It should be communicated in a clearer way.
32.0 GB (27.8 GB usable)
AMD Ryzen 7 8845HS w/ Radeon 780M Graphics 3.80 GHz
Very cool either way!
Once in depth mode I saw movie posters but it zoomed out until they where gone.
(Not working on Firefox mobile, but my guess is hugged to death)
N.B. I visit HN multiple times per day. I've been a member of HN since 2010. My last comment was made in 2018. I logged in for the first time in years just to make this comment :-)
An ocean of film posters is a cool visualization but doesn’t solve the problem well.
I had another job last year where they kept asking me to fine tune an LLM to answer questions about text, but I was only allowed to feed in the raw text of a single document but not generate an actual Q&A dataset. I kept telling them that was not how it worked but they kept insisting I continue for weeks. I think in his mind he just thought I was just doing it wrong over and over and he was secretly training me by paying me to practice or something.
I guess it's kind of sad how many jobs I've had that were a waste of time. I'm weirdly used to it so looking back it doesn't bother me that much somehow.
It does irritate me in the moment though. I have a client now that wants to make an agent system for a certain platform. The previous day we had just been discussing MCP servers and what they do. But the next day he seemed to think that each MCP server was going to have an entire full custom UI and agent loop in it per agent instead of just providing tool commands. So MCP server became an agent. I think I explained how it normally worked and he seemed to maybe get it, but then I was trying to ask how we were going to allow people to use all of the existing MCP servers in their agents so they could combine them and the non-response made it seem like he thought that might not be important. Did not get an answer so far. I remain ever hopeful for our upcoming meeting though.
Is this a sign of old age? Answers to something specific just become rambling walks down memory lane? It's interesting how it seems that I am generally well adjusted to how futile my life has been.
Anyway, I'm sure it's just a coincidence that Silverlight sent me down this tangent.
"This gallery features a collection of the 50,000 most popular* movies according to TMDB"
Yeah, there's the problem. I went to the TMDB site and tried sorting by popularity, rating … all I got was a whole bunch of recent films that look like all the reasons the film industry is crashing (you know, the kind of stuff you would see in a Red Box when they were still a thing).
If you really don't know what to watch you could do a lot worse than checking off films on this list (the 1001 Movies to See Before You Die):
https://1001films.fandom.com/wiki/The_List
(And you have already seen a number of them I expect.)
Elitist? Maybe. But if you can actually get through the list (I've been at it, casually, for several years now) you'll feel like you have acquired a film school degree.
I think it's fine to watch "8 ½" and hate it — at least you'll know that you hate it now. But it would be sad if you never saw "La Strada" because you thought all of Fellini's films were impenetrable.
"Marty", "Stella Dallas", "The Hustler" — just a few great American films that came to mind that I would have otherwise missed…
But if the latest Guardian of the Pirate's of the Marvel Universe movie are your cup of tea you probably have no problem finding something to watch anyway and can pass on "The List".
I'm sure one could make a "top 5 Marvel of the 783 released so far" movies and these 5 are actually pretty darn good.
The joke is though that the list of "top 5" could probably be picked from a pool of 15 good ones and it wouldn't matter. The 15 are all pretty good.
But if you watched all 15 good ones you'd be bored out of your mind because it's the same formula.
And if you watched all the ones which are merely "ok" it's like watching the more boring variants of anime where very little happens over very many hours.
(Specific numbers just made up for the argument.)
There’s some real gems on lists like this but some of the older stuff would definitely be a slog for most people and would probably need to be put into context to make sense.
Recently I’ve engaged in a practice of just hitting play on any old movie on Prime with at least an actor or director or even studio I’ve heard of. Essentially any film from the 40s to the 00s that had a budget and was remembered enough to make it to Prime. It’s been quite rewarding, all sorts of hidden gems out there even from the 70s through 90s. Even the mediocre stuff from then is often more nuanced and creative than a lot what you get today.
