GPT 5.2, realtime video editor, AI stereo videos, mobile AI agents, full body control: AI NEWS
Rapid breakthroughs in trajectory-based video control and real-time editing give AI filmmakers finer manipulation over object motion and post-production clean-up without traditional rigging.
What this lesson covers
This news roundup covers major AI releases, including GPT-5.2, Alibaba's Wan-Move for trajectory-based video motion control, and real-time video editing tools. It highlights advancements in photo reflection removal, realistic image generation, and mobile AI agents.
Key takeaways from the creator
AI-extracted notes, not independently verified product claims. Timestamp links let you check each point in the original video.
- 00:11 ↗
Alibaba released Wan-Move, a video tool allowing users to control object motion and camera movement by drawing trajectories on a start frame.
- 01:11 ↗
A new AI tool called Window Seat effectively removes stubborn window reflections and raindrops from photos while correcting lighting and colors.
- 03:21 ↗
RealGen introduces a detector reward mechanism to eliminate AI artifacts and improve image realism, outperforming standard Flux models.
- 06:48 ↗
Wan-Move requires substantial VRAM (~70 GB for the official model), but community FP8 ComfyUI integrations reduce requirements to around 16 GB.
Workflow outlined in the video
- Explore Alibaba's Wan-Move via GitHub or ComfyUI integrations to test trajectory-based motion control for video generation.
- Utilize Window Seat reflection removal tools to clean up reference photography shot through glass or windows.
- Test RealGen checkpoints to generate hyper-realistic photographic elements with natural grain and motion blur.
Before you use this workflow
These notes describe the source video at its publication date. Model access, pricing, connectors and interfaces may have changed. Check the original source and the provider’s current documentation before installing an add-on, connecting an account or spending credits.
ReelStack has not independently tested this workflow. Preview one representative shot and check motion, continuity and output quality before applying it to a full production. No result, cost saving or model capability is guaranteed.
Explore the tools
How to interpret the numbers
109,748 views captured 23 October 2026. ReelStack recommendation score: 39/100. Formula: radar-v5. This is ReelStack’s calculation, not a YouTube rating or a measure of factual accuracy.
Show the calculation
- 45% performance: 0.93× views vs 14 other channel videos observed at a similar age. View-evidence factor 99% (views / (views + 1,000))
- 25% freshness: 0/100 with a 14-day half-life
- 20% momentum: 2 views/day vs 3 channel baseline (7 comparable uploads), with the same view-evidence factor
- 10% engagement: 92/100; likes + 4× comments, smoothed with a 500-view neutral prior