Tesla Optimus stands 5’8” with 28+ custom actuators while MicroDuck is a 25 cm biped with 15 motors, ToF LiDAR and an 800 g frame. They perform household chores using vision neural networks and lightweight reinforcement learning, with the MicroDuck occasionally tumbling and self-recovering.
Key points
Optimus height: 5’8” and 28+ custom actuators
MicroDuck specs: 25 cm size, 15 motors, ToF LiDAR, 800 g frame
Sources
Dreamina Seedance 2.5 used for AI music video creation
Dreamina Seedance 2.5 was used to generate reference-to-video, performance and lip-sync for the AI music video “NON MI TROVI”. The workflow also incorporated Magnific, Suno, Topaz Labs and CapCut for various production steps.
Key points
Tool: Dreamina Seedance 2.5 for reference-to-video generation, performance and lip-sync.
Tools: Magnific for character, location and visual references; Suno for assisted song production; Topaz Labs for upscaling and refinement; CapCut for editing, colour grading and post-production.
Sources
Grok Imagine song creation and Grok Bot video 2:58
Grok Imagine wrote and generated a song for X Vibe Chat. Grok Bot produced a 2:58 music video with lyrics, beat-synced edit, the group logo and members across 8 cities, requiring no editing skills.
Key points
Song creation: wrote and generated the song with Grok Imagine
Video output: 2:58 music video made by Grok Bot with lyrics, beat-synced edit, logo and members across 8 cities
Sources
P-H3-Image handles costume changes automatically
P-H3-Image handles costume changes fully automatically by analyzing two input images with a self-made QwenVL-based model node, outputting JSON, and swapping it into a P-H3-Image prompt with a self-made JSON formatting node.
Key points
Analyzes two input images with a self-made QwenVL-based model node.
Outputs JSON and swaps it into a P-H3-Image prompt using a self-made JSON formatting node.
