Go to bottom

How To Camera

category: gfx [glöplog]
That's an interesting topic for me because my level is unfortunately at zero.. (object in center, focus 99% of time coding the effect and forget to care about camera, never wrote some non linear paths for camera ever)
added on the 2016-04-03 14:53:35 by Optimus Optimus
I've only had time to read the first post so far so sorry if these have been already said.

use longer lenses (that is, turn down your fov to 15 or 10 or 5 degrees)

Not all the time, but I guess a simple rule here is: the larger things you are looking at, the lower FoV to use, as it helps large things look large.

Second: linear interpolation looks okay, everything spline-y is questionable. Mostly use linear movements (maybe with variable speed, but keep cam movement on a line) and cut when you need to go to a new shot. No hermite-flying-around.

This one I don't quite agree with. Linear is good if you only want to move between 2 points at a constant speed, but it will break with anything more (obviously).

For me bezier became the jack-of-all-trades because you have precise control over timing and speed between two keys, and it's pretty much the only way that can work with curves using multiple keys.

Example: you are pulling the camera towards your subject and want it to stop. With bezier you can make it progressively slow down and stop without any sudden "clicks". (Sidenote: use bezier for fades too, much better than linear.)

Third: look how camera is done in actual movies and try to simulate that. Put your virtual cam on a crane or a dolly and control its parameters.
Fifth: Improbably fast camera movements are forbidden, especially if things are large.

These two are important to give the camera weight, as you put it. Don't do anything with your camera that you wouldn't be able to do with a real camera. Some AAA Hollywood movies tend to forget this too, making million dollar CG shots look like unbelievable crap.
added on the 2016-04-03 15:01:46 by zoom zoom
There's also this :)

BB Image
added on the 2016-04-03 18:58:06 by gloom gloom
this video is quite useful too:
Instead of bsplines try to use 3 different sums of multiple sinosoids.

Wolfram mathworld " like - curves" tells You how to use 3 stepping functions with multiple sines. One curve foreach dimension.

Automatically integrateable and very easy to compress.

Alternatively use signed distance Field functions and have The camera be like a boid that moves through a magnetic Field of extruded normals with a target and some momentum.

Both methods are Close to an audio codec or to swam Behavior. Both common things already in demos.
added on the 2016-04-03 19:43:17 by ollj ollj
The very modest thing I learned coding Nautile, is that you can get cool effects when the camera is slower than what you are filming: https://youtu.be/JEMIeSU1r7o?t=91
added on the 2016-04-03 21:43:52 by Soundy Soundy
[...]I don't think randomly changing zoom and lighting is a good idea. In the real world, zoom lenses are expensive and heavy and have worse optical properties compared to fixed lenses, so if they don't need the variable zoom (and if there is a fixed lens available for that level of zoom), a director will choose a fixed lens.[...]If there is no "motivation" for the light (like a lamp) that could plausibly move, we add confusion to the scene which is not what we want most of the time.

Naaah, you should not ad any kind of animation without a purpose. E.g., if you zoom on something you also imply that you´re focusing on it. If yu´re moving the camera position you imply that you´re sort of passing it (being more a spectator than part of the action). Circling around something points out that you´re examining it from all sides etc.

Same goes for lighting, a moving light source is usually a bad position (unless you want to depict a flashlight or a moving car), but sunrise/sundown, clouds/smoke/mist, closing doors, moving sunshades etc. also create a more or less slowly varying lighting while providing a deeper impression of what´s going on.

Also, I don´t think one should just copy what movies do. they´re sort of stuck to old restrictions while we are completely unbound from any real-world limitation, and they also focus on pleasing the general audience. It´s appropriate to imitate that for creating a cinematic look, but not useful as a general advice.

Generally, your suggestions involve using time to make the shot interesting, but in line with the title of the "every frame painting" video series mentioned above I think: A shot should work without movement. Or at least the beginning and ending of the shot should.

That´s correct, but showing the same shot for an extended time period without any alteration is pretty boring, no matter how good the single frame is.

