Walter Murch on 3d. Let's chat.

[Clyde DeSouza]

However, the line has to be drawn somewhere (don’t you think?). The fact that S3d can be used to amplify truly horrific subject matter I don’t feel is a good thing, a 2d image on a 2d screen is a good psychological safety barrier (I think). As filmmakers were not going out of our way to warp people’s minds, video games already do a good job of that already, and unfortunately S-3d is going to make that even more of a slippery slope..

Eric you bring up an interesting point, and something that's always been on my mind with technology, that of it having two uses. Good and bad. As we know almost every technology has been spearheaded and/or granted/funded by the military.

In a similar vein, just look at VR and Simulators. (I've written and spoken about this at talks on Stereo3D)... The one reason that Simulators use stereoscopic 3D visuals, is because it allows the true "simulation" of a location or situation, to acclimatize (some might even say desensitize) soldiers or operations forces to these situations.

...what if we could actually turn this to good? To appeal to the "good and humane" side of people/soldiers with these very realistic visuals?
Would then s3D play an important part? just as it can play a destructive part in games with Stereoscopic Gore?

There's always two sides to leading edge technology. It's our choice which side to push it to. Hopefully it's for the better.

Regards,
Clyde


P.s about texture and detail.. the same reasons some Cinematographers want high dynamic range sensors, and some even swear by the "film look" or film rez, yet they forget about "spatial resolution" that a Stereoscopically captured movie can present. They might also not realize that not every nuance and "bump" that makes real world "textures" gets converted in 2D to 3D conversions.

PPS. Sony's SXRD projectors will become popular (others are also soon launchinh 4k projectors). 4K with Above/Below lens in front.. full 2K 3D, with no time multiplexed strobing, triple flash etc...


PPPs Thanks for bringing up the excellent point on drawing the line between amping realism with S3D!

 

[John Fairstein]

3D conversions

3D conversions

Tom, you are right on. Converting something that was originally acquired in 2D to 3D is the wrong approach. This is especially true for classics such as those Peter M. mentions. Those films were carefully crafted to be displayed in 2D. I think James Cameron is on the warpath about 2D - 3D conversions (and bad 3D in general).

Mike, why are you assuming that Blade Runner would need to be converted? I thought the question posed had to do with whether the picture might have been enhanced if it was SHOT in 3D? BLADE RUNNER might be epic in 3D, I think. Who's to say?

 

[John Fairstein]

S3D is not a gimmick

S3D is not a gimmick

Yes Clyde makes an excellent point. S3D is not a gimmick, it is the natural way of acquiring a scene. The stereo camera has two "eyes." Just like when we acquire stereo audio with 2 microphone "ears." Stereo conversions in post are a gimmick. Maybe folks are confused about how an S3D camera works.

The common mistake we make in comparing true Stereoscopic 3D capture over and over again to things such as B/W versus Colorized, or Mono v/s Stereo sound etc.. or even 2d to 3D converted movies etc, is that these are done "after the fact".

 

[John Fairstein]

Schindlers List

Schindlers List

Re Schindlers List, an S3D shot from above looking down at the prisoner in his cell, a shot of the evil colonel looking down on the yard from his balcony, etc. Also there is no problem with S3D in BW. It works fine.

Schindlers List: ...from the same article above, Can 3D replicate the feeling of claustrophobia, or of solitary confinement?

Regards
Clyde

 

[Eric Lange]

Yes Clyde makes an excellent point. S3D is not a gimmick, it is the natural way of acquiring a scene. The stereo camera has two "eyes." Just like when we acquire stereo audio with 2 microphone "ears." Stereo conversions in post are a gimmick. Maybe folks are confused about how an S3D camera works.

To a certain degree, a stereogram has the capability to re-present to our eyes and brains what it would naturally see. However this is really only the case for a very narrow range of depth, for example examining the surfaces of an Egyptian tomb or carved bas-relief in a French Mesolithic cave. Things start to become rather unnatural when the depth range and corresponding range of parallaxes start to increase; this is the point where apparent scale and the break-down of view accommodation vs. convergence starts to kick in. There are solutions to this but these involve re-sampling or reining things into safer ranges of on-screen parallax. The way stereograms are presented on screen are not the same as how your eyes would access and interpret a true physically built 3d scene.