You're right. The book ("1001 Movies to See Before You Die") gives that context. I was surprised though that, as I went through them — slogging even, as you say, at times — in time I started to develop perhaps a new way of looking at film. And then slowly I found I was "slogging" a lot less. I began even thinking a bit more fondly back on the films I had seen earlier on that had been more of struggle for me to watch.
I think that's the gist: anything goes, and maybe start with 3 of anything!
...based on genre (Comedy): Tampopo (1985), Matchstick Men (2003), Force Majeure (2014)
...based on subgenre (Heist): The Killing (1956), Thief (1981), Sexy Beast (2001)
...based on director (Kelly Reichardt): Old Joy (2006), First Cow (2019), Showing Up (2022)
...based on production studio (Studio Ghibli): Only Yesterday (1991), Porco Rosso (1992), The Wind Rises (2013)
...based on location (Vienna, AT): The Third Man (1949), Bad Timing (1980), Before Sunrise (1995)
...based on interest (Desires): Deep End (1970), Exotica (1994), Ghost World (2001)
...based on language (French): L'Année dernière à Marienbad (1961), Un homme qui dort (1974), Portrait de la jeune fille en feu (2019)
...based on movement (Czech New Wave): The Sun in a Net (1963), Valerie and Her Week of Wonders (1970), Morgiana (1972)
etc.
- Ikiru (1952): A fantastic movie, also in the above list
- A Clockwork Orange (1971): Alright, nobody is going to believe these are random if I keep pulling these...
- Tracks (2013): Haven't seen this one, doesn't seem particularly astounding but I'd probably watch it, especially if my wife was wanting to watch something we haven't seen together.
- Songbird (2020): This one looks a bit garbage to me, probably wouldn't watch.
- Osmosis Jones (2001): Watched a few times, maybe a bit of nostalgia but it has been a few years so I would throw this up. Not deserving of the top 1000 by any means... but also not Leprechaun 5: In the Hood (2001).
And that seems pretty reasonable. I'm only about 800 movies in (since I've started tracking), of those only ~400 are in the original 1001 Movies You Must See Before You Die list. To me watching straight through the various compendiums of the best movies is actually more dull than mixing in other things like the latest pop films or some stuff that wouldn't make many "top" lists but isn't necessarily bad. It's a bit like touring the best restaurants in a new area - if you do them all in a row then you just lose some of the feeling of variety. I doubt Tracks will leave an impact on my soul to the level of Ikiru, but I still want to watch things like that often.
[1] for my personal taste
And as for absolute shit? Yeah, there's Andy Warhol on there.
- Everything Everywhere All at Once (2022): Newer than the cutoff date of the list.
- Your Name (2016): I'm not usually a fan of Anime but this newer film hit different.
- Sunshine (2007): This one I know I consider far better than most do, but it's one of my favorite variants on this kind of story for some reason. Also Cillian Murphy is my favorite actor in the modern era.
All 3 of these happen to be from after the original 2003 book from Steven Jay Schneider as well (though they aren't in the newer list either). I find he did a really good job of covering the classics, but I'm sure if I dug through my top 100 I'd hit something older I liked a lot that didn't make the cut. I'm just too lazy to :D
We created this as a demo for OpenSearch on AWS (we are AWS partners) but I really enjoyed making it and I thought it turned out to be really fun.
Took the same type of dataset at the OP (I think we have 80,000 posters) but we used AWS’ multimodal embedding model to enable semantic search across the posters, as well as the ability to do a similarity search based on the posters.
Eventually we will include AWS’ personalize as part of the service to show off some of the movie recommendation engine features they developed.
Not commercialized there’s definitely some bugs, we built it for fun
A little bit better way to navigate back and forth on results would be amazing.
But loving it, nice demo.
[1] https://www.metafilter.com/209750/Godfrey-Reggios-Koyaanisqa...
[2] https://www.damninteresting.com/the-baader-meinhof-phenomeno...
To those criticizing the UI: agreed. To those criticizing the movies: also agreed :D
As some commenters have already pointed out, it’s more of a tech demo than anything.
sarthaksoni•10h ago