Why is a 3D rig expensive?

[Michael Hastings]

There are many 3D rigs that have been utilized in recent history, yet there's a dearth of real understanding of the underlying complexities. Don't take my comments personally. I just think its important to make a clear distinction of these concepts for people reading this thread that haven't built and experimented with their own rigs yet.

Thanks, Jeff. I don't take it personally, because although sometimes these forums can suck when they get bogged down in endless "A is God, B and C suck 'cause they bring out new cameras every 3 months" posts - this thread is an example of the beauty of the internet: We've already got great input from yourself; Joey; Markus; Pedro; J.J.; Mark L.; and we're only 20 posts in - that's a hell of a (virtual) room.

I've been in the video industry for 25 years and consider myself an expert in a fair part of it - but other than very simple stuff 3D is new to me. I can pick up a fair amount of it from research but it is extremely helpful to have these professionals weigh in and clarify things.

 

[Petri Teittinen]

This is an excellent thread, please do go on, gentlemen. I'm thinking about making my own rig once the 8x Scarlet is out, so this is gold for me.

(sorry, off-topic) I think someone brushed briefly on this earlier... my pet peeve, so to speak, with modern 3D films has to do with the use of shallow DOF. While it's perfectly fine (understating it radically) in traditional moviemaking, it bugs me immensely when used in 3D features. The main reason for this is the fact that quite often I would like to focus on a different plane than the one decided by the director.

I was really frustrated by this when viewing Avatar; there was so much interesting stuff on screen, but Cameron wouldn't let me see it all on my leisure. I guess this is why I think CGI features are better suited for 3D. Not only do CGI features generally appear to cultivate a much deeper DOF than live action films, they have the capability of practically infinite DOF, with everything in the frame in perfect focus. I wonder if that could actually work with current 3D technologies, and whether it would allow my eyes to focus on whatever plane/depth they wanted in the frame.

 

[Markus Stone]

Hi Petri,

One of the practical issues is that you lose a stop of light simply by putting the cameras on a mirror rig, which puts you behind the eight ball a bit when it comes to adding even more light to the scene to get a deep depth of field. It's possible, of course, but not necessarily practical.

-Markus

 

[Pedro Guimaraes]

Stop = more light= more $$$$$

Add to that 1 stop loss on the beam splitter, iso 200-320 of the RED, low budget.....shallow focus is unavoidable.

Make a strong argument for using smaller sensor cameras for 3D.

Do you know how much light they used in muppets 3D to shoot at F11? enough to have a few of the puppeteers pass out under the stage from heat exhaustion!!

Coraline had lots of issues with the mini "set & actors" shrinking and expanding due to heat from lighting! A big issue when your shooting stop motion 3D with on DSLR motion control rig. For example a bed might of been 2" inches long at the beginning of the shot then hours later it had expanded due to heat to 2.2" !! Eventually they figured out the problem and addressed it. just goes to show some of the issues people have had on trying to shoot with a deep focus and lots of light required for it....

Talk about making actors uncomfortable and trying to keep them from sweating! So There real world issues in achieving a F8 - F11 indoors on location. Hell even a 5.6 can be hard to get on the RED one/beamsplitter combo.

 

[Pedro Guimaraes]

OK, let's clear this up before we all confuse the hell out of everyone reading this thread. Some of you sound like you're confusing the Inter-Occular distance and the Convergence angle. They are not the same and they must be isolated and controlled independently from each other.

I'm not confused.

IO is one thing and convergence is another.

In Simple terms think of IO as a 3D volume knob. Convergence on the other had mainly determines where the screen plane is.

Actually they don't have to be controlled independently. For example the ET rig automatically adjust the convergence when you adjust IO to keep the screen plane where you set it. It's a great feature! So in the ET case they are linked.

The inter-occular is typically not adjusted during a shot and 99% of the time is set at 65mm.

While we don't "pull" IO that often we it's done more that just 1% of the time.

I can also tell you from numerous 3D productions I have been on that in fact we rarely stay at 65mm IO. So your figure is definitely incorrect in that regard.

I'm doing a 3D shoot tonight using the ET rig/Reds and a Cunima beamsplitter on a technocrane. I can almost guarantee most of the time we will be below 65mm....except for some of the wider shots. Good chance of us changing the IO during the technocrane shot.

This is typically adjusted during a shot to maintain the subject within the Z axis. This angle can change constantly and dynamically in a shot, yet it should be done within a 20 degree range so as not to induce negative responses from the audience.

