Friday, October 04, 2024

Straight from the imagination to the screen.... Meta's video release

This is the promise of AI. And we’re getting tHere faster than anyone could have imagined. You become a video director, without the eye watering production costs. GenAI has moved way beyond text to the image, video, audio space.

Meta’s new ‘cast’ of models allow you to create but also edit and personalise images, videos and audio. Personalised means based on your or anyone’s face.


This is another astounding milestone... models that generate:

·     1080p HD

·     Up to 16 secs (16FPS)

·     Different aspect ratios

·     Synchronized audio 48 kHz

·     Instruction-based video editing

·     Personalised videos based on a user’s image


Note that 16 seconds doesn’t sound long but call them takes and they’re very long but TikTok built a global business on videos with a 15 second limit (now longer).


All from text prompts.


https://ai.meta.com/research/movie-gen/


LEARNING

Any teacher can create videos of them showing students how to do anything. That’s surely useful, especially in vocational learning but also in science, art, music and many other subjects. Call this personalised teaching.

 

It gives anyone the ability to create short (here’s 10 for startes):

Instructional step-by-step procedures

Explainer videos

Trigger videos

Branched scenario-based

Simulation videos

Animations

Video flashcards

Scenes too risky to film

Scenes impossible to film

Short messages from CEO etc

Ads for new initiatives


Long-form video suffers from the transience effect, but there are ways to make it more relevant and effective in learning, beyond the lecture and talking head.


I’ve been involved in many scenario-based video sims and this has just made them attainable on a low budget. Your video production costs on management training sims, such as interviewing, having difficult conversations and so on has just plumetted.


In performance support, huge numbers of short videos can be created ready for delivery at any moment of need in the workflow.


CONCLUSION

I can see this move towards longer takes, even drama. To be honest that may already be here. You’ll see a lot of creative fun stuff, like personalised birthday messages. The personalisations is interesting. Thought also has to go into impersonation and the avoidence of explicit material.

No comments: