on this visionary artist - his future at the intersection of VFX, music, and AI promises to be a fascinating one.
WINNERS OF THE WEEK
Attention, connoisseurs of creation! Once again our COW warriors are moo-ving us to the edge of our seats and we are udderly ecstatic to present their masterpieces!
Galloping into the COW colosseum, tazi artfully peels away the canvas in Mindscapes, an ode to the allure of the psychedelic. Simultaneously, Arson sets the gallery ablaze with Accursed Artifacts, igniting a trail that hurls us into the inferno of unexplored realms.
Head on over to the Community Events on the Discord and earn your chance to be highlighted!