New DeepSeek & Gemini, Wan 2.5, Alibaba dominates AI, 3D scenes, new deepfake lipsync: AI NEWS
As open-source 3D-guided video generation and advanced segmentation tools like Lynx and Hunyuan3D-Part emerge, filmmakers gain unprecedented control over character consistency and camera motion without relying entirely on costly closed-source platforms.
What this lesson covers
This video explores 'video from 3D', an AI workflow taking a reference style image and a moving 3D model to generate consistent video output with precise camera control. It also covers ByteDance's open-source Lynx reference-to-video generator and Tencent's Hunyuan3D-Part segmentation tools.
Key takeaways from the creator
AI-extracted notes, not independently verified product claims. Timestamp links let you check each point in the original video.
- 1:01 ↗
Video from 3D combines a style reference image with a moving 3D object or scene to generate consistent video output while maintaining precise camera control.
- 2:20 ↗
The video from 3D process uses a sparse anchor view generation module followed by a geometry-guided generative inbetweening module to interpolate and smooth frames.
- 3:11 ↗
ByteDance's open-source Lynx allows users to upload a reference character photo and a prompt to generate high-action videos with strong facial accuracy.
- 5:18 ↗
Tencent's Hunyuan3D-Part uses P3 SAM to convert a 3D model into a point cloud and segment it into meaningful parts, which are then reconstructed using XART.
Workflow outlined in the video
- Visit the GitHub repository linked for 'video from 3D' to review instructions for local installation and execution.
- Download and test ByteDance's open-source Lynx model locally under its Apache 2 license for reference-based character video generation.
- Experiment with Tencent's Hunyuan3D-Part online demo to explore 3D model segmentation and part decomposition.
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.
How to interpret the numbers
103,403 views captured 23 October 2026. ReelStack recommendation score: 22/100. Formula: radar-v5. This is ReelStack’s calculation, not a YouTube rating or a measure of factual accuracy.
Show the calculation
- 45% performance: 0.42× views vs 7 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: 0 views/day vs 12 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