I wouldn't even say "typically". In reality alot of the productions I have been on have not "pulled convergence". This is a phenomenon that Cameron is a big fan of. He uses it like focus pulling. Controlling in a way what the audience is seeing. A moving "window" is not always something you want.

Completely opposed to Cameron's convergence pulling method, there are lots of good reasons and lots of good people that like to shoot 3D parallel and converge in post. You eliminate the keystone distortion that you create with toe'in the cameras. Shooting at 4k you have plenty of pixels to shift the images and adjust the convergence in post. In fact if you keyframe your convergence pull that is way more accurate than even highly skilled hands of a convergence/foucs puller. Plus once you coverge during shooting, your kinda baking that in. Shooting parallel keeps that creative choice for the 3D edit review room.

Naturally if you shooting a live 3D event you need to converge at the camera. Or if you have zero budget for 3D corrections in post (which is scary thought). Otherwise there are very strong arguments to shooting parallel.

There are many rigs you can use, but they all perform the same basic principles: replicate human vision. 3D is an attempt to replicate human vision. That's why it works. You can deviate all day long from this paradigm, but it has consequences with how the audience experiences the movie. Just because you can shoot your subjects outside these parameters, doesn't mean it will automatically translate into effective 3D footage. Stay Human! It works 99% of the time.

If you only had to change the IO from 65mm 1% of the time than every 3D rig would have a fixed IO!

anyways,

Rigs are sometimes very different in what they can perform. Especially when you on location in heat, humidity, cold, rain and on a tight schedule. This is why the best ones are more expensive. They more accurate in less time! If your spending $100k for a shoot and you don't "make" your day because you spent so much time fiddling with your rig. That is a big problem! Angry producers!

I always strive to NEVER have the production waiting on "camera" to be ready. having a good rig is key to make this a reality. I love yelling out "camera's ready".


I would say, 3D is an attempt to FOOL your eyes and your brain that you are seeing depth. Not that it's an attempt to replicate human vision. Cameras do not work the same as eyeballs. Also we don't have a "frame around what we see. We need to deal with a phisical screen plane and a screensize issue. We have to balance a variety of factors focal lengths, IO, convergence, near&far points, object in the distance, not having any window violations etc.....all in an attempt to fool your BRAIN thru your eyes that you are seeing depth. It's not a replication of human vision or else the dolls in Coraline would lool like 2" dolls. They don't, because like you mentioned they scaled the IO to trick our brains into thinking those dolls being photographed were much bigger than they were.

To be able to execute this magic trick properly we need to balance many factors....including the IO.

Look it's really simple. Take a slide bar and your DSLR. Go shoot your best friend from across the room with a 65mm IO. Then with the same lens and IO put the camera 2' in front of his face. The results will not be good. Reduce the IO to say 1.25" and you will have better results. He will not look like half the size he was. IO does of course affect scale but there are other factors involved.

My best advice to learning about 3D is ......go out and experiment! Grab your iphone if thats all you have and start shooting some stuff. If you have a DSLR try shooting things in your backyard from different distances and with different focal lengths....you will witness and learn the most intriguing things!

Best way to learn is to just do it.

 

[Michael Althaus]

Lot's of good information in this thread. Does anybody know if there is a simple software to simulate shooting 3D? I imagine something like a simple virtual set where you can mess around with all the parameters (IO, Convergence, f, etc.) and observe the results in realtime. Of course you could do it with any kind of 3D modeling program, but I'm looking for a very simple solution just for learning purposes....

 

[Jeff Coatney]

I'm not confused.

IO is one thing and convergence is another.

In Simple terms think of IO as a 3D volume knob. Convergence on the other had mainly determines where the screen plane is.

Actually they don't have to be controlled independently. For example the ET rig automatically adjust the convergence when you adjust IO to keep the screen plane where you set it. It's a great feature! So in the ET case they are linked.

While we don't "pull" IO that often we it's done more that just 1% of the time.

I can also tell you from numerous 3D productions I have been on that in fact we rarely stay at 65mm IO. So your figure is definitely incorrect in that regard.

I'm doing a 3D shoot tonight using the ET rig/Reds and a Cunima beamsplitter on a technocrane. I can almost guarantee most of the time we will be below 65mm....except for some of the wider shots. Good chance of us changing the IO during the technocrane shot.



I wouldn't even say "typically". In reality alot of the productions I have been on have not "pulled convergence". This is a phenomenon that Cameron is a big fan of. He uses it like focus pulling. Controlling in a way what the audience is seeing. A moving "window" is not always something you want.