Everybody agrees dimensionalized movies can be sucky.

I think the technical applications of s-3d are freaking awesome and so are the quasi documentary and live event applications. However do I want S-3d in my art? I don’t think so (personally).

The term gimmick is a bit strong as it implies a sort of crooked deception.

Another word or noun or adjective is needed instead. When I say Gimmick I basically mean…

Esthetically challenged unnecessarily flashy BS

I can’t find a good adjective or noun for that; I’m sure in German that there is a wonderfully ironic single word to sum that all up.

(come on you Germans!)


I think the fun factor for s-3d is awesome, the kids love that kind of thing, and various horror pics for the older kids. Of course it’s certainly possible to create “high art” in s-3d, but I just don’t think it’s going to happen that often (if at all). Fifty years from now, maybe there will have been about three movies in total over that entire period that constitute "high art" (at least for cinema) and the rest of the hundreds of s-3d movies made over that same period will all have been ‘engineered” to thrill; there’s nothing wrong with that, that’s entertainment.

The spatial ambiguity of the 2d image is artful. For example the original movie of the Hustler has amazing photography in it, Eugen Schufftan won the “Oscar” for best cinematography. Some of the shots take a substantial number of seconds to figure out your frame of reference, which is pretty amazing considering half of the action takes place around a pool table. This movie would have been a complete disaster in s-3d and Schufftan certainly would not have won the Oscar. In this case s-3d would have been completely wrong and inappropriate for the script.

On the other hand I think the recent re-make of BSG (tv series) was an amazing drama with very complex long arcing plot lines, so it didn’t matter if things were not too realistic “cosmetically” but it certainly would have been more thrilling to see that in s-3d; but the s-3d would not detract from the good drama and plot lines. There’s a weird balance in there somewhere.

This is why I think that TV and broadcast once they get going are going to be able to develop most of the yet undiscovered grammar for s-3d story telling and presentation perhaps over cinema.

 

[Hrvoje Simic]

No. Convergence is at screen plane. For example, the shots of planets, stars, etc in Avatar in the beginning are at screen plane. If you took off your glasses during the beginning you'd see that the shots of the planets etc look fine in 2D - because they're converged to screen plane.

Disney movies - even Murch's own Captain Eo for that matter - converge the stars etc stars a way behind screen plane. Sure it looks cooler but you get ghosting and are also causing eyestrain. Personally, Captain Eo caused quite a bit of eyestrain - Avatar caused no eyestrain but Eo managed to cause pain in 20 minutes.

Bruce, I always appreciate your posts. Here's the cause of the disagreement: convergence can be near or at the screen plane, but it severely limits the storytelling. Specific example cannot prove general adequacy. One can match focus and convergence near the screen plane on some shots, which does not prove usability for all of them, because every scene is different.

Sense of depth depends on perceiving convergence caused offset. Brain calculates depth based on convergence. You can try to squint your eyes a little and slightly blur your vision and you will still see depth. Because of the convergence. If you have an object with a volume, no matter how close the convergence is to the screen plane it has to vary throughout the object, unless it's a flat shaped object parallel to the screen plane, otherwise you wouldn't perceive the object's depth. If there is a space and you want the viewer to perceive it, not the object in the space, but space itself, convergence will have to be well behind the screen.

If you'd have every scene with convergence stuck to screen plane, you'd limit the sense of depth, of watching into space, the space would mostly be "behind" the stuff you are looking at. You'd also tell a story with everything in the foreground.

Deeper convergence does not dictate ghosting, which (unless it's a poor 3D conversion artifact) is often a consequence of crosstalk, dependent on glasses, projector or panel.

There is a simple test. You can go to the nearest TV store where they have Panasonic promo material on VT25's. Cover the horribly designed glasses so the light doesn't leak in and bounce inside them, and watch all those shots cca. 4-5 feet away. You should see one wide shot with a train station in perspective. Notice the feeling of depth, and watching into space and notice the convergence caused offset without the glasses. There is also a fly-over above mountains. Notice the sense of depth and check the convergence. Then try to imagine if it would be the same effect if your eyes were converged to the screen surface.

