20th Century Fox Is Using AI To Analyze Movie Trailers, Find Out What Films Audiences Will Like (theverge.com) 123
20th Century Fox is using AI to predict what films people will want to see. According to a recently published paper, researchers from the company are using machine learning to examine trailer footage and compare the objects and events it identifies with data generated for other trailers. "The idea is that movies with similar sets of labels will attract similar sets of people," The Verge reports. From the report: As the researchers explain in the paper, this is exactly the sort of data movie studios love. (They already produce lots of similar data using traditional methods like interviews and questionnaires.) "Understanding detailed audience composition is important for movie studios that invest in stories of uncertain commercial," they write. In other words, if they know who watches what, they will know what movies to make.
To create their "experimental movie attendance prediction and recommendation system" (named Merlin), 20th Century Fox partnered with Google to use the company's servers and open-source AI framework TensorFlow. In an accompanying blog post, the search giant explains Merlin's analysis of Logan. First, Merlin scans the trailer, labeling objects like "facial hair," "car," and "forest" (taking into account how long these objects appear on-screen and when they show up the trailer). By comparing this information with analyses of other trailers, Merlin can try to predict what films might interest the people who saw Logan.
To create their "experimental movie attendance prediction and recommendation system" (named Merlin), 20th Century Fox partnered with Google to use the company's servers and open-source AI framework TensorFlow. In an accompanying blog post, the search giant explains Merlin's analysis of Logan. First, Merlin scans the trailer, labeling objects like "facial hair," "car," and "forest" (taking into account how long these objects appear on-screen and when they show up the trailer). By comparing this information with analyses of other trailers, Merlin can try to predict what films might interest the people who saw Logan.
The problem is the measurement (Score:5, Insightful)
"People" are not a fungible quantity; they differ greatly as individuals, groups, and even between regions. Instead of trying to make a movie to make a generic consumer happy, make a quality movie that people who like quality cinema can recognize. Others will emulate them.
Re: (Score:3)
Instead of trying to make a movie to make a generic consumer happy, make a quality movie that people who like quality cinema can recognize.
They don't want to make quality movies. They want to make movies that make money.
A quality movie might come out as a waste product.
But then, it depends on how you define quality. Run Run Shaw was once asked, "What type of movies do you like best?"
He answered, "Those that make money."
Re: The problem is the measurement (Score:1)
They want to make trailers that sell.
Re: (Score:2)
Is it really so difficult, they want a good story well told. They probably don't want the same story told over and over again, often told worse than it was before. You can cheat, scope out all other media and check out what stories people liked and tell them in the movie format but remember to tell them well, so fuck off the nepotistic no nothings, that make things look good but tell really shit, clumsy, stupid stories because that is the limit of their intellect and they are to egoistic to hire smart peopl
Re: (Score:3)
Re: (Score:1)
There is plenty of room for movies that "make sense" or "tell a good story" but you'll generally be limiting yourself and your audience to people who want to think and analyze.
Yes, but this smaller audience is still worth catering to, as many of them simply won't go seeing those robot/explosion movies. Smaller audience merely means smaller budget, but that is hardly a problem. Use decent but unknown actors - they are cheap. This kind of movie don't need much special effects or props, also cheap.
Re: (Score:3)
A friend of mine owns a small chain of theaters here in Ontario, they're in mostly small cities/towns, where the old theaters died out and they never replaced them. For him, this was the first fiscal year in the last 20 where he's operating on a loss. Roughly, he had 35% less "asses in the seats" then the previous year. People are tuning out on movies, too many remakes of poor quality for example. Or too many trying to push an agenda. These things are either not creating that escape, or they're simply
Re: (Score:2)
At this point, for people like me, the movie theatre isn't really ever going to be an option again.
Well you're not a minority in this by any stretch. To me it looks like movies are now following the path of TV, engaging in simply making shit and expecting their audience to lap it up. I'd laugh my ass off if the Star Trek:TNG prediction about "that form of media not living past the mid 21st century" turned out to be true. (or was it early 22nd century? been a long time since I watched that series)
Re: (Score:2)
make a quality movie that people who like quality cinema can recognize.
How big is that market segment?
They are trying to make money, not win an award at Sundance.
This is what the AI is going to recommend: Fast and Furious Nine
Re: (Score:3)
This is what the AI is going to recommend: Fast and Furious Nine
Of course it will. They want to be able to churn out widgets exactly like the music business figured out how to do 15 years or so ago.
In the music industry's case they discovered that what Simon Cowell likes on his various TV "talent" shows can cheaply be packaged up and sold for a brief period before the next sound-alike comes along and starts the process again.
The movie industry wants to be able to do this too, and Fast & Furious 9 (or 10 or 11) is perfect for them.
Re: (Score:3)
Fast and Furious Nine
Honestly I'm looking forward to that. 7 was great. The Rock took out a drone by landing an ambulance on it.
