Distillery.dev

#genai AI can't replace the unique touch of designers, but it's an incredible ally in saving time, crafting initial concepts, and sparking new ideas. Followfox was developed to provide creative minds with a 'digital brush,' equipped with built-in functionalities that support their workflows.. Below is a simple example of how a non-designer can generate a visual for social media (Halloween-themed post for a shop called SOLA). Current functionalities can already be used to use your brand name and change the details of the image with in-painting. Have a use-case that you want to try? - happy to support. #aiapplications #graphicdesign

🥃 Tips to become a professional AI art maker As part of our open-source policy, we at FollowFox are committed to sharing our knowledge. 'Learn generative AI with FollowFox' is a series of posts where we share micro-learnings ❓ What is the 'Seed' parameter and how to use it? If you use the same Seed number in your prompt, you will generate the same image every time. Using the --seed parameter can greatly assist in iterative experimentation, allowing users to observe the effects of modifying prompts and other parameters on their image generations. In the example below, we used the same prompt two times and the output images were identical, due to adding ''--seed 123'' to the prompt. The seed number can be any random number. ❓How does it work and how to add it to the prompt? Stable Diffusion begins its image generation process from a random noise. By setting a specific seed number, you ensure that the starting noise remains consistent every time, leading to reproducible results. To add it to the prompt, simply add --seed (seed number) to your prompt in Distillery 👉 Follow us not the miss out on upcoming micro-learnings! 👉Distillery is an AI text-to-image generator created by FollowFox. We provide 25(!) different parameters to control your image generation. Want to give it a try? Join Distillery, we provide 10 free generations per day for everyone http://FollowFox.ai

🔥What it takes to create unique art? - deep dive into example of a prompt ℹ The image below was created by a Distillery user. We will dive the prompt to see what it takes to write a unique prompt. The prompt of user 'moloko' draw inspiration from a diverse and eclectic mix of artistic styles and fictional characters, creating a unique and imaginative scene. It includes references to: 1️⃣ The Flying Glove and Snapping Turk from "Yellow Submarine": These are characters from the 1968 animated film "Yellow Submarine," which is known for its psychedelic imagery and colorful, abstract designs. The film's art style was heavily influenced by the pop art movement. 2️⃣ Lady Rainicorn and Marceline from "Adventure Time" by Pendleton Ward: These are characters from the popular animated television show "Adventure Time." Lady Rainicorn is a colorful, rainbow-like creature, and Marceline is a vampire queen. The show is known for its whimsical, surreal, and sometimes bizarre characters and settings. 3️⃣ Peter Max: An artist known for his psychedelic, colorful artwork that became iconic in the 1960s and 70s. His style is vibrant and often features bold, abstract shapes and forms. 4️⃣ Junko Mizuno: A Japanese artist known for her unique blending of cute, childlike aesthetics with darker, more surreal elements. Her work often features bright colors and is heavily influenced by manga and Japanese pop culture. 5️⃣ Yoko d'Holbachie: An artist whose work is characterized by bright, vivid colors and fantastical creatures that blend cuteness with a sense of strangeness and unease. 6️⃣ Keith Haring: An artist and social activist known for his graffiti-inspired artwork, which often features bold lines and bright colors. His work is often filled with simple, yet impactful, figures and symbols. 7️⃣ (((Todd Schorr))): An American pop surrealist known for his highly detailed and meticulously crafted paintings that often contain references to popular culture and cartoons. 8️⃣ (Hokusai): A famous Japanese ukiyo-e artist from the Edo period, best known for "The Great Wave off Kanagawa." His work features traditional Japanese themes and techniques. 9️⃣ Tibetan Thangka: These are Tibetan Buddhist paintings on cotton or silk appliqué, usually depicting a Buddhist deity, scene, or mandala. They are known for their intricate details and spiritual symbolism. 🔟 The combination of these various elements suggests a vibrant, surreal, and eclectic image, blending traditional and modern, Eastern and Western, as well as fantastical and real-world elements. Created with http://FollowFox.ai Credits to our user 'moloko'

Have you ever wondered what goes into fine-tuning those cool AI image generators like Stable Diffusion? Here is a step-by-step guide by Followfox that breaks down everything you need to know to train your own custom Stable Diffusion model. 🖼️ Find Training Images The authors used Midjourney v5 images since they found synthetic images train SD models faster. They compiled around 4,000 image-prompt pairs. 🧹  Clean Data & Prepare Captions They downloaded the images and used the original prompts as captions to retain Midjourney's prompting style. 🔩 Define Training Protocol They used EveryDream and trained for 100 epochs on an RTX3090 GPU, saving checkpoints every 25 epochs. 🪄  Fine-Tune the Model The 100-epoch training took 20 hours. They observed the training and validation loss to choose the best checkpoint. 🧑💻 Test the Model They tested different checkpoints on various prompts to compare quality and choose the 25-epoch checkpoint for release. ✔️ There are definitely further improvements possible as the authors mention, but even the initial v1 model produces fun and creative images. Let us know if you give it a shot or if you have any other tips for training AI image models! #ai #artificialintelligence #stablediffusion #imagegenerator

