Distribution of colors in movie posters between 1914 and 2012 | Hacker News Reader

Distribution of colors in movie posters between 1914 and 2012

vijayp.ca

244 pointsfallenhitokiri84 comments

I am awe-struck with what people with a passion manage to get done "in a couple weeks" just for fun.

I can't imagine that I could have hired someone to do the same analysis, research, programming, scripting, graphics creation, and web site layout for anything less than tens of thousands of dollars. (Or is my intuition wrong about that?)

Well, a week is about $2500, so I guess it'd take around $5k.

There are many people who would call $2500 their two day rate. So, $10k+ doesn't seem that far fetched for a project of this complexity that delivers business value of some sort.

People who charge $150+/hour are rarely working 8 (billable) hours a day.

I think that's far. I mostly posted because I don't want people to believe that $5k is the most you could get for this project.

There are infinite ways you could add a ton of value to this project that would be worth more than $5k. Automating the data retrieval and plugging it into mechanical turk for verification, being able to critique or comment on the deeper meaning of the change in colors. Or even online marketing to get the backlinks for a business could potentially be worth $10k.

I'm sorry, isn't this just a histogram of all the posters for that year? Am I missing something (e.g. a page I didn't see)?

It doesn't take that long to make a histogram of a "movie posters year X" Google image search...

No, you're right.

But from your comment "it doesn't take that long.." its clear to me we're thinking about this in two different ways.

You're saying, "Here's how long it would take me..." hourly * estimate.

I'm saying, "Here's how to add more business value..." business_value * my_cut_to_decimal

If you can, perhaps by not even programming, make business_value a large enough number then (business_value * my_cut_to_decimal) > (hourly * estimate).

I'd go one step further and argue that the higher price is a win for both parties. You get paid more, and the company has just discovered a slot that they can slide money into and get more money back out off.

Oh, sure, I agree with that. It's just that the OP was talking about what it would take for him to hire someone to do this, so I went with that.

I don't mean to be facetious, but there are also developers, and I mean that in the loosest sense of the word, that call $100 their daily rate. My guess is that both parties need to maximise the value/profit ration by negotiating. $10k sounds right as we are delving into hypotheticals.

Nowhere does he say that he spent all of the 2 weeks working on this entirely. My guess would be that he did a 2-day sprint to hack the prototype and a bit more to polish it across the 2 weeks. Point is - he would not bill 2 working weeks were it a contracting gig. But then again, where'd you get such a motivated contractor (cf. working on own project)?

If he did it in a 2-day sprint, that makes it even more amazing -- which was my point. It seems unlikely to me that I could find someone to do the same quality of work as a contracting gig (whether it took 2 days or 2 months) and charge less than tens of thousands of dollars.

That's why people spouting "open-source is always less quality" and "without paying artists there would be no culture" are so dead wrong. People who feel the urge to create and do, will do so. And to me, things that were created for one's own enjoyment are pretty much always better than things created to be monetised.

Right. Can you imagine a film with the quality of Toy Story being produced just for fun? Not a chance. ThAt film took thousands of man-years to make. The whole open source everything movement is just communist nonsense that typically comes from people who have never had to make a payroll. Rails for example was the open source result of a profit-creating piece of software: Basecamp. The profits enabled the open source not the other way around. When I can feed my family, I'll contribute to open source, but who pays open source developers? Usually they're being subsidized by closed source work. For example, in my current day job, I'm building software for a startup. That pays the bills so I can afford to dabble in open source for fun. The idea if give everything away is just silly. Who is going to pay for the computer you're usin to develop on? Who pays the electric bills? Certainly not te little open source fairy godmother.

Many great films are made on a fraction of the budget required to produce something like Toy Story. And anyway I don't think it's hard to imagine Pixar artists saying they were more motivated by the pure joy of creating something than by their paychecks. There are bad movies that are created specifically to be monetized -- that people need money is beside the point that often those who create the really good stuff are probably doing it more for fun than for money. I think I see what you're saying, but it seems like an overreaction to what didn't strike me as a controversial comment by the parent.

If that were true, there would be other movies like Toy Story, but "open".

Well, it isn't quite as long, but http://www.bigbuckbunny.org/ was pretty good in terms of quality.

That project had a budget of about EUR150k and artists were paid.

But is Toy Story really the best of our culture? Just because it was expensive and hard doesn't necessarily mean it was te est (for the record, I have loved almost everything Pixar has done).

Asked another way: if the Pixars of the world were to disappear, would it be a net loss for our society? I'm not sure the answer to that is "yes".

It's hard to imagine a film the quality of Toy Story being made for fun, but I frankly find it just as hard to imagine a world powered completely by open source software, from the OS on our computer through the network stack, to the databases, web servers, and programming languages that run them all. Yet we live in that world, and the closed-source software that you assert is the only way to make a profit is somehow being written as a thin layer of film atop an ocean of free, open source software.

I'm not sure how it has escaped the world's attention that you didn't approve these thousands of man-years of work, but it's clearly an oversight that must be corrected immediately. Let's have you tell everyone it doesn't make sense and get this communist nonsense shut down immediately—it's not philosophically viable!

