When small interaxials just don't cut it! - Page 5

I think you're right, but for most of the footage I've shot it hasn't been a problem. In my Missouri Botanical Garden piece only a handful of shots have been unusable. I can get around the issue with virtually anything I shoot by converting the JVC footage to 720p, but the resolution drop is too severe. I should have that 3D Blu-ray disc (that I promised a while back) in the mail for you tomorrow. There are a couple of shots I'll probably have to cut because of the stuttering, but for the vast majority I think it's almost completely unnoticeable. Certainly, the resolution drop going to 720p results in a 3D Blu-ray that looks a lot worse. What I'm hoping for is an extension to the 3D Blu-ray spec that allows us to keep the footage in its native 60i format. 60i 3D straight from the JVC camcorder to ANY of my 3D TVs looks phenomenal. Going to 24p 3D Blu-ray is what causes me grief sometimes.

No argument from me. I've actually been sold on wider interaxials since some of the first videos around your house. Even there, the advantage was obvious to me.

Thanks, Frank. That's going to be the most relevant for me anyway. No hurry.

Was there ever a question on distant large stage that a wide IA is necessary?
The question was that originally you were of the opinion that extreme wide was good for everything. Or, that shooting in a small room of activity that the JVC or Sony wouldn't work.

I see the need for 4 types of cameras for 3D on wide angle.
1. The Bloggie3D for typical stereo at distances of 2-6 feet.
2. The SonyTD10 / JVC TD1 for scenes between 4 ft and 15 ft stage depth
3. The sport cams ( IA 75mm ) for distances between 10ft and 50Ft
4. The dual cams on a bench for shooting wide angles of landscapes and large stages at great distances. THis variety can be mounted with an under-through glass to allow for one rig to do close as well.

Each will have it's advantages and disadvantages but in all cases the dual cam system that is adjustable can work for all stages including very small subject to very large. What it doesn't offer is portability and ability to look "normal" in what society expects. In other words, as we go from system 1 to 4 we draw increasing attention to what we are doing. This usually is not a desirable feature.


As for the Tall ships video, I couldn't resist so I too downloaded your left and right clips and the MP3 and rendered out the best settings and burned to a BD-RE. What a difference in quality on the Passive monitor. Every line of rigging on the ship was fine and without jaggies. The color was very good as is and it plays very smooth. I rendered at the 1080 24p 25Mbit so the only motion blur I saw was when the ship rocked I saw some blur on the rigging. I am really pleased at your success on using the Gyro. It doesn't look too cumbersome either. Most of my future projects would not require a Gyro on a bench system as a tripod would be best. But now I know the idea does work.

"The SonyTD10 / JVC TD1 for scenes between 4 ft and 15 ft stage depth."

Don, this ignores, does it not, the zoom lens of the TD10? The "window" you give is relevant only for the extreme wide angle of the TD10. At full extended zoom (10X), the window (in theory) is 25-65 feet, right? And something in-between for middle zoom lengths. The TD10 is much more flexible on window than 4-15 ft.

Everyone seems to ignore the zoom lens, although didn't you give an example of how zoom affects depth?

I don't think there's anything close to a 1:1 relationship between the depth and the zoom on these small IA camcorders. IMO, as I zoom in with the narrow interaxial of my TD1 everything flattens out dramatically. There's not a lot of 3D effect left at max zoom, and the TD1 won't zoom in nearly as much as the TD10. My contention, since I got my JVC, has been that if I frame carefully, with objects in the foreground, middle ground and background, the sense of depth can be pleasing, especially if I'm at max wide angle. But, it doesn't have the same look and feel as the wider interaxial of Frank's Canon rig. That's why I asked Frank to do the test, so we'd have a couple of objective clips that could demonstrate that difference. I've seen it in several of his earlier videos that he posted in the TD1 thread months ago, but setting up the cameras so as to eliminate other variables should make the differences clear to all of us.

1. "There's not a lot of 3D effect left at max zoom, ..."

2. "My contention, since I got my JVC, has been that if I frame carefully, with objects in the foreground, middle ground and background, the sense of depth can be pleasing, especially if I'm at max wide angle. But, it doesn't have the same look and feel as the wider interaxial of Frank's Canon rig."

I disagree with 1. That is not my experience, and I have plenty of experience with the zoom lens of the TD10. You have to zoom in on distant subjects to get a 3D effect, but when you do you can even get pop-out. I have uploaded videos to show this, and I think Don has one example as well. Perhaps the issue is whether you get conspicuous 3D effects at, say 65 feet, at extended zoom (you do), or whether you get 3D effects that have a 25-65 foot window at that zoom.

I agree with 2 completely.

For me, the 3D effect when zoomed in with a narrow interaxial camera is unsatisfying. It's like a pop-up book. Telephoto lenses have the effect even in 2D photography of compressing space front to back. Even though I can tell what's closer and what's further away, the effect is more cardboard cutout. Frank has some shots taken with the TD1 in which tree trunks look flat, not round. You can tell which trees are in front of other trees, but not much looks natural. I avoid zooming in when I shoot 3D with my JVC.

Agreed, Mark. But reread my post. I was addressing wide angle.



Joe- have a look at Post #82 this thread. Get your red / cyan glasses.

Yes, there's a good sense of depth in that shot. It's hard for me to tell what the different "layers" are like from that image, though. I'd really like to run some of the tests I suggested Frank do, but with a Sony and JVC. I may have the chance, since a buddy of mine who runs a local high school media program intends to get the Sony this fall when school starts. That's less than a month and a half away. I'll be sure to post the results here on AVS if I get that opportunity.

Still, I know the "feeling" I get when I see 3D shot with Frank's Canon rig and I really like it. I don't think the Sony or the JVC could ever duplicate that sense of space front to back, no matter the zoom setting. But that's why I asked him to perform the test. I think I know what I'm going to see, but I won't know for sure until I see it.

Joe- If you're looking for a scientific explanation why you get that "feeling" I told you to go get the book "3D Movie Making" with the included DVD and watch the DVD with the Autodesk 3DMax Demos section showing how all these relationships affect each other. Mendiburo tries to explain it in the book on paper but until I saw the explanation in a 3D wireframe drawing in animation it wasn't really clear.

In a simple "trust me this is why" explanation of your feeling it has to do with the fact that what Frank or anyone gets with his wide IA is an advantage with wide angle of view. The same z depth can be obtained with closer IA but only with a sacrifice of angle of view ( that is zoomed in). Additionally, the proximity of what your mind sees as what the physics suggest you should see is accurate to the extent of your vision accuracy. Imperfections in your vision can affect the Z depth in the scene just as imperfections in the twin camera alignments shooting the scene can affect the Z depth. The physics only works accurately assuming both camera alignment AND your eyeballs are accurately aligned. This alignment includes not just camera position but also focus and zoom setting, not to leave out the cameras themselves must be a match.

I think anyone really serious about doing 3D needs to get a good background education in what the science already knows. Casual 3D hobbyists can go through life fascinated with the magic of 3D but if you really want to DO 3D you need to reduce the magic to science and that is done by learning how people have done it before you.

I have several books here on the science of 3D, but for one book to really cover it all I recommend 3D Moviemaking by Bernard Mendiburu. The others cover it too but you'll spend more money getting all the volumes and in the end learn the same thing. The only negative about this book is the author is French and sometimes the language is a little difficult and it makes you read a couple times to be sure you didn't misunderstand the English but he has lots of drawings as opposed to high math. If you prefer pure optics physics and lots of high math this is probably not the best book. Even though I have had enough math as an engineer, optics is not my field so I prefer the drawings for explanation.

In another observation on Frank's demonstration, I want to specifically target the subject of "Popout" or negative Parallax where the image is between you and the monitor surface.

The one thing I have not been able to successfully achieve ( that I recall) is a popout when shooting extreme distances zoomed in with the TD10. I have not tried to create this effect intentionally but have shot lots of scenes from distance requiring full zoom. In Frank's ships the popout is quite obvious at times. Mark- you claimed you have gotten full zoom popout. I need to see that. By the physics, it should be possible but I just haven't seen it in my shooting. Maybe I was just too far out for my Zoom range. If this is so, that might also be an explanation why the JVC doesn't achieve as good a Z depth at distance zoomed in because it has only ( I recall) a 5 to 1 zoom while TD10 has a 10:1 zoom (in 3D mode)

I'll get around to reading more about 3D, but right now I almost prefer to discover it for myself (or beg Frank to do an experiment or three ). I saw the cardboard cutout effect with the JVC before I owned it, because I asked Frank to zoom in as an experiment with the JVC months ago. I didn't know what it was called, but there it was, plain as day. That's why I don't zoom in often when I shoot 3D.

For now, as I learn, I'm relatively happy with the JVC. My 3D "box" is much smaller than Frank's, because of the narrow interaxial, so I adapt and learn to live within the limitations. Although the IA is a serious drawback, I love the image quality.

"Mark- you claimed you have gotten full zoom popout. I need to see that."

Butterfly on tree, full zoom, at about 1:43.

http://youtu.be/tdhsOcDpDeY?hd=1

Longer lenses flatten the 3D image (cardboarding): http://www.captain3d.com/temp/cml/cml_volume.html

It's related to the projection equations:

3D Projection:
xL = D*(x - O)/z
xR = D*(x + O)/z

2D Projection:
x = D*x/z

As D increases (zoom), the relative contribution of O (camera offset/separation) decreases. Similarly, as z increases (distance from camera to object), the same thing happens. The difference between xL and xR diminishes as zoom and distance from the camera increase. Thus, at the limit the projection of xL and xR converge on x as D and/or z approach infinity (resulting in a completely flat, 2D image). When filming far away objects and using a zoom lens, increasing the camera separation will unflatten the 3D image as the relative contribution of O increases.

When depth mapping tools become more widely available, it will be possible to shoot close up objects with (somewhat) larger separations and then remap the 3D volume in post to provide a comfortable final view (flatten the space in the foreground and background as necessary (+ not necessarily linearly)). Sophisticated image processing is required to round out a flat/cardboarded 3D image (requires 2D to 3D technologies).

"Originally Posted by markr041
"Mark- you claimed you have gotten full zoom popout. I need to see that."

Butterfly on tree, full zoom, at about 1:43.

http://youtu.be/tdhsOcDpDeY?hd=1

That's a great example of the Card-Board Effect Yikes!

Making objects "pop out" is easy. Using Vegas you can pop out or in whatever you please. I do it all the time."

The problem with your approach is that you do not recognize that with your rig there is no butterfly 3D video at all:

1. You could not haul your gear up that mountain trail in 90+ degree weather. Even if you left behind the gyro.

2. Even if you did, and survived, by the time you set up your rig with the "correct" interaxial distance, there is no butterfly. bye, bye butterfly.

In fact, for most interesting video, long-distance vista shots are a minor adornment. Look at any film, tv show, documentary. No one is going to watch unending vista shots. Up-close people shots are what makes a video. People playing, reacting, interacting, doing something. Don's kid in the pool is still among the best 3D video I have seen - it has a kid enjoying and learning and interacting with parents/grandparents with great 3D and color and action. It's human. 3D Bloggie. A nice 3D vista shot of the pool by the house with the big rig might have added something (about a 6-second shot), but that's it.

The small interaxial distance of the all-in-one cameras is the correct choice - it's for taking good video of people, not for specialized vista shots from power boats, which would constitute a trivial (but impressive) aspect of any video that a person would actually want to watch.

I do not think it is a coincidence we have not seen a video of any length - beyond "tests" - from this touted rig.

It's been fun to see all this, [I appreciate the work and expense, and I learned a lot about 3D. But let's get real.

John- did you look at some of those "experiments" he did? There appeared to be something radically wrong. After carefully studying I figured out what it was.
As you zoom in the background goes tighter ( correct) but the foreground goes wider. (incorrect) The only way this can be possible is if the camera is on a dolly and is pulled back as the lens zooms in.

The treatise is nice if only it was correct. Am I missing something?

The top data is an interesting trick. But he demonstrates fantastic roundness due to the dimple pattern on the balls which can give it better roundness as it places a pattern of anaglyphic disparity over the surface. Roundness in 2D is only achieved by shadowing and highlights. How would the golf balls look in the same experiment as round balls with a smooth surface. I cry a bit if illusion trickery in this experiment.

Without further study, I believe this nicely written piece is full of holes and needs further study.

Rather than draw scenes that support a theory, how about seeing a real world example with real cameras. I believe any test would demonstrate what Mark and I have been seeing in our shooting and what we have seen in Franks shooting.


Well, I have to put these threads to bed for a few days, I'm off on a shooting venture. Some will only be in 2D as those are just speaker presentations. I'll try to get in some 3D shooting as well.

2D - apostate!

Good!

Mark- I looked at your link and saw no popout of any distant scene. It looked similar to the depth I get here. The popouts I did see were of objects in the scene that were close in camera to subject distance and this is not what I was getting at. Try to get a popout of an object that is 50-150 ft away fully zoomed in. Everything I have done to date has been good 3D depth but all from screen plane to behind screen. Still good 3D but no real popout. I haven't played around with the 3D stereographic effect yet in Vegas to do this. Only have used it to position graphics in Z axis. See my YT video of the Liberty singers for example of that technique. BTW- Thanks for the compliments of the Pool video.

What Frank has demonstrated is some great stereo popout to great depth using his approximately 9" IA cameras. I agree with you that the rig is cumbersome but some of us have experience with working with bigger equipment. Not for amateurs, that's for sure, but in my younger day I have shot betacam in an underwater cave system past serious restrictions, up to a mile in an underwater cave. But as I got older I couldn't do the multi stage mix gas dives pushing 250 pounds of gear for a two mile swim against the current then decompress for 9-15 hours. Those days are gone! I'm impressed Frank is able to do what he does. I just hope he doesn't over do it. His rig will be required to achieve my dream video of Bryce canyon. I don't think the trek with his gear will be difficult physically.

Don , that butterfly was 40+ feet away. With a fixed wide-angle lens no pop out is possible at that distance. Of course you are not going to have pop out from 100+ feet away with the TD10! The point is the zoom capability gives you a window for popout that extends beyond the 15 feet you mentioned as the (correct) wide-angle limit, as this video clip shows.

And with that fixed, wide-angle lens, no way that butterfly hangs around when you approach it at 3 feet to get pop out. Not even Mr. Titanic gets that shot (without cgi, of course).