Microsoft is moving away from OpenAI for Office 365 and bringing Anthropic on board. Claude models will be used for certain Copilot features in Word, Excel, and PowerPoint in the future—especially where they perform better according to internal tests, such as in spreadsheet
@kimmonismus
-
Movie-Grade Visual Control: Text and Audio-Driven Animation Technology
By
–
5. Why this is powerful: Movie-grade visual control: Combination of text-controlled global movement + audio-driven micro-expressions – appears lifelike rather than “doll-like.” Long-form videos: Minutes-long stable sequences via frame compression (similar to “hierarchical
-
Key Strengths: Visual Control, Long-Form Output, Non-Human Characters
By
–
4. In summary, the key strengths are: Movie-grade visual control: Audio + Prompt control global and local movements.
Long-form output: Minutes-long videos through hierarchical patchify method.
Non-human control: Also works with animals, animations, or stylized characters. In -
Audio-Driven Lip Sync with Realistic Facial and Body Movements
By
–
3. A still image + short vocal track creates realistic lip movement with subtle head and eye movements.
— Chubby♨️ (@kimmonismus) 9 septembre 2025
This demonstrates finely tuned audio-driven motion combined with text-based control of body posture. pic.twitter.com/diCk06jb0I3. A still image + short vocal track creates realistic lip movement with subtle head and eye movements. This demonstrates finely tuned audio-driven motion combined with text-based control of body posture.
-
Wan2.2-S2V: Stable Video Generation with Synchronized Gestures
By
–
2. A short spoken sentence is reproduced with synchronized gestures and facial expressions—stable, without image drift.
— Chubby♨️ (@kimmonismus) 9 septembre 2025
Wan2.2-S2V thus demonstrates its ability to deliver longer, consistent video output. pic.twitter.com/u4QrECZZtY2. A short spoken sentence is reproduced with synchronized gestures and facial expressions—stable, without image drift. Wan2.2-S2V thus demonstrates its ability to deliver longer, consistent video output.
-
Open-Source Wan2.2-S2V: Audio-Controlled Cinematic Video Generation
By
–
1. Open source, cinematic look, audio-controlled: a novelty. Today, I'm testing Wan2.2-S2V—an open-source speech-to-video model that generates cinematic facial expressions and body motion from audio. This feature makes it possible to accurately match audio to motion using open
-
Apple Intelligence Live Translation Feature Test
By
–
Live-translation by Apple Intelligence… well…. lets see how that works
-
ASML’s €1.7 billion investment in Mistral AI: European AI funding analysis
By
–
A few comments on ASML's investment in MistralAI There are currently many expressions of joy coming from Europe that the only well-known European AI company, Mistral AI, is now receiving relatively large funding of €1.7 billion from ASML based on a €10 billion valuation. This
-
XPeng Iron Humanoid Robot Mass Production Begins 2026
By
–
XPeng Iron is an intelligent humanoid robot developed by the Chinese electric vehicle company XPeng. They will start mass production in 2026.
— Chubby♨️ (@kimmonismus) 9 septembre 2025
In 2026, the fierce battle for the mass market for humanoid robots will begin. pic.twitter.com/WomxJahbWsXPeng Iron is an intelligent humanoid robot developed by the Chinese electric vehicle company XPeng. They will start mass production in 2026. In 2026, the fierce battle for the mass market for humanoid robots will begin.
-
AI Ends Podcasting Industry with Elevenlabs Podcast Generator
By
–
It’s so over for Podcasting. Another profession that was ended by AI. Elevenlabs nailed it with the Podcast generator pic.twitter.com/7bJVmTp2lX
— Chubby♨️ (@kimmonismus) 8 septembre 2025It’s so over for Podcasting. Another profession that was ended by AI. Elevenlabs nailed it with the Podcast generator