Also, the current 3D hype is a quite two-sided beast: On the one hand, a camera movement which is bad in 2D will look even worse in 3D / VR. On the other hand there are a lot of interesting image compositions which look awesome in 2D, but hardly work in 3D (and thus are pretty rare in 3D focused movies nowadays)
added on the 2016-04-03 23:29:52 by T$ T$
Imho general rules are really the worst approach to this topic. Camerawork surves a purpose. It's a perception altering motor by itself. The very question should be: what am i trying to convey here? Then choose perspective accordingly.
added on the 2016-04-03 23:38:02 by rp^frstl rp^frstl
true that, but general rules are the right approach 'when you don't have the eye for it'. and even when you have the eye for it, knowing what's considered good or bad helps in making the right decision.
i mean, i can imagine Anton Corbijn going 'for this next shot you'll all pose in a Fibonacci-set up on the photo' to U2... it just happens to be the nicest looking picture of the shoot. they use it as an album cover and every composition analyst goes wild over how brilliant that was... ;)
What Maali said. There is some proverb to the effect of "you must first know the rules before you are allowed to break them".

There are some examples of demos with such nauseating camerawork which could certainly benefit from following some general guidelines.
added on the 2016-04-04 01:18:36 by drift drift
I actually think Maali point (and similar) is too simple or even ignorant.

It is true that all rules originally comes from what the audience seem to accept as good / impressive / pleasing / intriguing, not the other way around. So you can even say that in principle there are no rules. However, there are certainly 'things' that are _known_ to stimulate a positive reaction in the audience, as well as 'things' that stimulate a negative reaction.
And those are the things I believe we are discussing here.

Some people obviously have more 'eye for it' than others, but I would bet not everything comes from their own experience and observations. It's like with every other discipline: you _need_ to learn from others to progress, one brain seem to be not enough (even Maali's).

So, please guys, continue with the 'rules'! For example those 'Every frame a painting' series are a real eye opener for me. Thanks!
added on the 2016-04-04 01:59:28 by tomkh tomkh
and thus spoke Captain Autism...
Maali: that's discrimination for autistic people right there! Nah, just kidding. Also sorry by pointing finger on you, but you know you deserve from time to time, do you?;) Let's just not put this discussion into trash, as usual, please :(
added on the 2016-04-04 02:16:53 by tomkh tomkh
What Maali, drift and tomkh said. It's better to follow "rules" (even if you don't really need to) than forgoing them all and trying to be completely original. That's how almost all great artists start anyway, by following the common, well understood and known patterns and eventually finding their voice and style among them. Of course if you're absolutely brilliant go ahead and do your own thing from the get to, but remember that statistically you're most likely very mediocre, just like everyone else.

It really is important just to know the rules and how and when to apply them. Just like in programming patterns, you should not only know what they are, but which apply to which problems. The goal isn't to make demoscene look like films. The visual language of cinematography goes much further than that. Borrowing from films doesn't imply a "cinematic" look. Similarly knowing music theory doesn't mean you now need to strictly compose classical music or pop tunes.

I also completely agree on fibonacci point. These things are in my opinion almost exclusively intended for visual and aesthetical analysis afterwards. Nobody lays out a complex set of golden ratios while taking a photo, and for articles to imply that is so misguiding. Rules of thirds is very useful though, but even with that you need to know when to break out of it as to not make your compositions look forced.

I guess I could also say a few words about Hydrokinetics as it was brought up (appreciate it :)). The goal with the intro was to imply a relatively realistic scene even with the crude geometry, and one way to approach it was to also use relatively realistic camerawork. I basically decided on one still shot that just worked and decided to build everything around that. The rest comes down to two approximations of a "crane" shot, one establishing shot (used twice), and a close-up. Nothing is cut directly to the music, but to movement and action, while the action is set to mirror that the music does, which is what I found to be a very balanced approach. The whole thing would've been a disaster had I moved the camera all around with cuts to the bass drum or something. One more close-up or another revealing shot a couple points wouldn't have hurt and I might've had the bytes to spare, but eh, what I ended up with worked for me.
added on the 2016-04-04 02:29:14 by noby noby
At the risk of stating the obvious, I don't think there is a set of ThingsToDo for your demo to look good.

In that regard, saying in the blind that you shouldn't center your frame for example, seems pointless to me. There's the rule of thirds, there's the golden ratio, there's centered frame. Some films use the latter to achieve memorable images (see Wes Anderson, as already pointed out). There are only a library of rules that will, in certain cases, convey a certain meaning. Read classics, know your basics, read articles that analyze film cinematography, watch films with a critical eye, read more, get out with your camera, try things out, keep reading and watching films.

