If youtube is to be believed, thumbnails make a dramatic difference to first-glance engagement. A model trained to find great(by what metrics?) screenshots from the film sounds like a fun project.
I'm looking forward to classic movies with AI-generated thumbnails that take the main actor or actress and make them do the "Youtube Thumbnail Surprised Face". Imagine The Maltese Falcon or Citizen Kane but with thumbnails that look like [1].
That would require paying a contractor for ~5 minutes of time to skip through the movie until they find a good scene, take the screenshot, trim it, and upload it. Ugh. We've already got all these GPUs over in AWS, just spin up an image generation model and prompt it "Make a thumbnail preview image for the film {{film.name}}", good enough.
That'd take a human with taste a terrifyingly large 5 minutes, whereas AI trash can be entirely programmatic, with maybe an intern picking the best of 4 in a couple seconds. Think of the savings, and with only a moderate impact on the reputation of your service.
Good point. I wonder if Amazon is experimenting with this with the idea that they will be able to make more "eye grabbing" thumbnails with genAI as opposed to pure stills from the given film. The click through rate may be a key metric internally, and since prime video is an add-on to prime the risk of customers dropping the service over this is low (compared to if Netflix did this)
What percentage of people will notice the primary means of marketing the film?
Almost all of them, surely. Maybe not consciously... but unconsciously certainly.
Films are incredibly expensive products to produce, an average of $100 million dollars apparently [1]. Do the incentives really not align to pay for professional artists to create a high quality thumbnail? I doubt it. Hell, youtubers have figured out it's worth doing this on every video which cost and pull in 4-5 orders of magnitude less money.
I loved that remake! Zombies intervene on jury deliberations, devouring one or two dissidents. The rest are able to negotiate a detente and quickly reach a unanimous verdict.
As someone from a country that uses subtitles over dubbing, I'll let you know that subtitling movies/tv shows is an art form. Good subtitles are more akin to interpretation than translation. You have less space to fit, timing is important, you need to translate idioms, etc.
Probably an art form that's going the way of the dodo, but nevertheless...
> Good subtitles are more akin to interpretation than translation. You have less space to fit, timing is important, you need to translate idioms, etc.
Timing particularly! You'll often get a mystery where the detective says "And the killer is … [long, dramatic pause to survey the room] … the butler!", but the subtitles, displayed at the beginning of the sentence, just say "And the killer is the butler!"
I've noticed HBO seems to have the highest quality subtitles in terms of timing and position on screen relative to the character speaking.
Then there's other shows I watch on Netflix will do that exact thing you mentioned - spoiling the punchline because the subtitle appears before a single word is spoken.
Oh, I forgot that Hulu has an infuriating thing where, if you turn on English-language subtitles for a show that's mostly in English (as those of us with poor hearing often do), and the show has burnt-in subtitles for portions of speech in some other language, then the burnt-in subtitles will be overwritten by a subtitle reading, say, "[Speaking in French]."
> Good subtitles are more akin to interpretation than translation.
Translation is interpretation to start with, but subtitling definitely adds constraints around what works well that don't exist for, say, generic prose in a purely written context. Similar to (but different specific constraints) to translation for dubs.
> Good subtitles are more akin to interpretation than translation.
Why would I want english subs for english audio to be anything than the lines coming from actors' mouths. English subs are the only language you're gonna get from english audio transcription regardless of the underlying technology.
Text-to-speech is AI, but we used to stop calling things AI once they became practical. AI used to mean "things computers can't do yet". That was actually the main reason AI never took off.
Prime Video has consistently had the worst quality subtitles of any streaming service I've tried. I'm surprised they haven't been subject to a ADA-type lawsuits over it.
I'm suspicious that a rights holder would be submitting an AI thumbnail. Still, anything that Amazon puts in Prime is something they should be willing to stand behind.
Why is that suspicious. The rights holders core competency is in making the content. If they can offload poster/thumbnail creation then I'm sure they would.
Part of that core competency is promotion isn't it? A part of that used to be posters to put in the lobby of a movie theater advertising what was coming soon. Much later home media needed a cover for the boxes. If I search for the movie on another amazon property I'd expect to see that poster.
I think it's important to note that this happened on a very old movie. In the 67 years since it's original release, most people creatively attached to it have long since died. The person uploading it to Amazon is nothing more than a corporate bean-counter.
> Still, anything that Amazon puts in Prime is something they should be willing to stand behind.
Have you looked at their catalog? They publish movies that creators can't even bother to burn to DVDs.
I wonder if it could have an impact on the development of the brain in children, if they grow up consuming realistic looking media that has subtle errors.
Watching a movie takes too long. We just need an AI movie-summarizer that'll convert a 2 hour long movie into a bite-sized 60 second long TikTok for human consumption.
This is the future of all online communication: the sender uses an LLM to expand a few sentences into a few pages, and the receiver uses another LLM to summarize it.
A few weeks ago I was looking at albums in Apple Music. I read the blurbs on several albums in a row from a particular artist, and I noticed that they were all overly wordy, banal, and extensively reused the same rare words. So I searched on the subject and of course all the album background blurbs were AI-generated. Apple bought some company specifically for that purpose a few years ago. So now I know to not waste time reading the album blurbs on Apple Music.
Annoying, because you can waste a few minutes reading the site before you realize that it's junk.
It even pretends to be a person, with statements like "Personally, this song resonates with me on multiple levels. The story within the lyrics reminds me of the value of genuine connections and the joy of reuniting with loved ones!"
I was trying to figure out some of the lyrics to "Harold of the Rocks" a while back, stumbled across that site and holy shit:
> The lyrics of Harold of the Rocks are metaphorical and open to interpretation. They tell the story of a character named Harold, who is portrayed as an outcast in society. He stands out from the crowd and refuses to conform to societal norms, embodying a rebellious spirit. The song challenges the idea of conformity and expresses the band’s belief in the importance of staying true to oneself, regardless of the pressures to conform.