Cameron said, "We just came up with a whole new stereo aesthetic from scratch, and it goes like this: you put the convergence where you're looking. Period. Now, it's never, ever been done that way, and any old-school stereographer....would tell you, 'That's not what you do. You put the convergence at the screen plane and you let the objects come forward into the viewer space.' [That's] all bullshit, it's complete bullshit. That's not how it works, and you can't make a movie that way because you can't cut [creatively in post-production from shot to shot]. So we came up with a whole other way of doing it.")

Cameras converge and focus to a same point in space, but for the viewer in the cinema, the focus has to be on screen plane, while convergence varies in simulated space dependently to scene. It has to vary because it is essential in depth perception. How it varies lies in the hands of a good stereographer, director and editor. Tricks around the fundamental laws may fool the brain a little but reduce the realism adding fake looking depth, which makes a lot of shots look like composites, not just the shallow DOF and long lens circus ones.

When speaking about the illusion of 3D or "stereo" one must differentiate the specific solutions between general adoption as well as depth tricks between realistic immersion. Each has a price. With over simplified and generalized perception of 3D storytelling circus as it currently is, these two are often perceived as one, which leads to a lot of misunderstandings and misuses.

What also causes eye strain is wrong point of interest for the viewer, the point of the scene for which cameras weren't aligned to. Our eyes always scan. They never stare at one spot unless one is intoxicated, unconscious or lost in thought. Eye want's to scan the 3D image and fixed convergence limits that. With that, it is highly unlikely that the viewer will always choose the same point of interest as the director wanted. Also, natural human interest to explore the shot/scene leads to looking around the shot. All this puts additional strain to the optic nerve.



Btw, I have used the word "convergence" 13 times in this post,
without this one time which doesn't count. Guess I should have used copy/paste.

 

[John Fairstein]

S3D reproduces what we naturally see

S3D reproduces what we naturally see

Eric, yes, this simple concept is exactly what folks are having a hard time understanding:

To a certain degree, a stereogram has the capability to re-present to our eyes and brains what it would naturally see.

But I'm not sure what you mean by a "narrow range of depth." My experience shooting S3D stills has been that you can reproduce virtually any depth. Beamsplitter S3D motion picture rigs allow you to adjust the interocular distance over a wide range: closer for closer subjects, and wider for more distant subjects. With the appropriate equipment, microscopic objects and distant stars can be imaged in S3D.

There is a full description of the formula for determining interocular distance at http://www.berezin.com/3d/Tech/lens_separation_in_stereo_photog.htm. Many of us just use the 1/30 rule: divide the distance to the subject by 30 to obtain the interocular distance to place the subject at the window.

The remaining adjustment is convergence for subjects close to the camera, e.g. less than 20 feet (see my earlier graph). I think James Cameron has solved convergence with his PACE system, but I don't know how it works. Clyde deSouza knows a lot about separation and convergence.

I am prejudiced against converting 2D to 3D in post, so I refer to it in the pejorative as a "gimmick." I guess it is more polite to say 2 1/2 D.

 

[Eric Lange]

Eric, yes, this simple concept is exactly what folks are having a hard time understanding:



But I'm not sure what you mean by a "narrow range of depth." My experience shooting S3D stills has been that you can reproduce virtually any depth. Beamsplitter S3D motion picture rigs allow you to adjust the interocular distance over a wide range: closer for closer subjects, and wider for more distant subjects. With the appropriate equipment, microscopic objects and distant stars can be imaged in S3D.

There is a full description of the formula for determining interocular distance at http://www.berezin.com/3d/Tech/lens_separation_in_stereo_photog.htm. Many of us just use the 1/30 rule: divide the distance to the subject by 30 to obtain the interocular distance to place the subject at the window.

The remaining adjustment is convergence for subjects close to the camera, e.g. less than 20 feet (see my earlier graph). I think James Cameron has solved convergence with his PACE system, but I don't know how it works. Clyde deSouza knows a lot about separation and convergence.

I am prejudiced against converting 2D to 3D in post, so I refer to it in the pejorative as a "gimmick." I guess it is more polite to say 2 1/2 D.

