Russell Scott

clearly... :)

*stokes fire* The irony is that Yogi revolutionised the animation industry by introducing a cheaper and more production efficient techniques at the expense of established animation rules. Animation purists were outraged at this assault of standards and looked down upon such techniques as a cheapening of their craft. The public didn't care of course, caring only for expressive, interesting content... Not that this has any relation to any dicussions that happens so often here ... :D

To be clear - you generally have the choice between no decay, linear decay and inverse square decay. Having a light that decays physically (i.e. inverse square) is not the same as photometric (although of course photometric lights will all decay physically).

On big features you'd have specialist lighters. They would light an animated feature like you'd light a live action set. You have lights that you place in 3d space, give them an intensity, a colour. Most 3d systems have photometric lighting setups so conceptually its pretty similar. In some respects its easier than live action, because you have no physical limitation on light placement (e.g. you can light a scene with a light in front of the camera but hide it from the camera so the light itself is invisible). In other respects it is much harder, the 'photometric' nature of the lighting is based entirely on approximations and that leads to much wrangling to get the desired result. You also don't get instant feedback, so when you move lights around you don't see how that light has affected the scene, you have to crunch the numbers on the render machine to get the result (which could take hours).

Posting in a public forum, with your whole name, about a film you're working on and then casting aspersions on the directors abilities was never going to end well. It wouldn't take 2 seconds to find the director involved - airing grievances in this manner is not OK. With friends like these...

Hugo Cabret

I don't think its worth converging the shots, simply because the extreme nature of the stereo in the shot means it won't help. In this case it isn't just the foreground elements that are the problem. Often there are objects captured in one eye and not the other, or captured with very different perspective vanishing points; converging the stereo differently won't help. The general answer is yes, you are forced to never include objects coming out of the screen. You can do it occasionally with quickly moving objects because the mechanism with which the eye tracks objects is slightly different at pace (poor explanation, sorry!). You can also get away with objects in front of the screen if your stereo is minimal, but its best to avoid cutting the screen (left or right sides) an even then, generally inadvisable.

Its not really going to help that much, adjusting convergence only changes where the depth starts and stops, not now much depth you've locked into the scene. In this case the problem is the depth and there isn't really a 'post' solution. Having said that, you very rarely want objects coming out of the screen. Generally you want to be staring 'into a window'. For various reasons images coming out of the screen are more taxing than those going back. Pushing convergence too much also has other problems especially when you have strong perspective cues (and you convergence doesn't match those cues). you can see that when she opens the drawer at 0.12. The perspective strongly conflicts with the convergence you've chosen. (best shown using the 'parallel' mode on youtube) At a glance I'd say your stereo is about a minimum of 10x too strong and for some of the close scenes about 50x. One trick you can use to get away with having too much stereo is to film with a close background. So that even though the depth is too strong, the amount of depth visible in the scene is limited (this is probably why when you 'blew up' the image it worked. Generally digitally zooming in stereo simply increases the overall stereo)...

Hi Mike, what is your intended transmission format? i.e. do you hope to put this on a TV or screen? If so, your camera separation is too much. A DVX rig is about 13cm side by side (?) which means for semi comfortable stereo on a tv you'll need the zero parallax at ~4m away(and nothing in front of that!). For close character shots with a dvx you'll definitely need a mirror rig (you can get suitable ones for the dvx for about 3k i think). You've also chosen a lot of shots with strong foreground elements, more specifically, elements coming out of the screen. You want to minimise this as much as possible, especially with strong stereo. The example I'll point to is at 2:40. The foreground element is both way too strong and acts as a distraction, your eyes are drawn to it rather than the action, also as you pan the camera left and right, the foreground element cuts the screen, which -particularly for foreground objects - is not good. At 2:51 you have an element in one eye and not in the other, this = pain for the view. You also need to watch your reflections - in a couple of the shots there are stray reflections in one eye and not the other. hope that helps in some way...

Hi Hal, I've done this a few times. Its particularly effective doing this with a vertigo shot. Much harder to do with physical camera of course and if you don't want to noticeably stretch the depth you have to have a distinct kind of shot to use this on. e.g. going from a long shot to a close in shot... "flying over a bunch of bunch of buildings then flying down to see our hero walking along the street.. camera ends up framing his face". It is possible to smoothly transition from showing the depth of the buildings to the depth for a closeup. but generally you wouldn't want to do it...

just to expand on this. The ideal i/o isn't 64mm for the camera - you want an apparent 64mm separation on the projected image. The bigger the screen and closer the audience sits to that screen, the closer your cameras will need to be (also dependent on what your camera is looking at and where that object is in relation to the camera). The i/o of the cameras should be chosen with some reference to the medium its projected in, eg a 64mm camera separation for TV might be fine but not for cinema. This is one of the main reasons people get pain in cinemas - the i/o that is chosen is far too much. and as a separate point I would also argue against trying to mimic perfect depth and go for the less is more approach, though that's is just my opinion...

but I worry about it daily. :blink: :lol:

if I were to guess, the left and right images are only using a fraction of the lens, and the CCD will be small too. this would act as a separation multiplier. Or since its likely this is only for close action, perhaps it will only have a maximum effective separation of ~1-2cm and you need to have a long lens on it.... but I'm only guessing. very interesting

LCD projectors do polarise the light, there are ways to use them, but there are more problems. dlp is the way to go. for the rest it depends on how fussy you are. it is *possible* to watch stereo via a white screen, but its generally going to be terrible, with lots of ghosting. same with cheaper filters (careful not to burn your filters) - you get more ghosting. you can even paint a wall with aluminum flaked paints and you'll get a reasonable result but silver screens are the only way to ensure a good result.

yes I figured you'd say that (about hte toe-in/ no toe-in :P ) my answer was that, as long as we're using the same phrases, the film plane is the correct rotation point. (I've never called it a 'film plane" so I'm assuming what you mean by that). Having said that, your keystone won't be improved, but your stereo would be. Keystoning is a result of the vertical parallax that toe-in introduces, so while the keystoning would be more accurate when you identify the correct rotation point, it doesn't follow that by correctly rotating , your keystoning would be reduced. an example would be if your original rotation point caused you to effectively rotate the cameras *less*. if that happened, your correct toe-in would increase the keystoning.