🖼 AI and the Arts: How Machine Learning is Changing Creative Work Recently, we have stumbled upon a report from the University of Oxford talking about interplays between artists and their machine learning-based tools. Even though the report dates back to 2022, we can clearly see that for artists to create their own model and workflow that fits their needs, they have to go through extensive work. Distillery, founded by gaming executives, who have close contact with visual artists, is built with the main goal of providing a solution that simplifies the creative process while preserving the artist's unique flair, adaptability, and vision. Our vision is to empower creators across diverse genres, enabling them to maintain their distinct style and workflow, while effortlessly achieving remarkable results. https://followfox.ai

📖 When we launched Distillery a few months ago, we made a promise to always be transparent with our methods, and to be committed to the open-source cause. We created a series of blogs where we we started from an empty canvas, and step by step, we are building up SDXL workflows. The goal is to build up knowledge, understanding of this tool, and intuition on SDXL pipelines. 1. [Part 1] SDXL in ComfyUI from Scratch - SDXL Base 2. [Part 2] SDXL in ComfyUI from Scratch - Image Size, Bucket Size, and Crop Conditioning 3. [Part 3] SDXL in ComfyUI from Scratch - Adding SDXL Refiner 4. [Part 4] Advanced SDXL Workflows in Comfy - img2img, LoRAs, and controlnet You can read all 4 parts on our blog - https://lnkd.in/dfhyKeXn

🧠 Ho fatto alcuni test usando #Distillery, un sistema di generazione immagini creato da Followfox. 💡 Distillery si basa su #Cosmopolitan un modello open source creato attraverso il fine tuning di #StableDiffusion 1.5. 🦾 Oltre ad essere open source, in questo post viene spiegato in dettaglio come si è ottenuto il modello a partire da SD: https://lnkd.in/dS9gXacS 👉 Una dimostrazione di quanto stia diventando "semplice" implementare un sistema efficace da un modello open source. 🤔 Fantastico! Ma come la mettiamo con filigrane e riconoscimento? #AI #GenAI #GenerativeAI #IntelligenzaArtificiale

AI image generation has limitless use cases. At Followfox, we continuously explore user applications, believing many are yet to surface. A prominent current use is crafting website visuals. Below is a workflow of how we used Distillery to create a background image for our website from scratch 1️⃣ Ideate: Begin by brainstorming your desired outcomes. Explore other websites for inspiration or engage with GPT for initial ideas. At Distillery, we aimed for a blend of realism and magic, symbolizing open windows to infinite creative opportunities—signifying Distillery as a gateway to integrating AI into creative work. 2️⃣ Develop Your Prompt: Crafting a powerful prompt is key. Getting a good prompt takes time but it is worth every second of it. Here, you have a few options: 1. Start by scrolling through Distillery channels on Discord. The get-inspired channel is a great starting point. Gather the visuals that you love there. Don't worry about images being of another subject, we will take care of that later. 2. Use our 'Crazy' function that can suggest you a crazy version of any prompt. All you have to do is write a few keywords or a simple prompt and add the --llm crazy command. It will return you a prompt that has a touch of 'crazy'. 3. Ask ChatGPT to offer you a prompt. We won't dive into details here, there are bunch of available information about using ChatGPT for prompt creation. 4️⃣ Find a prompt you like on social medias or http://civit.ai Gather all the prompts that you like and start experimenting by mixing them and using the parts of the prompts you like. Again, don't worry about the subject of the prompt, focus on visuals, colours and style for now. For us, one of the generations of our users served as the base start. 5️⃣ Select Your Subject Image: Identify an image representing the final result you envision. In our case, we chose an open window photo found on Google, serving as the canvas for our blended prompt. 6️⃣ Blend the Prompt and Image: Use the Distillery function '--adapt [img url]' to blend your carefully crafted prompt with the chosen image. This sets the stage for your initial generations. 7️⃣ Experiment with Distillery Parameters: Refine your creation by experimenting with Distillery's versatile parameters. Try different models (sdxl, screwdriver, vodka), adjust CFG values, play with aspect ratios—fine-tune until you achieve the desired outcome. Use --control and --image functions to incorporate images you like from earlier generations. Note: It may take several attempts—around 20-30 tries for us—to refine your vision. 8️⃣ Seek Support from Distillery Founders: We are happy to support your creations on public channels of our Discord. All our users have 10 free generations per day. If those won't be enough, for this particular project, we will grant you more free generations. Just reach out to us on Discord. Join Distillery - https://followfox.ai