I am willing to bet that the same person couldn't even get as far in 2 days/weeks if you were hiring him for this project. Contracting comes with significant additional overhead that quickly eats up the time.

You might find the results you seek if you offered to pay someone for two days to work on whatever they wanted. However, even then, you're apt to not get anything of quality back. I'm sure the person who has created this work has thrown away many more projects.

I immediately starting scanning the colors to see if this blog post from a couple of years ago was vindicated about all movies being teal and orange:

http://theabyssgazes.blogspot.com/2010/03/teal-and-orange-ho...

It is.

It definitely does seam to go teal heavy from the 80's onward.

I don't think "vindicated" is quite the word - as far as I know, the author hasn't taken any significant flak for it, nor have I heard much about it beyond "Huh, yeah, they do look like that now." I would just go with "this analysis is in agreement with this other one."

In the associated blog post he says:

  Methodology:
  I downloaded ~ 35k thumbnailed-size images
  (yay wget -- “The Social Network” inspired
  me to not use curl)

Could some who's seen the film enlighten me on why wget is better than curl?

Thanks.

I don't remember any discussion of wget vs. curl in the movie and couldn't find anything specific to the movie about that on Google; I'm guessing that the the OP just decided to use wget instead of curl because that was what they were using in the movie. (If I'm wrong, somebody please correct me.)

Here is a comparison of wget and curl that was posted on Hacker News a little over two years ago: http://news.ycombinator.com/item?id=1241479

I think this is mostly one line in which Mark Zuckerberg says something along the lines of "time for a little wget magic". Certainly not a major plot point.

Here's what I think is the relevant scene on youtube: http://www.youtube.com/watch?v=odOzMz-fOOw

Wow, I was surprised on how realistic that was. Maybe I should watch the movie after all.

It's actually quite a good film if you ignore the character assassination of Zuckerberg (not that he really needs defending, just it seems a little cheap).

I don't think it's fair to call it character-assassination. He didn't paint Zuck with a pretty brush, but I'm not sure any first-tier founders made it to where they are without delivering a few cheap blows themselves. There's definitely some sense of admiration for "Zuck the Doer", not "Zuck the Dreamer".

Especially the empty search part. In my experience, the three search terms most likely to return all results are:

  1) %%% (many sites limit the minimum search term length to three so a single % will not do the trick).
  2) '   ' (trhee spaces)
  3) '' (empty string)

Edit: formatting

That's about as much tech talk as there is in the whole movie - you may be disappointed. Bit of a storm in a tea cup I thought.

HTTPie[1] is better than both. ;)

1. https://github.com/jkbr/httpie

(I am in no way affiliated with this project. I'm just a happy user.)

But wget and curl aren't for humans, whereas HTTPie is. I've done plenty of `wget -m -np`s (and `puf -r` when I've had to saturate my Internet pipe...) which isn't what HTTPie does.

Ok, httpie is prettier and has some syntax shortcuts, but it doesn't even do a fraction of what curl or wget does. No recursive downloads, no simulating diferent HTTP protocol version, no SSL options, no even retry options! #fail

(I am in no way an httpie hater.)

While beautiful, only a seriously limited amount of movies was analysed for this and the selection seems weird. The year 2000 for example only has 48 movies in this. I am not sure you can interpret anything from it.

Agree.

I'm surprised at the amount of comments here that reflect this data as if it's gospel and whether it's fair to interpret anything from it at all. We can't even vet and recreate the methodology of, as you said "limited movies", and see if it's correct for what it does say.

Not to mention the fact that prior to 1929 there is not more than 9 posters of data being analyzed.

Printing has changed also. Number of color, type of press, computerization all would contribute to difference in what designers would do (the analysis did mention this) but as mentioned the data was limited.

I'd be interested seeing someone (not fit enough in this topics to do it myself) applying color psychology and mapping it to the years / events / industry.

Could explain the shift from warmer colors to more technical ones (just speculating)

Maybe they couldn't use some colors in older times due to technological limitations of chemical/print process?

Looks like "Peak Oil" back in 1977 made people feel blue or resulted in different chemicals being used in films. The blue band is much wider for that year all of a sudden.

Or it was Star Wars. ;-)

Plausible; as it happens again in 1980 (Empire Strikes Back)

Interesting to watch the color gamut open up as color printing technologies improve over time. Blue/Cyan is the hardest pigment to work with, and you can see the blue-range grow over time as the technology to support it improves.

I would say the color gamut has narrowed recently.

Really good blog post on Hollywood going mad with teal and orange: http://theabyssgazes.blogspot.ie/2010/03/teal-and-orange-hol...

There's plenty of blue in 1919-21, but it disappears for the following decades. Either the data is seriously biased or something happened here.

The eventual settling on teal and orange is interesting. I'm curious what the feedback loop was that told the marketing department "this is what will make more money". It seems such a subtle thing, and I would think red (as an alarming color) or green (as the color we see best) would end up figuring more prominently.