As a photogrammetrist we tend to push things a lot more than for example generalized stereograms as we are looking to squeeze out every possible drop of 3d data out of the imagery. So for example if it’s a historical facade, “gothic ruin” we might stereo record with a 1:5 or 1:8 base separation to distance ratio i.e. the same ratio of having an object a foot and half away from your face. This would be a 60mm lens on a 60x90mm camera, or even 100mm lens on a 130 mm by 180 mm plate camera. This gives you a ratio that is very life like for examining a complex 3d surface. However if you have a gothic arch where you see trees 200 meters into the distance (on the other side), and simultaneously have to view parts of the facade, as you know you have to dramatically reduce the base separation (or so called IOD). The problem here is that to accommodate such a colossal range of parallax and depth that you have to use a very narrow IOD like 1:30 or smaller as you describe. The problem here is that now you have a carboarding effect that both flattens the trees considerably as well as the foreground detail. To me this is un-exciting stereo, and is not a very naturalistic effect to me. The Victorians did that one to death I think. If you use wider IOD to get more of an effect, then the range of on screen parallax is too much and it is not possible to stereo-resolve the scene simultaneously foreground and back ground, no matter how you adjust HIT. This causes bad headaches. So for large depth ranges with associated large ranges of parallax foreground to back ground you either have cardboard or you have pain unless you re-sample all of the parallaxes through a different more advanced set of techniques. In this situation there is not a happy medium using conventional techniques.

Photogrammetrists have been using convergent stereo photogrammetry for almost a hundred years, and providing you crop in there is a lot you can do with it for particular isolated circumstances and can allow one to use wider “IOD” safely.

 

[Robert Sanders]

3D is just another brush in the artist's toolset. Those who like 3D embrace the new brush and gladly add it along with the other brushes. Those who despise 3D want the brush removed completely from the toolset and find it an unnecessary tool.

This debate will go for a long time, I think. Personally, I'm a fan of 3D and have been since I was very young. I am loving the current 3D renaissance.

I would be remise if I didn't admit that I feel my voice is being shouted down vociferously by the naysayers. Actually, I'm kind of taken aback by the vehemence of the push back. But I try not to let it bother me as I know that the studios and the consumer electronics manufacturer's are pushing the new brush in a big way.

 

[Eric Lange]

3D is just another brush in the artist's toolset. Those who like 3D embrace the new brush and gladly add it along with the other brushes. Those who despise 3D want the brush removed completely from the toolset and find it an unnecessary tool.

This debate will go for a long time, I think. Personally, I'm a fan of 3D and have been since I was very young. I am loving the current 3D renaissance.

I would be remise if I didn't admit that I feel my voice is being shouted down vociferously by the naysayers. Actually, I'm kind of taken aback by the vehemence of the push back. But I try not to let it bother me as I know that the studios and the consumer electronics manufacturer's are pushing the new brush in a big way.

I’ll drink to that! let’s hope it’s a brush not a broom!

Cheers,

Eric

 

[Clyde DeSouza]

Converge on focus...
...is 3D just a tool or a new medium for visual story telling
...do tools need a new language to be developed..or are usage instructions sufficient?...

...in real life do we move 10 steps backward when focussing are attention on someone in the foreground speaking, and then take 10 steps in front as the person in the background delivers his lines?..

This is addressed in this article: http://realvision.ae/blog/2010/03/is-3d-a-tool-or-new-storytelling-medium-for-cinematographers/

Regards,
Clyde
( note the harshness of tone in those articles is reserved only from some die hard moviemakers who refuse to open their minds eye to seeing anything differently, and is not at Cinematographers as a whole)

 

[John Fairstein]

Large ranges of depth

Large ranges of depth

...However if you have a gothic arch where you see trees 200 meters into the distance (on the other side), and simultaneously have to view parts of the facade, as you know you have to dramatically reduce the base separation (or so called IOD). The problem here is that to accommodate such a colossal range of parallax and depth that you have to use a very narrow IOD like 1:30 or smaller as you describe. The problem here is that now you have a carboarding effect that both flattens the trees considerably as well as the foreground detail.

