VP Land
Posts
The Biggest AI & Video Updates in 2024

The Biggest AI & Video Updates in 2024

Joey Daoud
December 27, 2024

In partnership with

With 2024 ending, let's look back at some of the most significant video tech updates (and a few highlights from VP Land).

Have a great New Year, and see you in 2025!

Joey

What happened this year in AI and video

AI Video Generation Takes a Big Leap

In February, OpenAI released demo clips from their new AI video model: Sora.

It took the internet by storm. It was a huge leap in generating realistic video clips from just a text prompt. We called it 'GenAI's Lumiere Train Moment.'

However, access was kept super limited. During that time, we saw really impressive models come out of China: Kling and Hailuo.

We saw fully generated commercials using Sora, notably Toys”R”Us and Coca-Cola (but most AI videos still require extensive compositing using traditional VFX techniques to achieve a high-quality final frame).

OpenAI finally opened access to Sora just a few weeks ago, 10 months after announcing it (which is basically 10 years in AI time), with a staggering $200 a month price tag.

With all the new models that had come out since then, including big leaps with Runways' Gen-3 model, the reaction was a bit ho-hum.

Then, to fully dampen the buzz from the release, Google surprised everyone with its own new video model just a few days later, with Veo 2.

Still under limited access, the general consensus is that it consistently produces far better quality without needing to pull the AI slot machine handle multiple times.

This all happened in less than a year. 2025 is going to be very interesting for fully generated, photorealistic videos.

More Ways to Control AI

We also saw a lot of new ways to control AI outputs beyond just a text or image input.

Runway rolled out Act One, where a human performance can drive a reference character. They also added new camera control visuals to see the camera movement in 3D space.

Midjourney has had style references but took it to a new level with Moodboards. Feed it a board of images with the style you're looking for and generate new images following that style. Really powerful way to generate with a specific look by using images instead of convoluted text prompts.

Playbook has some of the most powerful tools for creating images that adhere to a composition, including the ability to control and frame a 'real' camera in 3D space.

Luma revamped their AI Dream Machine portal with a variety of new prompt features - context pills offering new suggestions, Boards for organizing and developing visual concepts, Brainstorm for idea generation, and Reply for image modification and style transfer.

New Cameras: More Affordable, More Powerful

We got a handful of new, exciting cameras this year. Panasonic released the GH7, which among other updates, supports ARRI's LogC3 color space.

Canon released the C400, a 6K full-frame camera with triple-base ISO.

And we got a bunch of new cameras from Blackmagic. The PYXIS, their first box-style camera with their Pocket Cinema Camera's 6K sensor.

Their URSA Cine 12K camera, their first build targeting high-end cinema cameras, yet relatively affordably priced at about $15K. Building off that is a 17K version, which was the precursor to one of the most interesting camera announcements the entire year - the URSA Cine Immersive camera, a product specifically built to shoot content for Apple Vision Pro.

Editing Gets More AI and Cloud Support

Every major NLE now has some form of built-in captioning/transcription using AI (even Final Cut).

We're also (finally) getting better tools to tag, sort, and manage footage in the editing process. Strada is doing this in the cloud. Axle AI is doing this on-prem.

We also saw better cloud support in post. The biggest one for our team: Blackmagic Cloud. I could go on and on about how well integrated the entire system is (though there is still room for improvement), but check out our videos.

Postlab also had a complete revamp of their excellent software with version 2.0. If we were using Premiere or Final Cut, I would get the team synced with Postlab - they had the easiest, most reliable implementation of LucidLink and check-in / check-out process for project files.

SPONSOR MESSAGE