Re: (Score:2)
make a quality movie that people who like quality cinema can recognize
There are many dozens of "quality" movies released each year - Great story, great acting, great cinematography. Most of them go bust because they majority of the movie viewing public wants "The Fast and the Furious Part 11" not "Children of Men."
Re: (Score:2)
Children of Men was commercially successful and will continue to generate revenues for decades.
The Fate of the Furious (to pick the most recent in the franchise) earned far more money far more quickly so I can understand investment going into films like that but it certainly doesn't preclude making good films.
That only applies when you're making something (Score:2)
When it comes to maximizing profit you can absolutely look at people and their tastes in aggregate form. Focus groups work. They produce a mediocre product at best, but it'll make the optimal amount of money.
Re: The problem is the measurement (Score:2)
Re: (Score:2)
Same with games (Score:2)
When they make a huge budget game that tries to please a mass audience, few succeed and it is a make or break endeavor for the studio/publisher.
Those who focus on their niche and try to please their core audience will foster loyalty and have guaranteed repeat sales that sustain them for the long term.
The Metallica Problem (Score:2)
I think of this as "The Metallica Problem" referring to what this band faced after they first started to get big (Master of Puppets era).
They could sell out, and get huge instantly, or keep making material in line with their past efforts, and be niche.
The advantage to being niche is that you do not then have to maintain huge band status, and get more of a chance to do what you wanted to without being obligated to the desire of your audience for your music to always become simpler, dumber, and more exuberant
examined footage (Score:5, Funny)
Lawrence of Arabia was an okay movie but it really could have been great if it had facial hair, car, and forest.
Re: (Score:2)
I guess they won't see the new Star Wars then as John Boyega is one of the most underated actors to hit Hollywood from UK for a while. See "Attack the Block", amazing low budget SciFi, very intense performance.
Simple (Score:3)
In other words, if they know who watches what, they will know what movies to make.
Make movies that don't suck *or* halve the admission price -- problem solved.
Re: (Score:2)
The people that would only watch movies at half price won't buy overpriced waste corn and sugar water.
Re: (Score:2)
That's a pretty incorrect comment. Movie theaters have manatee and weekday cheaper tickets that sell out and people buy lots of concessions. There are cheap theaters near me that show movies a few months old and they make all their money on concessions. People want cheaper tickets.
Re: (Score:2)
Does the manatee ticket come with a garbage bag full of popcorn and a 128-ounce drink? If so, I'm there.
Re: (Score:2)
Re: (Score:3)
Make movies that don't suck *or* halve the admission price -- problem solved.
I've never understood why good movies cost the same as bad movies.
Also, movies should cost more on the opening weekend, and then decline in price each week.
Re: (Score:2)
Up until the 80's that was the case, with A movies & B movies, the successes of B movies like Star Wars and others through the late 70's/early 80's erased the line.
The same thing can be said for length of movie. Something like Lord of the Rings being ~3x the length of some kids film, yet the price is the same at the same theater.
Re: (Score:2)
Make movies that don't suck *or* halve the admission price -- problem solved.
I've never understood why good movies cost the same as bad movies.
The movie industry financial folks have calculated that the demand curve is extremely inelastic. Even if they lower the price . . . more people won't go to see bad movies. Those folks will probably have a complex linear programming model to support this, but, in general, people tend to suspect that anything steeply discounted isn't top quality.
Try these two propositions on your mate:
"Let's have a fine bottle of wine with dinner, and then go see a good movie.
"Let's have a box of Walmart wine with dinn
Re: (Score:2)
I've never understood why good movies cost the same as bad movies.
Because they rely on people not knowing that a movie sucks and wasting their money on it. Before the internet existed it often took a while for word of a bad movie to get around. Nowadays that information spreads at internet speed. They haven't caught up yet.
Just look at movie trailers. They really don't want you to know the truth about most movies. Charging less for bad ones would just signal to people to avoid seeing them.
Also, movies should cost more on the opening weekend, and then decline in price each week.
Hollywood hates anything that devalues movies over time. Discounted DVD sales, renta
Re: (Score:1)
bigdatadayla.com (Score:1)
This was already mentioned that 20th Century Fox was using AI for trailers in the Big Data Day LA (bigdatadayla.com) symposium this year in 11 August 2018 at USC. I sat in on this particular session, towards the end of the day. It was hosted by Miguel Campo, SVP Data Science and Analytics, 20th Century Fox. He showed one trailer that was a control (without modification) and that had been recut using "Collaborative Filtering". Impressive, to say the least. Yes, still along way to go, but positive.
LOL (Score:2)
Re: (Score:2)
If you liked "Spiderman" you will probably like "Spiderman 2" or any of the awful super hero movies that have been out in the last 10 years. Genius.