Photogrammetrists have been using convergent stereo photogrammetry for almost a hundred years, and providing you crop in there is a lot you can do with it for particular isolated circumstances and can allow one to use wider “IOD” safely.

Eric, yes, now I understand the point you raised earlier. So how should an S3D cinematographer address a situation like you describe? Should the cinematographer avoid large ranges of depth altogether, <strike>or apply some convergence</strike>? [Applying convergence to distant objects is a duh idea on my part! Adding convergence only makes distant objects more difficult for the eyes to accommodate] Or take another tack: what if you set the base (IOD) for the near point, then let focus fall off slightly in the distance to mitigate the narrow base? (Clyde may cringe at this suggestion.) And how do our eyes/brain seem to be able to accommodate large ranges of depth?

 

[John Fairstein]

Clyde's article is excellent

Clyde's article is excellent

This is addressed in this article: http://realvision.ae/blog/2010/03/is-3d-a-tool-or-new-storytelling-medium-for-cinematographers/

Clyde, this is a great article. I wish every cinematographer read it.

 

[Ketch Rossi]

Some interesting points in the article for sure, thanks for sharing it Clyde

 

[Clyde DeSouza]

what if you set the base (IOD) for the near point, then let focus fall off slightly in the distance to mitigate the narrow base? (Clyde may cringe at this suggestion.) And how do our eyes/brain seem to be able to accommodate large ranges of depth?

Fortunately, we aren't doing a photogrammetry documentation of a scene in a typical narrative 3D movie, so with proper lensing, we can use selective focus to do the job.

The trick seems to be to be completely un ambigious in the intent, I.e use extremely shallow depth of field, or none at all. Audience headaches occur with semi focused objects in a scene..For some reason the brain locks in very fast on anomalies in a 3d scene then discards it moves to the subject of s story. While no instant harm may occur, eye muscles never complain like other body muscles, the headache usually kicks in much later after the film is over. Given time we will resolve faster the intent as we now do in 2d movies.

So foreground subject of scene is in sharp focus and background is completely blurred out, and in racking focus, this might be done while *moving * the camera to reframe at the same time, so as to move the foreground subject at least a bit out of the way of the frame, and guide audiences properly.

Tron uses some of these techniques very nicely, granted being a cgi and blue screen heavy movie, it made it somewhat easier.

Using bigger sensors, helped with the depth of field to an extent as well.

Thanks guys for the kind words,
Regards
Clyde

 

[Eric Lange]

Eric, yes, now I understand the point you raised earlier. So how should an S3D cinematographer address a situation like you describe? Should the cinematographer avoid large ranges of depth altogether, or apply some convergence? Or take another tack: what if you set the base (IOD) for the near point, then let focus fall off slightly in the distance to mitigate the narrow base? (Clyde may cringe at this suggestion.) And how do our eyes/brain seem to be able to accommodate large ranges of depth?

I’ve been busy so have not been able to reply etc.

My main plank is to make sure that stereo is rich, detailed engaging and easy on the eye and brain. My worry is that if the ranges of parallax are too much then this causes pain (plain and simple). If the ranges of parallax are safe and comfortable for corresponding large depth ranges then, frankly the fine 3d detail is flattened to the point of almost being completely eliminated. In such a situation props , sets, and talent appear as cardboard cut outs disposed at various apparent depths. If you like, to reach more optimal and engaging stereo, something has to be thrown out unless you want to use very advanced digital re-sampling techniques.

To answer your question in a cursory way…

1. “Should the cinematographer avoid large ranges of depth altogether”

In a lot of cases I would say yes.

So stereo cinematographers are leaning to control the backgrounds, i.e. in very clever ways. One set of stereo show reels I saw had large busses parked in front of the middle back ground and in front of a textureless white wall, and then all you could see beyond was clear blue sky. The blue sky had no visual texture to lock onto, and the busses in effect “blocked” or limited any further detail to be resolved at a greater distance. This meant that the girls on the car bonnets could have slightly wider IODs given over to them without having to compete with objects in the far distance. By extension a lot of things are being shot on green/blue screen so that virtual sets and backgrounds are inserted later (there are problems with this also, beyond the scope of this thread). Some work has been carried out on manipulating the ranges of parallax for CG work in basic ways; a good early example of the thinking behind these techniques was employed on Beowulf. So various techniques can be combined, preferably as a 3d composite.

