As predicted, Hollywood is done.
Over the past couple years, we discussed how Hollywood is seeing threats from all over the place. Much of the conversation centered on the epicenter of entertainment and how Southern California was losing that.
Other geographic regions have stepped up. They are taking a significant portion of the filming, for both television and movies. Unfortunately, at least for them, this might be short lived success.
Hollywood's death comes from two technological waves:
The first is the destruction of the monopoly on distribution that came as a result of the Internet. It is something we saw evolved over the past 15+ years.
Now, we are seeing the green sprouts of the second wave: AI.
Image generated by Ideogram
People fail to recognize the disruption that technology poses. Often, this can be overlooked due to the slow time to evolve. When changes happens over decades, many are not aware.
As stated two years ago, this was not the case with AI. We were looking at a radical transformation in the technology. In spite of that, many protested.
We heard pushback such as "AI cannot be creative" or "it will never write music as well as humans". To focus upon Hollywood, some claimed AI would never write legitimate movie scripts.
Naturally, all of those have been dispelled. AI can do all of that. Of course, one can debate the level of quality that is presently being offered. However, the output has surpassed the expectations that most of the naysayers had.
Then we have video generators.
Before getting to that, let me lay out my forecasts.
First, I claimed that online ads (social media) would see a significant production from AI generators by the end of 2025. This is the low hanging fruit where quality is the lowest. Since most surf on mobile. the screen size makes overall quality expectations lower.
The second was that feature length films would eventually be replaced by generative AI. This was gong to proceed to the point where anyone would prompt what they desire to see.
Does this mean all human movie making goes away? No. However, the majority of what consumes most of people's time would be of "good enough" quality.
A number of video generators have been talked about. Sora is probably the leading one, coming from OpenAI. Many Hollywood studios are testing this product, helping to train it.
That said, we are now looking at the emergence of open source models which some believe are rivaling the closed ones.
This is leading to much longer length videos coupled with consistent quality.
The latest breakthrough comes in extending video duration beyond the typical few seconds, with two new models demonstrating the ability to generate content lasting minutes instead of seconds.
In fact, SkyReels-V2, released this week, claims it can generate scenes of potentially infinite duration while maintaining consistency throughout. Framepack gives users with lower-end hardware the ability to create long videos without burning out their PCs.
Here is a sample:
There are a couple flaws in the video but, overall, this is fairly impressive. We can also see how it is 30 seconds. Of course, this is far from a feature length film yet it is a starting point.
Video is a much tougher medium to work with as compared to text. Nevertheless, we are seeing an improvement curve that follows that of text and images.
The key is to not look at where things are today. Instead, focus upon where they are going. What will this technology be like in 2027? That is only two years away.
It seems like the pace is becoming clear. By the end of the year, the above video will be suitable for social media advertising, with this promoting a cosmetic company. Put some audio over the top and it is a full 30 second ad.
Open Source can move a lot faster than closed once things get going. If the open source video generation market is starting to take off, development could be at a frenzied pace.