One thing that photography (amateur, lazy one at that) has told me is to love fixed lenses. If you don't already have experience with real world lenses, I think cupe's advice to experiment with zoom is a good one. That being said, fixed lenses have the great teaching value of forcing you to learn to work within constrains and learn what you can and can't do with a given size, instead of blindly adjusting your zoom until it fits. Same goes with aperture and depth of field.

On film making vs demo making...
One the one hand, the rules used in film making come from dozens and dozens of years of experience, and we could do worse than following them. Moreover, we have been used to the film look and its grammar, so anything outside of it will require additional care to work well with the audience.
On the other hand, film rules are dictated in part by technical or practical limitations. As we are free of these limitations, we should definitely explore beyond them. See for example how the rise of smaller and lighter cameras like the Canon 5D or RED Epic based rigs yielded new types of shots from film makers.

Apply that to 24 vs 60fps: 60fps looks less cinematic (see also the criticism of The Hobbit for its use of 48fps), but also has this "more real than real" smooth look, which could be exploited for better impact. Has anyone tried using variable framerate to convey meaning? (one scene at 24fps; another one at 60fps) A similar example I can think of would be Tron Legacy, which used 2D for its real world scenes, and 3D for its grid realm scenes.

Those were my 2c, keep the change.
added on the 2016-04-04 08:50:33 by Zavie Zavie
I still don't believe that the 24FPS looks better, because XYZ.

It doesn't look "better" per say, but it does have a look that's associated with movies and quality. That's just how we've been conditioned after decades of movies on the big screen.
added on the 2016-04-04 10:16:21 by gloom gloom
I didn't read through all of this but it's def on my list for tonight. However, i can recommend quickly:

Camera shakes from Dead Pool: http://premierepro.net/editing/deadpool-handheld-camera-presets/

Doing camera movement in the traditional wiggle style doesn't work, using noise is better but makes it harder to interpolate in a modern style keyframe editor.

We (nuance) do it very wrong in all demos, however we spent actually almost a day on that camera for this scene:


What cosmic and i did: We went outside with a DSLR and actually walked around a statue filming it. We then went back, tracked that camera with the Nuke Cameratracker in AE and exported the camera into our engine. I'm still quite happy with the result, however the statue we filmed was too small, having a bigger statue would have helped to improve the effect. What we also did was to slightly offset the actual focus change to simulate some kind of Auto Focus delay.

A similar but obviously procedural camera was applied in debris (that cube scene where they go down the parking garage, beeing observed by a shaky camera)

That tip from Cupe to actually simulate the rig the camera is attached to is definitely the way to go!
added on the 2016-04-04 10:25:30 by pro pro
Here are some of the highly praised books on Cinematography by people in the industry that I've found through some research (these seem to be the ones that most people recommend):

The Five C's of Cinematography
Matters of Light & Depth
Master Shots Vol 1,2,3
Film Directing Shot by Shot
Cinematic Storytelling
Set Lighting Technician's Handbook
In the Blink of an Eye
Lighting for Cinematography: A Practical Guide to the Art and Craft of Lighting for the Moving Image

Painting with Light
Cinematography: Theory and Practice
The Filmmaker's Eye
The Visual Story

American Cinematographer Magazine

added on the 2016-04-04 10:36:58 by therue therue
Sorry - with general rules i didn't mean the existing rules and boundaries. Rather the generalization of approach. Creating tempo and stress demands different measures than tension or calmness.

Whatever you're trying to convey thus has an impact on every scene. If you aim for fast cuts it makes sense to use DOF to create focus and set lights accordingly. Whereas slower scenes can reveal more subtle details.

This is the kind of stuff that needs planning ahead and then choose how to apply which technical solution.
added on the 2016-04-04 11:36:45 by rp^frstl rp^frstl
BB Image
added on the 2016-04-04 12:33:36 by cupe cupe
I still don't believe that the 24FPS looks better, because XYZ.

It doesn't look "better" per say, but it does have a look that's associated with movies and quality. That's just how we've been conditioned after decades of movies on the big screen.

But unless the movements are very slow, you'll need some motion blur to make it work, otherwise the stuttering will be too distracting.
added on the 2016-04-04 12:57:50 by zoom zoom
On the broader topic of applying film techniques to CG:
added on the 2016-04-04 14:46:19 by Zavie Zavie
Zoom: possibly, but that's also why movies do add motion blur to fast scenes, and why certain scenes are shot in very predictable ways (to avoid those types of things) which again reinforce the "look" of movies.
added on the 2016-04-04 16:32:05 by gloom gloom


Go to top