pouët.net

gloom: Movie cameras shoot at 24 frames per second with a 180 degree shutter which gives you a 1/48th of a second exposure. This gives all films a constant amount of motion blur, you can't really change the amount of it unless you do it in post. Digital of course gives you the freedom to choose your own shutter speed (though most films still generally use the tried and tested exposures), but in my experience it's almost always the choice to have less motion blur in fast paces scenes since it tends to disorientate the viewers more due to each frame having less information about the movement.

Also regarding 24 Hz in general, there's a lot more to it than it just having associations. There are serious artistic reasons to sometimes choose a frame rate in the range of 24-30 due to how our visual systems work. And yes I do agree that you should also use motion blur (180 degrees is a good starting point, but you should experiment) especially in these cases, and especially if you have any faster paced action.

"due to how our visual systems work" = "because it looks better"

:)

Or could you elaborate a bit?

Yeah, big [citation needed] on the 24fps issue beyond "because we're used to it". Esp in light of current VR research that says kinda the opposite, and the often neglected fact that analog cinema not only suffers from 24Hz stuttering but also from double images due to projectors doing two exposures per image (24Hz flicker would be way too visible) - funnily film projectors let everything run "in the second frame", plus the nice conincidence that at least PAL CRTs do the same which further amplifies the viewing habits.

I do have a citation for it, but I'm trying to avoid steering this thread into that direction because I think it'll shift the focus too much away from the initial topic. I'll make another thread when I've written out my thoughts with some care. I mainly just wanted to throw it out there that refresh rate is a design choice just like everything else and that people should be more inclined to actually aim for something other than 60 Hz too. I'm biased towards the 24-30 range because I think for most purposes it looks better, and I've read material that has suggested physiological reasons why it might look better at least to some. But nothing is conclusive yet on this topic. I'd just like to see more experimentation on the subject and to see if people feel the same way as I do :)

@Pro: actually scanning real world camera motion is super cool. In your example the effect might almost be a bit over the top for my taste, but it still perfectly illustrates that the camera motion itself conveys a meaning. A little bit like in "clover field". For me the most impressive scene in debris is, when the "invisible guy holding the camcorder" moves a bit closer to get a better shot of the debris falling down (on him). For some reason this scene always reminds me of 9/11 footage. There are tons of other cool shots in debris, which might be a sign of fiver's intensive research.

Since I am on vacation and have plenty of time for reading, this is the perfect topic for more books. Keep posting. :-)

But as usual smash does have point: the majority of literature seems to be about framing people, which doesn't really apply to the stuff I wanna do.

Btw: I saw the "Revenant" on the plane, and I was totally blown away by the camera work. It's almost unbearable how good the camera is. So good in fact that it becomes distracting. They also seem to have a rule that maps "sickness of character" to "amount of camera roll shake". Any thoughts on rolling the camera and tilting the horizon?

People are just subjects :). A lot of the stuff about framing people can be applied to what ever your subject is. I don't like to just list books found by googling, but I've been reading Cinematography: Theory And Practice and found it to be the most applicable to demomaking of the ones I've looked into. Just consider skipping the segments that deal with pure technicalities like film exposure latitudes and whatnot.

Excellent, That's the one I've started with. Nice reading indeed.

Noby: I think it's the displaying frames twice which creates the film-like blurrines in 24-30 fps stuff. You can try it out with normal monitors, particularly CRT. Make your stuff move at 30 fps but with 60 fps monitor/screen refresh rate, and compare between displaying, (a) 1, 1, 2, 2, 3, 3, 4, 4, ... or (b) 1, BLACK, 2, BLACK, 3, BLACK, 4, BLACK... Most likely the motion will look much sharper with the black frames, although flashy or flickery, because of the black frames. I think this is called "black frame insertion".

I think that even though film looks nice in a film-like way, it kind of sucks, because motion is always blurry. If you want your demo to look like movies, then your motion is jerky. ;) People are just used to seeing that blurry see-everything-as-two motion.

Pixtur, you're right. We wanted to film a bigger statue, but that was the biggest we could find. So we tried to simulate filming a bigger statue, however it ended up in too much jitter. I think doing some kind of stabilizing / steadicam would improve the effect as well. That was just plain walking. We definitely want to pick that idea up again in the future when there is real time again....

yzi, yes- lots of the TVs sold as "120 Hz" are really 60, but with half of each frame time black. Cuts down on blur of the image across your retina as your eyes move to track the movements occurring in the video.

Seconded. This is the part that set "Debris" so much apart from everything else at the time (and almost, since). It's massively impressive on so many levels.

KB: is there really any more reason needed than "because we're used to it" though? There are more than enough articles on the subject of high-framerate cinema (like this one, focusing on Peter Jackson's use of 48 FPS in "The Hobbit"), and the consensus seems simply to be that people just find it odd/strange/"cheap looking". That reaction is directly derived from our own experiences. It'll probably change over time, but for most people I would argue that if people feel a multi-million dollar Hollywood production "looks like a cheap soap opera", then that's reason enough to at least consider 24 fps an artistic choice that's comfortable for a large group of people.

Extremely nice talk:
"Filming Giant Virtual Vehicles: Procedural Cinematography in Homeworld Shipbreakers"
They build tools that make the game engine find good camera angles in dynamic scenes. Some very nice ideas, including some that have been discussed here. Especially helpful because they provide a coder-perspective and analyze the image beyond less-than-helpful "it looks good because it's balanced, see?" comments.