> In a broader sense, Harold of the Rocks can be seen as a critique of a society that values conformity over individuality. It urges listeners to embrace their uniqueness and celebrate their differences instead of succumbing to societal expectations. The lyrics provoke thought and encourage listeners to question the status quo, making it a powerful anthem for those who feel like outsiders or outcasts.
As a rule, I avoid any text which, on the internet, purports to explicate the meaning of some work of art, high or low.
Even before large language models began remixing this drivel, it was drivel. There is no part of youtube more accursed than those videos which wish to tell me what some film or discerning work “means.”
One must elucidate such, if such exists, in oneself, and in the discussions of true critics. Never from that mumbling middlebrowery now being remixed into nonsense lowbrowery.
Recently I was reading the news and saw an article from USA Today with an AI-generated cover graphic of Donald Trump.
And with both this and 19 Terrifying Men, I'm at a loss why anyone would do that. Surely people have already made thumbnails from 12 Angry Men, why did we need to change to a worse one?
Similarly, there must be tens of millions of photos of Donald Trump, from every stage of his life, available to license. Surely USA Today has a subscription to an archive where they can pull a stock image of Trump for no marginal cost.
What is the thought process here? Do they have a new hammer and see everything as a nail? Or is there a reason this makes sense that I'm not seeing?
Easily the cheapest way to get decorative pictures where reality doesn't matter much - at least for now, until the law in various countries catches up to this practice.
68 comments
[ 0.26 ms ] story [ 131 ms ] threadThis reminded me of this article I read a few years back. I wonder if they still use the same method.
https://netflixtechblog.com/ava-the-art-and-science-of-image...
https://netflixtechblog.com/artwork-personalization-c589f074...
https://www.theguardian.com/media/2018/oct/20/netflix-film-b...
1: https://hard-drive.net/wp-content/uploads/2021/05/youtubers-...
Almost all of them, surely. Maybe not consciously... but unconsciously certainly.
Films are incredibly expensive products to produce, an average of $100 million dollars apparently [1]. Do the incentives really not align to pay for professional artists to create a high quality thumbnail? I doubt it. Hell, youtubers have figured out it's worth doing this on every video which cost and pull in 4-5 orders of magnitude less money.
[1] https://www.investopedia.com/financial-edge/0611/why-movies-...
I loved that remake! Zombies intervene on jury deliberations, devouring one or two dissidents. The rest are able to negotiate a detente and quickly reach a unanimous verdict.
Probably an art form that's going the way of the dodo, but nevertheless...
Timing particularly! You'll often get a mystery where the detective says "And the killer is … [long, dramatic pause to survey the room] … the butler!", but the subtitles, displayed at the beginning of the sentence, just say "And the killer is the butler!"
Then there's other shows I watch on Netflix will do that exact thing you mentioned - spoiling the punchline because the subtitle appears before a single word is spoken.
Translation is interpretation to start with, but subtitling definitely adds constraints around what works well that don't exist for, say, generic prose in a purely written context. Similar to (but different specific constraints) to translation for dubs.
Why would I want english subs for english audio to be anything than the lines coming from actors' mouths. English subs are the only language you're gonna get from english audio transcription regardless of the underlying technology.
> Still, anything that Amazon puts in Prime is something they should be willing to stand behind.
Have you looked at their catalog? They publish movies that creators can't even bother to burn to DVDs.
It's awful for various reasons, but I'm sure AI video will be pretty good, pretty soon.
Edit: Noted by jsnell elsewhere here: <https://news.ycombinator.com/item?id=40599828>
https://www.youtube.com/watch?v=axCBA3VD5dQ
https://netflixtechblog.com/artwork-personalization-c589f074...
Annoying, because you can waste a few minutes reading the site before you realize that it's junk.
It even pretends to be a person, with statements like "Personally, this song resonates with me on multiple levels. The story within the lyrics reminds me of the value of genuine connections and the joy of reuniting with loved ones!"
> The lyrics of Harold of the Rocks are metaphorical and open to interpretation. They tell the story of a character named Harold, who is portrayed as an outcast in society. He stands out from the crowd and refuses to conform to societal norms, embodying a rebellious spirit. The song challenges the idea of conformity and expresses the band’s belief in the importance of staying true to oneself, regardless of the pressures to conform.
> In a broader sense, Harold of the Rocks can be seen as a critique of a society that values conformity over individuality. It urges listeners to embrace their uniqueness and celebrate their differences instead of succumbing to societal expectations. The lyrics provoke thought and encourage listeners to question the status quo, making it a powerful anthem for those who feel like outsiders or outcasts.
It's a song about a crack addict, people. https://www.youtube.com/watch?v=LvNWhr9BmcQ
Even before large language models began remixing this drivel, it was drivel. There is no part of youtube more accursed than those videos which wish to tell me what some film or discerning work “means.”
One must elucidate such, if such exists, in oneself, and in the discussions of true critics. Never from that mumbling middlebrowery now being remixed into nonsense lowbrowery.
And with both this and 19 Terrifying Men, I'm at a loss why anyone would do that. Surely people have already made thumbnails from 12 Angry Men, why did we need to change to a worse one?
Similarly, there must be tens of millions of photos of Donald Trump, from every stage of his life, available to license. Surely USA Today has a subscription to an archive where they can pull a stock image of Trump for no marginal cost.
What is the thought process here? Do they have a new hammer and see everything as a nail? Or is there a reason this makes sense that I'm not seeing?
Easily the cheapest way to get decorative pictures where reality doesn't matter much - at least for now, until the law in various countries catches up to this practice.