Completely opposed to Cameron's convergence pulling method, there are lots of good reasons and lots of good people that like to shoot 3D parallel and converge in post. You eliminate the keystone distortion that you create with toe'in the cameras. Shooting at 4k you have plenty of pixels to shift the images and adjust the convergence in post. In fact if you keyframe your convergence pull that is way more accurate than even highly skilled hands of a convergence/foucs puller. Plus once you coverge during shooting, your kinda baking that in. Shooting parallel keeps that creative choice for the 3D edit review room.

Naturally if you shooting a live 3D event you need to converge at the camera. Or if you have zero budget for 3D corrections in post (which is scary thought). Otherwise there are very strong arguments to shooting parallel.



If you only had to change the IO from 65mm 1% of the time than every 3D rig would have a fixed IO!

anyways,

Rigs are sometimes very different in what they can perform. Especially when you on location in heat, humidity, cold, rain and on a tight schedule. This is why the best ones are more expensive. They more accurate in less time! If your spending $100k for a shoot and you don't "make" your day because you spent so much time fiddling with your rig. That is a big problem! Angry producers!

I always strive to NEVER have the production waiting on "camera" to be ready. having a good rig is key to make this a reality. I love yelling out "camera's ready".


I would say, 3D is an attempt to FOOL your eyes and your brain that you are seeing depth. Not that it's an attempt to replicate human vision. Cameras do not work the same as eyeballs. Also we don't have a "frame around what we see. We need to deal with a phisical screen plane and a screensize issue. We have to balance a variety of factors focal lengths, IO, convergence, near&far points, object in the distance, not having any window violations etc.....all in an attempt to fool your BRAIN thru your eyes that you are seeing depth. It's not a replication of human vision or else the dolls in Coraline would lool like 2" dolls. They don't, because like you mentioned they scaled the IO to trick our brains into thinking those dolls being photographed were much bigger than they were.

To be able to execute this magic trick properly we need to balance many factors....including the IO.

Look it's really simple. Take a slide bar and your DSLR. Go shoot your best friend from across the room with a 65mm IO. Then with the same lens and IO put the camera 2' in front of his face. The results will not be good. Reduce the IO to say 1.25" and you will have better results. He will not look like half the size he was. IO does of course affect scale but there are other factors involved.

My best advice to learning about 3D is ......go out and experiment! Grab your iphone if thats all you have and start shooting some stuff. If you have a DSLR try shooting things in your backyard from different distances and with different focal lengths....you will witness and learn the most intriguing things!

Best way to learn is to just do it.

LOL! Pedro, I have no doubt that different techniques prevail on different films. Everyone who's tried their hand at 3D has used their own experience to inform how they do it. You have yours and I have mine. There is one thing I have found to be true: while there's no school of thought that has yet prevailed above any other, everybody's an expert .

 

[Joey R.]

PEDRO made perfect examples of why you don't want to shoot at a set "human" inter-pupilary all the time. You also can't ONLY shoot parallel if your subject is 5' in front of you. And you can't shoot beamsplitter with subjects 500 yards away on a 2500mm lens. Ok you could but the rig would be completely unwieldy and the glass required would cost more than the rig.

And BTW, with the ET rig's control you CAN actually separate IO from C if you so choose but the whole point of the software is to prevent problems and to constantly calculate the proper CONV tangent whenever the IO is moved. And vice versa.

 

[Christer Engstroem]

Mr Cameron put it this way:
"Interocular distance varies in direct proportion to subject distance from the lens. The closer the subject, the smaller the interocular. The farther the larger. A shot of the Grand Canyon from half a mile away may have a 5' interocular. A shot of a bug from a few inches away may have a 1/4" interocular. Interocular tolerance is subjective, but there is a constant value of background split which cannot be exceeded."

More to read:
http://www.3dathome.org/webpage.aspx?webpage=1952

.c

 

[Markus Stone]

One thing I would add is that it also depends upon how far away the background is.

If your BG parallax is at an acceptable level when looking past your bug to the distant horizon, and then someone places a flat 4 inches behind your bug, suddenly your scene has no depth. You have to open up your interocular if you want to maintain the same sense of 3D.

It is even possible to have a long lens shot with a *larger* IO than a short lens shot, (but obviously in different situations). It depends not only upon the camera - subject distance, but also on the camera - background distance. These together determine how close something can come to camera (out of screen) and still be comfortable.

 

[Pedro Guimaraes]

like I mentioned there are numerous factors that can affect your IO decision. One of which is your back ground......is it homogenous? do you have many geometrical lines in your background? i.e: prison bars, certain buildings.........what the contrast like in the scene? lot so black and white things next to each other? i.e: like a checkered flag?