That's what I'm thinking too, this algorithm will state the obvious but if you're trying to pick one of these five super hero movies to make next I really don't think the algorithm will capture what's actually a good character and story arc compared to what's not. I mean it doesn't really make sense to have a superspiderbatman, they're the same but they also have to be uniquely different. Even doing a prequel/sequel it has to build on the lore and character of the first movie. I can understand trying to ana
Re: (Score:2)
They've succeeded too. Batman/Spiderman/Ironman/Whateverman is just the same movie released a couple of times a year since the mid 1990's.
Well Sure, Mr Smarty Pants... (Score:3)
Re: (Score:2)
No-one liked Spiderman 1. They just really wanted a good Spiderman movie and were hoping that Spiderman 2 would be better.
Transformers is the worst of that. I so want it to be good... And now the new one has the proper G1 character designs, I don't think I can resist watching it.
Adverts that dont ruin the movie (Score:1)
When I watched Simpsons, the adverts ruined every joke in the movie. Start with not doing that. During skyfall, here in Australia, they actually reviewed skyfall before the movie.. Why?
Then how about releasing a few original movies?
Also why not release movies to Netflix or online faster? If the cinema/theatre experience is genuinely better (its definitely more expensive), people would still go instead of watching at home. Instead, people probably pirate them because they don't want the movie to get spoiled
Facial Hair, Car, Forest (Score:2)
That seems like exactly wrong way to do it. If you liked Logan, you would have probably liked it if it took place in the desert with a donkey, and he shaved.
Maybe though, they aren't talking about liking the movie, but rather if the trailer will hook them enough to go see the movie.
Even then, I think their analysis is pretty stupid. (BTW, I didn't read the article.)
How to make a good trailer (Score:2)
Re: (Score:2)
Easy! (Score:2, Insightful)
No SJW, no diversity quotas, no feminist appeasing, no kids.
Re: (Score:2)
Cartman hated it
Re: (Score:2)
Measuring the wrong thing (Score:2)
Re: (Score:2)
This approach measures whether people (may) like the trailer, not whether they'll like the movie.
Came here to say this. Maybe the studios think that a good trailer will drive attendance even for a bad movie. For the opening day or two this is more likely to be true, especially if they can pay off enough reviewers who are our only alternate source of information. Sadly for the studios, internet reviews make word of mouth very fast; if they don't have simultaneous worldwide releases then many audiences know in advance to avoid a stinker, however alluring its trailers may be.
There is probably no way to measure audiances (Score:5, Interesting)
by analyzing movie trailers. Modern trailers for upcoming movies usually spoil anything possibly good or interesting about new movies. I seriously avoid watching trailers for movies I want to see. Although for other movies, trailers can convince me which movies to absolutely avoid.
Data analysis of movie revenue based on actors/scriptwriters/directors would be a better start, but that would cost way more money, because you'd have to work around "Hollywood Accounting" practices to figure out how much money movies actually created.
This type of analysis will create another miasma of Spiderman//Superman movie "reboots" because "This Time for Sure" Isn't there another Robin Hood movie coming out soon?
Re: (Score:2)
Good film though.
I hate trailers (Score:2)
Then again, the 50% that don't get pissed off I seem to get along better with. huh.
Why do I hate trailers? Long time ago there was a movie where Forrest Whittaker (first time I noticed him) was a chick with a dick. Those last 4 words were the entire movie's pivot. And a fucking trailer gave it
Re: (Score:2)
How can AI work in this project? (Score:2)
No wonder their films suck! (Score:2)
Humans are not AI.
And while AI is great for figuring out lots of things, human response is one of those tricky things.
Especially as film hype and enjoyment change with saturation of awareness.
Take the Zamboni scene from Deadpool.
Funny as shit.
Now stick it in every last trailer and play them non-stop in the weeks before the movie comes out.
Interest and response to this particular joke have dived off significantly.
AI simply doesn't capture that...
Or they could just ask us movie-goers (Score:2)
Spaceships, lots of weapons fire, scantily-clad women. Otherwise not worth my money seeing on the large screen.
No expensive AI needed.
Re: (Score:2)
Bullets, Bombs and Boobs... the 3 B's of cinematic excellence.
Another case of writing by committee? (Score:1)
The problem is, it doesn't work. This is a way to get bland generic pap. It will probably be pleasing enough but nobody will care if they see it or miss it. it's predictable and repetitive.
so... (Score:2)
So given AIs mix of awesome success and amusing failures, we will soon see a feature-length movie featuring disembodied facial hair floating through a forest for two hours, mixed with special effects and a blasting soundtrack ?
great (Score:2)
they will find the perfect trailer formula and then all trailers will look the same, except for the title of the movie.
ofcourse, this has to a large extend, already been applied to movies as a whole, that's why every movie is basically the same.
praise the indies!
it's amazing how much the video games and movie industry is similar.
Re: (Score:3)
Re: (Score:2)
I think this will result in trailers that look fantastic but the movies will be total shit.
oh wait, we have that now so essentially you're right.
Innovations of AI (Score:1)
Re: (Score:2)
I read in reverse order, so I haven't gotten to that story yet.