If you don’t have access to “heavy” digital techniques, then try to shoot with wide angle low distortion lenses, (preferably biogon/symmetric type, rather than retro-focal lenses).


2. “or apply some convergence?”

No, convergence in this situation makes things much worse for larger ranges of parallax, in essence if you have a foreground and a for off detailed back ground, if you converge on the foreground subject then the on screen parallaxes for the back ground subjects become ridiculously huge. I would use convergent cameras if I ONLY have a side by side arrangement where the cameras have a physical limiting width. I don’t like beam splitter boxes, as it’s very difficult to have such a large piece of glass to perform really accurately (for our applications) and they seem like such a contraption, (but a necessary evil in a lot of circumstances). So I would use convergence in favor of a splitter box, only if the back ground was a solid color or indistinct texture and there is only one foreground subject. In other words it’s a possible solution when you can’t get small enough IOD if you are using a side by side camera arrangement.

3. “what if you set the base (IOD) for the near point, then let focus fall off slightly in the distance to mitigate the narrow base?”

This is a slightly mixed idea but raises a good point, that Clyde talks about as well. The whole question of how blurry does a back ground need to be before the brain does not want to lock onto it, or try to resolve these stereo features. Carrying out tests with sinusoidal test patterns of varying frequencies set into the background set against a foreground subject would be interesting to find a useful limit of blurriness. I would say that the bokeh has to be very substantial to in effect produce a “safe” background. Hey, just experiment to find out what works for your particular situation.

4. “And how do our eyes/brain seem to be able to accommodate large ranges of depth?”

Well there are several things going on and is a topic that is much debated.

So for example, if you hold out your thumb, the eyes swivel in their sockets to fixate on a point on your thumb. The brain locks onto high frequency, high contrast features, and is able to perform something akin to pattern matching to find corresponding point pairs from which depth is “extracted”. Depth ranges in detail are extracted for things a bit in front of your thumb and things a bit into the distance are also stereo resolved or completely fused into a single coherent percept. The thing that’s really interesting is the stuff in the background as it is NOT stereo fused but appears as a double image. However, even more interesting is that fact the we have a sense of depth for where these double images fall into the background. I.e. un-fused stereo imagery still has an associated “sense” of depth associated with it. In other words our brains are still able to construct a spatialized context for un-fused stereo imagery. Fixate your gaze to a distant object and your thumb becomes a blurry double image. So it is a very useful mechanism to have the focusing of the eyes to focus to the correct distance when we converge on a point as it means that objects at different distances are automatically blurred out. In this case the brain does not try to stereo match the blurs, but still has a sense of its relative disposition. Part of the problem with the movie stereograms (like what Clyde is talking about) is that everything is in a sharp focus for a large range of depth. So the brain tries to lock onto the whole lot and does not have the ability to filter out more difficult depth ranges as these are in high focus. Secondly you get the breakdown of view accommodation and convergence, i.e. eye focus is different from what we would naturally use for a particular convergence angle (in most instances). In some sense, the Murch articale is right in saying that our brains are forced to processes a much larger volume of stereoscopic “data” than it normally would in real life.

This is just scratching the surface.... but in any case cinematographers will learn useful s-3d technique, either but osmosis, through little steps or giant leaps. I think it’s more of a case of recognizing the limitations and understanding what NOT to do, and then trying to maximize what it is that you CAN do.

In that respect it’s no different from how the grammar and techniques were devised and evolved for regular 2d cinematography. I like Clyde’s point that S3d is not a tool he says it’s the medium. I liked that bit.

Cheers,

Eric

 

[axel ebermann]

Having shot quite a bit of 3D years ago before it became 'sexy' I think there is a point to the argument that the current stereoscopic way of achieving 3D is problematic because of the difference between the focal plane (the screen) and the point of convergence for the eyes.

I think 3D should not be dismissed as a gimmick, but it is also not the solution to all our visual aspirations.


First and foremost I think it is necessary to define what we are talking about. Throwing all 3D in the same bucket does not really make for a useful discussion.