Really there is ALOT that can factor in to a IO decsion. That is why 3D pre-viz zoftware can only get you so close. You have to evaluate everything in the frame being shot and made a decision.

Jeff your right, everyone does do 3D differently. There are some rules but most can be considered suggestions. There are even some rules that should be broken on occasion for the right effect of course!

What matters is entertained happy viewers. Because they pay my bills = )

 

[Pedro Guimaraes]

Lot's of good information in this thread. Does anybody know if there is a simple software to simulate shooting 3D? I imagine something like a simple virtual set where you can mess around with all the parameters (IO, Convergence, f, etc.) and observe the results in realtime. Of course you could do it with any kind of 3D modeling program, but I'm looking for a very simple solution just for learning purposes....

Yes, The closest I know of is Frameforge pre-viz software.

It allows you to set up a scene (cgi), set your camera rig and settings put on your glasses and view the results of a dynamic camera move thru your virtual set either with anaglyph or with your 3D TV set (active glasses or passive polarization).

You can learn alot from playing with it. But it is no replacement for shooting stereo in real life for learning purposes. Like I said you can just use any old digital camera to start taking stereo pairs and learning. I'm sure you already have a digital camera of some sort and PS on your computer. There you have it a free stereo lab for your education.

But frame forge is cool and has it's value for many purposes.

http://www.frameforge3d.com/previz/

http://www.frameforge3d.com/Products/Stereo-3D/

 

[Stephen Pizzo]

I don’t know 3D, but I know what hurts my eyes ;-)

I don’t know 3D, but I know what hurts my eyes ;-)

As if there hasn't been enough discussion of variable IO, I would like to add:

Consider that we refer to a lens as normal when it yields a perspective similar to human vision. For example, in the 35mm motion format we call 50mm normal and in the 35mm stills format we consider 75mm normal. These two lens/format combinations give us roughly a 27⁰ FOV which when coupled with a 65mm IO can work very well in creating pleasing stereo images. As the FOV gets wider the IO needs to get smaller to maintain a natural look without miniaturization. Conversely as the FOV gets narrower the IO will generally need to increase to maintain a more natural look while avoiding the cardboard cutout quality.

Now I know this is just one consideration in figuring out the correct IO but Pedro, Markus and others have covered a lot of ground and passed on a lot of really excellent hard earned knowledge. We at ET are engineers and technicians, not filmmakers. As a result we have been around an awful lot of very talented stereographers and stereo supervisors. Through this we have seen many styles of calculating proper IO+C. Some conservative and some radical. In all of this I would say that there is no single IO distance that seems to be used more than any other. There is certainly nothing special about a 65mm IO distance unless you also have the magic FOV of 27⁰. Even then; subject distance, negative parallax in the deep background or even screen size will force you to change the IO to something other than 65mm.

I would also like to add that Pedro's suggestion of experimentation is right on the money and well worth trying. If you're here on this forum then it is safe to assume you like to make images. You will have a blast making stereo images and learn so much, so quickly.


Stephen Pizzo
Technica3D.com

 

[Bruce Allen]

This is such a great thread!

I agree with all of you in different ways - Pedro, personally I think you're dead on with everything you say about interocular. Disagree with you RE shooting everything parallel and doing convergence in post though! I'm on the Cameron side of things there

Even when doing CG renders for 3D, we still do a little toe-in. Feels a little more natural to us. I do render extra wide though (2048x1080 or 2200x1080 instead of 1920x1080) , so that we can adjust the overall depth when sitting in the 3D online session though. Yes, I've been known to keyframe this

RE: shallow depth of field... Avatar didn't bug me as much as I thought it would. I found that my eye was naturally led to the sharp point of the image - which was where the 3D convergence etc worked well. I think the problem with having EVERYTHING in focus is that then you eye can focus on something in the extreme foreground - but this doesn't feel natural because our eye muscles "know" we're actually focused on the movie screen?

The biggest problem to me is always when the 3D breaks the edge of the frame - you've just gotta stage things properly so that there are no bright foreground objects doing that, I guess! Although I'd be tempted to do a test with "fake" pillarbox black bars on the left and right 10% of the screen (which objects are in rare cases allowed to break). Don't think that'd work too well though.

BTW, what shutter angle are you guys using nowadays? For CG I did many tests with different shutter angles and usually ended up in the range of 60-120 degrees shutter. I know we need to move to 48fps, but oh man is that going to suck for render times...

Bruce Allen
www.boacinema.com

 

[Michael Hastings]

This is such a great thread!

