What is Sora? Open Ai's New Text-to-Video tool will kill film productions, or not?

I'm being tagged with these questions in various chats today :) Indeed, if you look at the videos generated by the new neural network (Sora) from OpenAI, it seems like the end of professional production or does it?

What do you think? Will clients switch to neural networks for creating video content? Write in the comments!

‍

Let's try to reason about it.

How will this change cinema?

What Sora has shown (I remind you, this is still a beta version and it's not publicly available) looks very cool. Imagine all these videos generated simply by text! The entire production in the hands of just one screenwriter! Oh well, without specialists like ChatGPT :D

But there's a catch - the videos have glitches: sometimes the bird unexpectedly disappears, the fur doesn't move, or the physics of movements fails. I understand that all of this will definitely be fixed, it's just a matter of time. But right here and now, this isn't cinema killer. Such a result won't be allowed on the big screen. Also, there are serious doubts about the neural network's ability to generate blockbusters with explosions and effects - physics, simulation of tens of thousands of particles, etc. So far, the ability of the neural network to do this is not evident from the current samples.‍

How will this change the advertising market?

Here, too, not everything is so obvious. On the one hand, the neural network will definitely generate some kind of fashion video shot on a cyclorama. But how accurately will it reproduce the client's product? Let's say our product is a jacket with a lot of patterns and seams. Will the neural network replicate it 1 to 1 and how accurately? In case of even small potential errors, the client will most likely choose the proven old shooting method.

But you know where it will catch on in the very near future? In testing animatics of advertising videos! The cost of an animatic will be next to nothing with such a neural network. We'll be able to test dozens and hundreds of creative ideas before their production and spending budgets on placement. Currently, the problem is that animatics are not photorealistic and tests are not always representative.

Will production specialists be left without work?

Definitely not. In the next 2-3 years. Over this period, the grid will definitely be refined to solve the tasks I described above. And the same amount of time will be needed for clients to take this tool seriously. We, as a production team, will definitely actively test and squeeze all the possibilities out of the neural network after its release. Freelancers should also start testing grids (and should have already) - the market will gradually transition to them. Those who integrate them into their pipeline most effectively will be at the helm.

Will production become cheaper?

No. It will become more expensive. Yes, it seems that agencies will initially present this as a marketing gimmick, like "part of your video will be made by artificial intelligence... wow, how cool," and will leave the previous price tags or even raise them slightly. But when clients figure out what's what and many productions appear that have already implemented the neural network, then yes, the battle for clients will be through price reduction.

Will anyone be fired because of this neural network?

I think creators of stock videos will feel huge competition very soon. Stocks are often used either for cutaway shots in any content, or they are used to create various corporate videos where the exclusivity of the shot is not particularly important. But there is always a problem when working with stocks - you want something a little different: to add a sunset to this shot, or to have two trucks instead of one here. The neural network will do this. Competition in stocks will be huge.

We are in for an interesting and fun future, we observe :) And while I go simulate in Blender and write prompts in ChatGPT.

‍