Seedance 2.5 & The New Age of AI Video (w/ Theoretically Media)
Topics discussed
Intro: New AI video models and the rise of Chinese tech
Guest host Tim joins; overview of Minimax H3, WAN 3.0, Flux 3
CDance 2.5 experience: fashion shoot attempts and rejections
Prompting CDance 2.5: short film breakdown, costs, and dialogue
30-second outputs, consistency, and reading the documentation
Prompting style differences: CDance 2.5 vs 2.0 and Halo-O
Minimax H3 local deployment: ComfyUI setup and installation
Local model capabilities: breaking time limits and quality tweaks
Benefits of local AI: cost, privacy, and unrestricted content
Ethics of AI video: IP usage, Seinfeld clips, and creator longevity
WAN 3.0 announcement: features, text document input, and open source
Flux 3 status: early access, updates, and open model rumors
US vs China in AI video: Meta, Elon Musk, and American efforts
Hank Green controversy and Google Omni model delays
Higgs Field terms of service controversy and user backlash
AI video platforms: Artcraft, Martini, and SynthWaved reliability
OpenAI models: Astra rumors, Claude Code, and token management
Google news: Demis Hassabis role change and Gemini strategy
Featured AI creators: Kevin's work, Dave's vertical video, Demon Flying Fox
Listen ad-free on Castria