I agree with all of you in different ways - Pedro, personally I think you're dead on with everything you say about interocular. Disagree with you RE shooting everything parallel and doing convergence in post though! I'm on the Cameron side of things there
Bruce Allen
www.boacinema.com

I think everyone is on Cameron's side when you talk high level production. In high budget projects where everything is well planned and you know exactly what you are going for then manipulating all of those things with a proper rig is important and worthwhile - probably essential.

The point of shooting parallel and converge in post is that for learning purposes or simple/low budget projects it is probably more straightforward and may be less likely to create bigger problems.

 

[Tim Whitcomb]

As if there hasn't been enough discussion of variable IO, I would like to add:

Consider that we refer to a lens as normal when it yields a perspective similar to human vision. For example, in the 35mm motion format we call 50mm normal and in the 35mm stills format we consider 75mm normal. These two lens/format combinations give us roughly a 27⁰ FOV which when coupled with a 65mm IO can work very well in creating pleasing stereo images. As the FOV gets wider the IO needs to get smaller to maintain a natural look without miniaturization. Conversely as the FOV gets narrower the IO will generally need to increase to maintain a more natural look while avoiding the cardboard cutout quality.

Now I know this is just one consideration in figuring out the correct IO but Pedro, Markus and others have covered a lot of ground and passed on a lot of really excellent hard earned knowledge. We at ET are engineers and technicians, not filmmakers. As a result we have been around an awful lot of very talented stereographers and stereo supervisors. Through this we have seen many styles of calculating proper IO+C. Some conservative and some radical. In all of this I would say that there is no single IO distance that seems to be used more than any other. There is certainly nothing special about a 65mm IO distance unless you also have the magic FOV of 27⁰. Even then; subject distance, negative parallax in the deep background or even screen size will force you to change the IO to something other than 65mm.

I would also like to add that Pedro's suggestion of experimentation is right on the money and well worth trying. If you're here on this forum then it is safe to assume you like to make images. You will have a blast making stereo images and learn so much, so quickly.


Stephen Pizzo
Technica3D.com

any chance I can get a message or call back from ET regarding availability of Rigs? No offense meant, but ive left messages and sent emails for a month and have had NO REPLY...

hoping posting here will get better results. thanks!

 

[Joey R.]

Tim, my apologies but this is the first I am hearing about you trying to get a hold of one of us. I will PM you my cell phone number so I can be reached directly.

Also as a side note, we have doubled our square footage and now have a dedicated facility for Engineering, Admin and demos. The move would have made getting a hold or us difficult for the past two weeks.

Joey

 

[Tim Whitcomb]

moving, totally makes sense. Ill contact you shorttly.

everyone else, thanks for such an amazingly helpful thread...

 

[Ketch Rossi]

The guys over at ET, surely do know what they are doing, and love the space available in their new facility, was great seen Stephen, Hector, and getting to know the rest of the crew, Chris is also very knowledgeable 3D Tech and Stereographer, and Joey and I spent a considerable amount of time on the ET 3D rig.

I have always considered ET to build excellent gear of precise and high quality, and after my recent visit I found the decision to go with their 3D rig for my project, a simple one, as the rigs are extremely well built, and have great electronic and remote controls, even so we will be shooting some scenes on the current large rig with two R1' we very much look forward to their Medium size rigs for the Epic and Scarlet.

Joey, was a pleasure discussing my project with you, and look forward in our tests as soon as you guys are set up for it, you guys rock, and any one interested here, and serious about any 3D work should not hesitate in contacting ET, even so they are now many 3D rigs out there and some cheaper then others, here you do get what you pay for, and 3D is an extremely complex animal, even in its simplicity is still complex, and if you don't have the right rig, you really are up to putting your self in more trouble then is worth.

 

[Charles Angus]

There's is absolutely no variability in a 3D rig except for the angle at which the two cameras converge. The inter-occular distance is always 65mm. It doesn't matter which format of camera you use, from IMAX to super8mm and everything in between, the inter-occular distance is 65mm because that matches the average position of the left and right human eye relative to each other.

From everything I know, this is not correct. Large interocular distances can be used for distant subjects, to increase the 3d effect, or to cause the subject to look smaller than it is. Vice versa for small interocular distances.

It is also not necessarily desirable to converge the cameras. Convergence can be adjusted more precisely in post (it's easy, too - you just slide one image left or right relative to the other). Shooting with the cameras parallel (ie no convergence) also has the benefits of not keystoning with wide lenses and making post compositing easier (largely because of keystoning, but it also makes it easier to adjust convergence as you are not fighting what was done on set). I know this from experience compositing on 3d stereoscopic projects.

I would personally only ever shoot parallel 3d...