@jsnnsa Can you please have the models create SKILL.md that can generate these type of videos for any project by analzying codebase so can use to create videos for any project. A skill.md will be more detailed about the how than asking and working like slot machine results
Some one from @Meta@MetaVR_Official explain why would team ship meta quest VR browser with default extensions but no way to install others including custom
@mattpocockuk Is that not what AI deep research for to research signal from the noise? At this point you are pro at writing skills why not write 1 to research and find the signal
@justinryanio When you finish have astra create SKILL.md based of final output and sessions history for recreating have it follow the skill and cross check new raw output matches existing final output
@RyanSael Possible business If AI could analyze multiple videos of same real concerts, figure out angles and mimic the concert to show what would look like then use to book the same artists next show
@indystryfilms Ya need to follow @AI_search and not sleep on Blender MCPs along with good prompting if prompting sucks it changes how well good tools work. The models exist and can be done with good harness architecture and tool calling. @spawn good example of architecture with opus models
@kushal_to Want access to my WebXR Infinite canvas, it imports markdown, PDF, and HTML converts to content on Infinite Canvas can organize how you want. Its still reading so now on to figure out how to make it represent the text content to different 3D experience
@superalesha StepI think answer depends on hardware running on. Frontier cloud models are running on good hardware. Step Flash 3.7 was my go to from @StepFun_ai