Video Generation

Today's Stories

1. 1

   Microsoft Research builds AI video that remembers what's off-screen — so scenes stop falling apart

   Microsoft Research, working with several universities, released a system called Mirage that gives AI video generators a kind of spatial memory (the ability to keep track of where things are in a scene, even when they're not visible). Previous tools would "forget" a room's layout the moment the camera moved away, causing jarring inconsistencies. Mirage stores scene information in a compact format called latent space (a compressed mathematical map) rather than heavy pixel data, so it runs faster and uses less graphics memory while keeping scenes coherent through long camera movements.

   If you use AI tools to create videos — for marketing, social media, or presentations — future tools built on this research will produce scenes that look much more realistic and stable, without objects mysteriously changing or disappearing when the camera moves.

2. 2

   xAI's Grok Imagine can now turn your photos into short video clips

   xAI, the AI company founded by Elon Musk, updated its image tool Grok Imagine to version 1.5. The new version lets users upload a still photo, write a text description, and receive a short video clip at 720p resolution (roughly standard HD quality). Multiple clips can be stitched together to create longer scenes.

   If you use Grok on the X platform (formerly Twitter), you can now animate your own photos or images with a simple text prompt — no video editing software or technical skills required.

3. 3

   Luma AI, a video startup, opens a robotics lab that anyone can access

   Luma AI, which built its reputation making AI video generation tools (software that creates video from text descriptions), announced on June 1 that it is opening a robotics lab. The lab lets any developer or researcher train robots using Luma's software, extending the company's AI beyond video into physical machines.

   This means the same underlying technology that turns text into video could soon help teach robots to perform real-world tasks — a shift that could eventually affect manufacturing, logistics, and home automation.

4. 4

   A Reddit user asks: can you make AI videos for just $10 a month?

   A user on Reddit's AI_Agents community posted a practical breakdown of what AI video generation actually costs for a regular person — specifically, creating 3–4 minutes of animated video per day from text prompts. The discussion highlighted tools like Veo 3 (Google's video AI) and Kling as popular choices among everyday users, with the challenge being that most good tools cost more than $10 per month.

   If you want to use AI to make videos for a YouTube channel, small business, or personal project, the tools are getting more affordable — but finding a complete package under $10 a month is still genuinely difficult.

5. 5

   Az8 Studio offers a new way to build AI videos using a visual "node" workspace

   A Reddit user shared first impressions of Az8 Studio, a video creation tool that replaces the usual "type a prompt and hope" approach with an infinite visual canvas where different AI tools (for video, audio, and images) are connected like a flowchart. Users can generate a character in one part of the canvas and automatically pass that information to adjacent video or audio generation steps, allowing several versions of a scene to run at the same time.

   For small creators or marketers who build videos regularly, this kind of workspace could save hours of repetitive uploading and downloading between separate apps — everything stays connected in one place.

6. 6

   Real users reveal which AI video tools are actually winning in day-to-day use

   In a Reddit thread asking which AI tools people are actually using across different categories, Veo 3 (Google's video generator) and Kling emerged as the two most-mentioned AI video tools among active users as of early June 2026. For voice, ElevenLabs was the clear favorite. The thread reflects a broader pattern: a handful of tools are pulling ahead in real-world adoption while dozens of competitors fade.

   If you're deciding which AI video or voice tool to pay for, Veo 3, Kling, and ElevenLabs are the ones most real users are sticking with right now — a useful signal before spending money on a subscription.