Image acquisition: Avatar where an entirely artificial 3D space was created, just as well as 3D animated movies are lightyears away from movies shot in 2D, that are later tweaked to be projected in 3D.

Anybody who has ever composed shots & scenes in 3D will most likely find it creatively very limiting.
Ideally there are several picture planes that move according to a certain pattern. 3D is pretty much lost and often straight-out irritating when shot without infinite depth of field or with a dark background etc., etc.

So as nice of a toy 3D might be I find myself often in the situation where I feel that I have become the slave of a gimmick.

Recently I was on a shoot with the new Panasonic AG-3DA1, which makes the entire tech side remarkably simply (admittedly with an image quality on the lower side of the spectrum).

The director, a very experienced and visual veteran of the film world was struggling all the way because everything that would have created a solid 3D feeling was counter intuitive to what he regarded as beautiful and artistic.

I like 3D - but like everything in the film world there are circumstances / stories where it enhances the experience - and circumstances where it is ill used.

My fear is that with 3D cell phones coming out etc. we will be subjected to a smorgasbord of terrible 3D with guttwrenchingly ill conceived points of conversion.
And suddenly every customer wants nothing but 3D - without understanding the implications of it...

 

[Eric Lange]

Having shot quite a bit of 3D years ago before it became 'sexy' I think there is a point to the argument that the current stereoscopic way of achieving 3D is problematic because of the difference between the focal plane (the screen) and the point of convergence for the eyes.

I think 3D should not be dismissed as a gimmick, but it is also not the solution to all our visual aspirations.


First and foremost I think it is necessary to define what we are talking about. Throwing all 3D in the same bucket does not really make for a useful discussion.

Image acquisition: Avatar where an entirely artificial 3D space was created, just as well as 3D animated movies are lightyears away from movies shot in 2D, that are later tweaked to be projected in 3D.

Anybody who has ever composed shots & scenes in 3D will most likely find it creatively very limiting.
Ideally there are several picture planes that move according to a certain pattern. 3D is pretty much lost and often straight-out irritating when shot without infinite depth of field or with a dark background etc., etc.

So as nice of a toy 3D might be I find myself often in the situation where I feel that I have become the slave of a gimmick.

Recently I was on a shoot with the new Panasonic AG-3DA1, which makes the entire tech side remarkably simply (admittedly with an image quality on the lower side of the spectrum).

The director, a very experienced and visual veteran of the film world was struggling all the way because everything that would have created a solid 3D feeling was counter intuitive to what he regarded as beautiful and artistic.

I like 3D - but like everything in the film world there are circumstances / stories where it enhances the experience - and circumstances where it is ill used.

My fear is that with 3D cell phones coming out etc. we will be subjected to a smorgasbord of terrible 3D with guttwrenchingly ill conceived points of conversion.
And suddenly every customer wants nothing but 3D - without understanding the implications of it...

Axel that’s a really awesome post.

I think that really sums up a lot of the difficulties and realities.

For a lot of us, speculators or folks developing new tech (such as myself) in the 3d and s-3d arena, we all want to know are we wasting our time money and efforts on s-3d type applications (for the longer term)?

My piece of litmus paper or yard stick for me, is to see how the Nintendo 3ds, (auto-stereoscopic hand held gaming device) will be received. It’s stereo capability is somewhat limited, and better apps/games will need to be developed. But for me if it gains acceptance then it will tell me that there is a basic human need and desire for s-3d media, or if it dies, then perhaps s-3d applications may in the future recede to more specialized applications.

So that’s one I’m going to watch very closely.

www.nintendo.com/3ds

comes out in march.

I would also add, that if the technologies that are currently being developed behind the scenes (as it were) are followed through on, then I think you will be surprised at how a seemingly broad and disparate set of s-3d technologies will converge into a fairly large “bucket” (as you say); but even then techniques will certainly differ substantially depending on the specific problem. ( I know that’s a bit arse covering and vague, but I think you get what I mean).

…and yes I think we are ALL very fearful of as you say of “gutwrenchingly” bad implementations of s-3d. I.e. that this may have the capability to sink “good” S-3d, as the bad ones may be far more numerous.