I take it you have not seen this yet (linked elsewhere in the thread, but it's been on HN a few times): http://theabyssgazes.blogspot.com/2010/03/teal-and-orange-ho...

I'm sure you've seen this[1] before, but the answer is probably more of a feedback loop. The big-money movies could afford to do more color alterations that caused movies to appear to "pop", getting audiences used to those colors, forcing less-expensive movies to do the same as the technology to do so became cheaper.

1. http://theabyssgazes.blogspot.com/2010/03/teal-and-orange-ho...

It would be interesting to combine this information with boxofficemojo and adjust the distribution of colors based on box office revenue assuming that box office revenue correlates to mindshare, you would have a closer map to what people 'thought' a movie poster would look like in that year.

More red/yellow before, more blue/violet in recent history. Cool data set, but the question is "so what?".

Well, it's interesting to see that Teal and Orange filters are used just as prominently in promotional material as in the films themselves. It's a terrible trend and annoying as hell. See here for more:

http://theabyssgazes.blogspot.com.au/2010/03/teal-and-orange...

That was a very interesting link, thanks a lot.

I was looking out for the Teal/Orange effect and was interested to see that it started in posters way back in the 80s already, 20 years before digital film color grading.

That's an interesting observation. I wonder if that's a more general kind of lag that can be found with other effects, with things becoming feasible in still design work some years before the analogous thing becomes feasible in film/video.

Huh. I'm fairly color-blind, so I can't see the orange (or teal) tones that he's complaining about. However, I have noticed that more films lately seem to have more detail, and everything just "looks better" to me.

I've read that some amount of color-blindness is context sensitive (e.g. http://www.designmatrix.com/pl/cyberpl/cic.html); I bet that the increased contrast in these films is part of what's making them so much more visually appealing to me.

I wonder what happened in 1921. It was out of control!

The most interesting part of 1921 is that the original(?) Brewster's Millions was released. In the original movie, the protagonist had to spend $2mil in a year and remain unmarried to inherit the full $10mil. By 1985, the remake dropped the marriage requirement (honestly, being a single rich guy in the 1980s wasn't too bad) and upped the spending requirement to $30 mil in 30 days to inherit $300 mil.

My, how times change. I wonder what Montgomery Brewster will have to do in 2049 to earn his inheritance.

Shows an entire movie in a single image.

Can anyone explain how this works?

I don't know for sure, but judging from the sequence shots, looks like it scans a pixel column depending corresponding to the part of the movie. If it's scanning a frame halfway of the movie, it will scan the pixel column in the middle of the frame.

Basically, it's the equivalent of putting a monitor face down on a flatbed scanner, and scanning while the movie plays.

Or taking a photo of the screen with a very slow CMOS (camera phone) sensor.

Wasn't this here before? It's still cool though :)

I am not sure, I did not know it and just found it on a blog and I did not see any timestamp

I thought that too, but then after reading it I remembered a previous post about the distribution of colours within actual movies, not movie posters. So perhaps that is what you remember too?

Interesting that there are no black columns ie no colors which consistently wouldn't be used. Green is represented very little though

Fascinating viewing. Wonder what changed us from being all sunshine and light to being all doom and gloom. I blame the Dark Knight.

Ha, you can see the teal trend getting stronger in the 2000s. I wonder why the colorfulness of 1919-1921 was so quickly reverted?

I'm curious about the distribution of ink costs (by color) during the same timespan - just how close that correlation is.

This is the question I was going to ask. I've heard that the reason most superheroe's costumes are primary colors is because it was less expensive to print them for the comics. I wonder if a similar effect is happening here?

+1, I think this must've played a large role in color selection, especially considering the large print runs these posters had to have.

If they were scanned in recently, doesn't this just prove that print materials do indeed yellow over time?

But why a stacked barplot? This way you can easily see the dynamics in reds but the apparent amount of blue is quite deceiving. IMO a better idea is fixed hue grid and modulated saturation for the counts.

Looks like Star Wars posters messed up 1977!?

http://www.vijayp.ca/movies/index.html#1977

Hmm again in 1980:

http://www.vijayp.ca/movies/index.html#1980

Maybe Lucas really was a visionary!

Before clicking I thought it was Jaws, but wrong year.

I wonder how much this adjusts for natural fading in prints.

I find the near constant use of red interesting. Given all we've been told of red and marketing (think cereal boxes), I figured it would grown.

OK, so who's going to do the correlation with en vogue colors from the same years (fashion, paint, cars, etc.)?

but hey, isn't there any color corruption during scanning/digitalisation of those posters?

Cool! Color combinations form sinusoid throughout the history.

What happened in 1977?

Star Wars?

And Close Encounters, and Saturday Night Fever, which both have blue-ish posters here: http://www.imdb.com/list/_EcfN1DMzJQ/

Also The Deep, which isn't on that list but did pretty well so likely had a lot of posters. All of which would be mostly ocean. http://www.imdb.com/title/tt0075925/

Wait, so they all fall on the visible spectrum?

I don't see The Matrix. I wonder if it was left out because it's on outlier (too green/yellow).

Reply on news.ycombinator.com