(This image is not from the talk, but shows that they know what they're doing:)

Gloom: Not disagreeing with you in the slightest, it's just that Noby made it sound like 24fps was _objectively better_ which I'd love to read citation on because I refuse to believe that. :)

Funny that you bring up the Hobbit because for me watching at least the first movie in 48 fps resulted in a few realizations:
- Yes, at first, the 48 fps looked completely wrong and all the motion felt too fast; and at the same time I knew that it was right which confused my mind even further - but after a few minutes I got pretty used to it. I can second the notion that it made the sets look cheap though; probably because you actually see more of them and the mind has less blanks to fill in.
- In fact, when it came to the action scenes I quickly realized 48 fps weren't even enough, at least for 3D. I was still perfectly able to discern single frames, and tracking eg. particles with my eyes was way better than in 24 but still not where it should be. This pretty much coincides with VR research that currently pins the minimum fps for actually perceiving motion (rather than "knowing it's supposed to be moving") at 90.
- The fun started when Bilbo put on the ring and the whole image went into Ring-O-Vision. With all the weird wobbles and feedback effects my demoscene trained eyes actually started to relax and my mind felt more comfortable again. Which probably says a lot more about me than i'd like to :), but also shows how much viewing habits actually come into play there.

After I returned from Revision and watched something on Netflix, I got the impression that everything is incredibly stuttery (which it is). Guess I quickly got used to high fps that quickly. Also, remember the debate about HD TV looking shit because you can see "too much"? That was actually brought up as an argument against buying a HD TV set. People are strange!

I feel the same way about > 24 FPS in movies.
However: I have the impression that it's the complete opposite with games and demos, in my view they look so much better in decent 60 FPS compared to 24/30 FPS.
An interesting example of combining high framerate and subtle motion blur is reel by mindforce.

jco: The Netflix situation is probably amplified by the fact that proper 24p in affordable TV sets and all desktop/mobile screens is still a myth, and most resample to 60 resulting in the familiar 3:2 stutter pattern. After 4 days of Revision where I make the fuck sure that all refresh rates arrive on the big screen exactly as they're supposed to look that's a big step back, yeah :D

Anyhoo, back to the actual topic: Watch that talk Cupe posted, it's great. It reminds me of a similar system I once coded for a racing game that sadly got canned before release; but it had replays and an attract mode where it played for itself, and we needed dynamic looking cameras. After a week or so I arrived at the following which looked pretty cool (really, too bad I don't have anything to show for it :( ):
- Artists could place "TV" cameras in the level, either static or with a position animation (they later figured they can place actual tv cam models into the scene there for added realism)
- The camera system checked visibility to the vehicles, determined a set of cars that it wanted to show and tried to keep their bounding boxes (scaled by 0.5 *fakefake* :)) in the safe area in the middle of the frame by panning/pitching and zooming
- I "simulated" camera equipment and operator by converting the cam->target vector into angles, low pass filtering the hell out of them (which also automatically delays) and then adding min/max/speed constraints to the rotation as well as the zoom.
- Now for the fun part: Every camera calculated the "interestingness" of what's going on in front of it (mostly moving object count, really) including a "omg noooo, don't use me" flag (nothing too see anymore / cars too close or fast). Inactive cams were ticked round-robin every few frames to update their scores
- A "director" periodically checked if either the current cam set its "nooo" flag or another cam got way more "interesting" and decided whether to cut to a new camera or not. Add generous min/max times for switching and penalties for bad/recent cameras and you arrive at something that looks pretty much like a TV broadcast of a race. :)

Funny, I coded something a bit similar too: multiplayer car racing game, we wanted to have replays and something to show to players in the waiting room. Artists defined some cameras along the track, and the system picked them based on interest, which included the number of cars, how long the camera had been used, whether there seemed to be struggle in the race, etc. Nothing fancy but the results were satisfying.

There, fixed :)

I think that part of the problem with higher frame rate video tends to be related to the viewer expectations: usually TV-style video imagery had higher frame rates, so I personally, when seeing high frame rates in cinema, switch into "TV watching mode". I am guessing I am not alone at that.

Anyway, thank you very much for very interesting thread. I wish there were more threads like this on Pouet.

So, demos must have jerky motion like movies, because movies are made in Hollywood by Big Boys? There's something wrong with that idea. I wonder if gamers also want blurred motion, because then it looks like movies.

yzi: no one is saying that demos should have or has to have 24-30 FPS. What people are saying is that some times it can make sense and look better. So don't take the 60 FPS granted but explore.

In some things I'v used 30 FPS animation speed with variable rendering speed which works ok. However it is not so much of an artistic choice but just a mechanism to cope situations where it can't be guaranteed that the frame rate is stable (but most likely still over 30 FPS).

Actually I tried to challenge the idea of trying to look like movies, if it's done just in order to look like movies. I also don't like orchestral music in demos or games, because usually the only real motivation is to try and sound "like movies" and because there's an assumption that "people" are going to vote for that sort of stuff. Which they do because they think others like that sort of stuff so it's fashionable and legitimate. Yeah yeah you'll get votes. But it's still crap. I've always hated the jerky motion in movies. It's a bug in the art form. The emperor has no clothes.

Regarding 24 FPS i read somewhere (can't remember where, sorry) that lower frame rates are recognised by the brain as not real, and kicks it into a kind of dreamy mode where it's very forgiving about what it sees. At higher frame rates the brain thinks the footage is real enough and stays in critical mode, which makes the usual unrealistic effects and props stand out and look weird. It's unfortunate that The Hobbit became the reference for how high frame rates look, because it looks like a cheap video game even at 24 FPS. Would be really interesting to see more natural footage at 48 FPS.

absence: So you're saying it's kind of like an uncanny valley for framerates? If so, that would make sort of sense.

dudes, so this is the real reason why Amiga demos are so much better! ;-P