 

[Karel Bata]

Just stumbled across this post. Oh yawn...

On the one hand healthy debate is to be welcomed. On the other, the way some self-appointed pundits, especially Ebert, keep banging on that 3D is a 'fad' and keep producing one poorly argued 'proof' after another that 3D doesn't really work is now way beyond tedious.

First, it has to be understood that 3D is not really 3D. It's stereoscopy. It's an illusion. In the same way that stereoscopic sound produces an illusion of audio width (and some depth), stereoscopic cinema produces an illusion of 3D. True 3D is the province of sculptors and architects, not film-makers. So to pick holes in it's ability to reproduce real 3D is somewhat missing the point. Cinema always was a creative verisimilitude of reality, and never the real thing.

Likewise to say that 3D imposes limitations not experienced in 2D is a failure to see that it is not really the same thing. 3D offers a different experience, and is not something you should just bolt on to a 2D production to boost sales. Though many productions currently do, and with dire results.

The example Murch quotes of his experience of editing a 3D project from the 1980s is, what can I say?, hopelessly outdated. Things have come such a long way since, and are getting better all the time. The problems he mentions are specific to display systems which aim to produce the illusion of 3D, not of 3D itself. A quick example - increased strobing is specific to single projector multi-flash systems like Real-D, and is not apparent in dual-projector systems. James Cameron is now talking about projecting Avatar 2 at 60fps, which many current projectors are already capable of doing. Problem thus 'solved'.

3D made a big splash a year ago, and it is still going strong. In mid-2009 there were about 5,000 3D movie screens worldwide. Now there are over 22,300. The number of announced 3D titles for 2011 is now approaching 60. 50% to 70% of box office revenues are generated by 3D titles*. This is not a fad. The train has left the station.

The situation is akin to the introduction of sound, or of the change from monochrome to color. Both had their vocal critics, who now look somewhat quaint. I respect Mr Murch immensely, but his 'salt shaker' example is a little baffling. Has he worked in 2D for so long that he now thinks we actually see the world that way? We do not. 2D is a set of conventions that have evolved over thousand of years from cave paintings, to Old Masters, to Orson Welles. But it's not how we see the world. Is he really unaware of the double images we get naturally, all the time? Here's a quicky graphic:
How We See The World

As to his criticism that the problem of convergence and focus are insoluble, he is talking rubbish. Avatar, regarded among stereographers as a very conservative application of 3D, locked both on to the screen plane. Again, problem 'solved'. But there are better, more elegant, solutions regularly being used now.

But one thing that I have come to realise, more so since I've started actually doing it, is that 2D and 3D are entirely different animals. To extract the best from either requires a substantially different mindset. A good 2D movie will never really work in 3D, and is somewhat pointless, and a bit of a con. Nor can a 2D movie simply be made by using only, say, 'the left eye' from a competent 3D movie. So seeing a 2D version of Scorcese' Hugo Cabret will simply miss the point. Many producers, with an eye to revenues, will try to tell you otherwise - that you can shoot 2D and 3D at the same time. I believe they are fundamentally wrong, and their thinking that way will produce compromises that will (more often than not) hinder the acceptance of 3D. It's an issue that, as 3D techniques evolve, will receive increasing debate.

My 2p.

* source: 3D - A goose that lays a golden egg?

 

[Eric Lange]

Wow,

Karel,

Thanks a million for your 2p!

When I go to various stereo based conferences and trade shows etc., we come away thinking wow lets pass round the cigars, we have the best solution on the planet for what we do, we are going to make billions blah blah blah, isn’t the future going to be super amazing!

Then a few months later reality sets in when we have to figure out how to bridge various gaps in technology from what we have versus what is currently being used. When you are at the coal face of trying to build businesses around technological solutions you start to look at things with a very cold eye to make sure that one's efforts are not being wasted one way or another; especially as the developments we have, have a much longer arc and associated implications than let’s say a "current fad".

I will certainly take some time to properly digest the PriceWaterHouseCoopers document; the "3d- a Goose that Lays a Golden egg?

Get back to you on ”this” some time.

Are you from the UK, (the 2p reference?).

Ta

Eric