2024‐05‐30 Show Notes

Replay link: https://x.com/i/spaces/1YpKkwzonkXKj

AI Art Today Space Summary

This is a summary of the "AI Art Today" conversation, highlighting key points and addressing the specific questions asked.

Hosts: Jen

Speakers: Artful Papers, Nifty Trap, Ghost, Pete, Chris, Terry, G., Mars, Travis

Timeline and Topics: The conversation revolves around AI art, with a particular focus on training models, using specific tools, and discussing artists and trends.

• Early Discussion (beginning - approximate mid-point): Introduction and initial technical difficulties. Discussion of Claire Silver's AI art contest, AI art tools (Midjourney, Runway, Comfy UI, Automatic1111, Wombo), and prompt engineering in Midjourney.
• Model Training Discussion (approximate mid-point - end): Deep dive into model training with a focus on Dreambooth, Laura's, and fine-tuning techniques. Participants discuss the pros and cons of different methods, the importance of captioning, and personal experiences with model training. The conversation touches upon the role of style transfer and ControlNets in relation to model training, highlighting their potential impact on the future of AI art creation.
• Data Visualization and Other Tools (towards the end): Brief exploration of data visualization using AI tools, specifically GPT-4, and its potential in AI art. The conversation also touches on face analysis techniques and the challenge of achieving character consistency in AI-generated images. It concludes with a discussion on area composition as a technique to create more dynamic and less AI-typical compositions.

Interview: There is no interview in the traditional sense. However, Jen, as the host, engages in individual conversations with various speakers, asking them about their work, techniques, and opinions on AI art-related topics.

Questions Asked:

• "Artful Papers, are you doing art or are you creating something?"
• "How's the new pieces going?"
• "Would you like to speak about your art or anything else?"
• "Maybe some news related thing you've been thinking about?"
• "And what you're usually discussing on this? Sorry, my English is not perfect."
• "But what do you usually do on these spaces?"
• "Do you know something about her contest?"
• "I don't know too much about Claire Silver's contest, right?"
• "Is there a particular theme? I remember her talking about like you, she wanted a contest where you had to train your own models on something."
• "But maybe she changed her mind or maybe she has a different theme."
• "What's the theme of the contest or is it just, you know, the best picture wins?"
• "Any particular pieces you're liking so far?"
• "Sorry. Can you repeat the question?"
• "Any particular pieces in the contest that you like so far?"
• "Nice. And you're using a mix of runway and mid-journey, right?"
• "Do you have any tips using mid-journey? I haven't really used mid-journey much."
• "How do you generate prompts for any strategies or something like that?"
• "And what do you use usually?"
• "Also, are you doing any new artwork? I haven't checked. Let me check."
• "That's pretty cool. What type of Laura is that, and how are you training it?"
• "You said Dreambooth, so are you using a Google App or are you using it locally?"
• "Also, hey, what's up, Ghost? How are you doing, Pete?"
• "Any advice or any tips on training?"
• "Hey, Chris, just real quick wanted to know is dream booth still to go to or am I missing out on something better?"
• "But anytime we're ready X wave, we love to ask a couple of questions about how you how you do training so you can optimize a little bit better."
• "Because I haven't seen someone say that Koya wasn't the best way to do it, but I could I could be wrong."
• "Hey, what's up, Chris? What do you think about all this training stuff? Have you trained a model? Have you made a Laura?"
• "Have you guys tried some of the style transfer stuff?"
• "How do you feel about it compared to just using a model of your stuff?"
• "Any twos or something like that?"
• "Journey or what kind of tools to create your art?"
• "Is those packages, does it allow you to do like the output?"
• "Does it become like a living output so that anybody could actually query it?"
• "Like by just tapping or touching to see what would happen?"
• "Are those?"
• "Did you see, like, I don't know, I want to go back to 2012. Did you happen to see the Watson display up at, it was in New York."
• "And I want to say it was at Rockefeller Center, but I could be wrong about that."
• "But it was amazing."
• "Like it was touch screen, it was mapping, it was data visualizations."
• "You could basically move the nodes around and get more information from it."
• "I always love that."
• "Yeah, I didn't get to see it."
• "Did it show a robot or is it maybe showing off some different computers?"
• "Have you tried face analysis?"
• "Is this something I always wanted for AI art ever since someone someone point."

Overall Sentiment: The overall sentiment is positive and enthusiastic. Participants are passionate about AI art, eager to share knowledge, and helpful in answering each other's questions.

Closing Remark/Slogan: "Tweet your AI art even if it is a problem."

Upcoming Events/Future Spaces: The provided text doesn't mention any specific upcoming events or future spaces.

Opportunities to Share Art: There are no explicit opportunities for community members to share art in this specific conversation. However, the context suggests an open and encouraging environment where participants frequently discuss their work and processes. G., an artist from Iran, shares his experience of showcasing artwork on the "object" platform (potentially Foundation).

Jokes: There are no clear-cut jokes in the conversation. However, some instances display humor and lightheartedness:

• Chris mentions having to "dig through the porn" on CivitAI, highlighting a common experience when searching for specific types of models.
• Terry describes GPT-4 "laughing" at his request for a circular histogram and instead giving him a pie chart.

This summary offers a comprehensive look at the "AI Art Today" conversation, providing insights into the participants, topics discussed, and key takeaways.