interaxial distance for landscape

[rablaw 3d]

Hello,

I intend to shoot a 3d video (buildings in my city) with 2 go-pro hero 6 (SBS) attached to a drone.

The cam settings are 2.7k Linear FOV.

I tried 115 millimeters distance between the 2 cameras and I can see fair 3d image on the screen (projector in the room - 2 meters wide), not to mention I changed the disparity on screen in post. I need the video to be in positive disparity ( all behind screen, so the screen will act as a window)

The closest object will be 15 meters and the furthest will be 100 meters.

questions:

I wonder what should be the interaxial distance to feel more depth on the screen?

what do I need to project it on large screen in theater with 40 seats ?

Should I keep the current distance (115 millimeters) and the image will look deeper on theater screen?

I appreciate it if anyone can help.

 

[threed123]

To get all of the image behind the screen, the ending separation between the image objects that are furthest way, should be around 2.5" apart and the nearest have no separation--they will be right at the window. That said, I don't think you have to worry as with a drone, you can't separate the cameras far enough anyway to create a depth issue. It's probably trial and error. When the cameras are separated more than normal eye separation, along with the wide angle of the lenses, you will get a miniaturization effect. So you have to consider that as well. If the separation of objects on your 2 meter screen are further apart than 2.5", your eyes will turn out slightly creating eye strain especially if you are sitting within 10-20 feet of the screen. Sitting further back will elongate the depth as well. Sitting closer will compress the apparent depth.

There are mathematical equations that can be used, but you really don't know what you are dealing with given the wide angle of the gopro lenses regardless of what FOV you have it set at. Sooo...I think you just have to experiment until you get it to look right. Have others see the results as well. What's good for you, might not be good for others, especially children, whose eyes can be closer together.

You don't have to drone your cameras to get this right. You could stand on a hill or in a two story building and take some landscape video and try different separations--of course you have to have objects between you and infinity to make this work properly.

 

[Gerry Siegel]

My take, and I do not have much experience with hyperstereos of distant objects is to get much farther than the 115 mm you are using. I think the limit will be with the bracketing and fixtures needed to attach same to the drone. Without a diagram, it is hard for me to visualize. You need a lot more spacing.

How much depends subjectively on the effect you are after. Me I like hypers to be real hypers, ignoring any distortion of size ...it is after all an illusion. I could be wrong. Let us know what you find. In other words, report back and maybe reproduce a frame or two... Aloha,gs

 

[3D Gunner]

25-30 cm (250-300 mm) interaxial distance would be OK for projection on a big screen (200"-300"/5-7,5m diagonal screen), for impressive 3D effect.

 

[Gerry Siegel]

As usual when these questions discuss projection, and to a large audience, I am curious to ask what kind of video projector will you use to get requisite illumination? Or are you still seeking advice on what to buy in that respect? Viewing is still my area of most ignorance I freely admit... I know about commercial theatr gear, but is there a consumer level goodie out there..with polarizers etc

 

[rablaw 3d]

To get all of the image behind the screen, the ending separation between the image objects that are furthest way, should be around 2.5" apart and the nearest have no separation--they will be right at the window. That said, I don't think you have to worry as with a drone, you can't separate the cameras far enough anyway to create a depth issue. It's probably trial and error. When the cameras are separated more than normal eye separation, along with the wide angle of the lenses, you will get a miniaturization effect. So you have to consider that as well. If the separation of objects on your 2 meter screen are further apart than 2.5", your eyes will turn out slightly creating eye strain especially if you are sitting within 10-20 feet of the screen. Sitting further back will elongate the depth as well. Sitting closer will compress the apparent depth.

There are mathematical equations that can be used, but you really don't know what you are dealing with given the wide angle of the gopro lenses regardless of what FOV you have it set at. Sooo...I think you just have to experiment until you get it to look right. Have others see the results as well. What's good for you, might not be good for others, especially children, whose eyes can be closer together.

You don't have to drone your cameras to get this right. You could stand on a hill or in a two story building and take some landscape video and try different separations--of course you have to have objects between you and infinity to make this work properly.

No its not further apart than 2.5" but I wonder how will it look like on large theater screen.

I am trying to make flying simulator, and it hard to calculate the size of screen for 40 seats, and how far the first line and last will be from the screen. since I have to know while shooting before its too late to fix.

As for the interaxial distance, I can reach 220mm(not easy) and still fly and shoot.

I want the viewers to feel its real while thinking about miniaturization on one side and deep 3d on the other.

 

[rablaw 3d]

25-30 cm (250-300 mm) interaxial distance would be OK for projection on a big screen (200"-300"/5-7,5m diagonal screen), for impressive 3D effect.

Thanks

Do you know what should be the viewers distance from the screen in this case?

we are talking about flying simulator.

 

[rablaw 3d]

My take, and I do not have much experience with hyperstereos of distant objects is to get much farther than the 115 mm you are using. I think the limit will be with the bracketing and fixtures needed to attach same to the drone. Without a diagram, it is hard for me to visualize. You need a lot more spacing.

How much depends subjectively on the effect you are after. Me I like hypers to be real hypers, ignoring any distortion of size ...it is after all an illusion. I could be wrong. Let us know what you find. In other words, report back and maybe reproduce a frame or two... Aloha,gs

I can reach 220mm - 250mm with 3d printed rig and some gopro mounts. Actually thats the hardest part. I will post a video sample, but it needs to be seen on a big screen to get the right impression.

 

[3D Gunner]

For a flying simulator you will need a panoramic screen and high resolution, 4K/eye by example. The "immersion" is better if the screen is larger and the viewer is closer.

In my opinion, the optimal immersion for 4K projected image is situates at 0,8d (d=diagonal lenght) for 16/9 screen format, which is almost the limit in which you will not see individual pixels (in 4K resolution).

I think is not a good idea to make a flying simulator in a clasic cinema environment, because only few persons which are situated close to best viewing point will benefit the best enjoinment.
So, a VR system vould be better, but also much expensive if you intend to make a show for 40 people simultaneously.

..........................................................

As for expected costs:
- You can start with ~6000$ (for viewing system only) for a cheap dual 4k projector formula, with passive interference glasses (Dolby 3D, Infitec), 150"-170" diagonal screen and moderate brightened image, to impress some friends or other non comercial aplications...
- To ~3000$/custom video player and HMD/client, for comercial applications, plus high additional costs for a good video production.
(only as an idea)

 

[Turbguy1]

I may be repeating the words of others...

You must use care to ensure the farthest (probably infinity) subject elements are no farther separated ("disparity") on the screen by the average inter-ocular distance (65mm or so). This avoids eye divergence that can cause eyestrain for some of an adult audience while they attempt to fuse the scene. As mentioned by others, children may still have an issue, as their eyes will be closer together.

This mean that projection size must be considered. Disparity separation adjustment is available in software, but will also effect the z-position of closer objects, so that they may appear in front of the screen (a potential "window violation"). This adjustment may also change the aspect ratio. And, some audience seats will just have better viewing than others. The farther away they sit, the deeper the scene will appear.

Using the typical 1/30th rule for interaxial distance for closest element in a scene that also includes infinity, your separation should be 1/2 a meter. YMMV.

Test, test, test with multiple viewers, too.

 

[3D Gunner]

Using the typical 1/30th rule for interaxial distance for closest element in a scene that also includes infinity, your separation should be 1/2 a meter. YMMV.

Test, test, test with multiple viewers, too.

Typical 1/30th rule is not working for this project. He will need to adopt 1/50 to 1/60 rule for this project, if intend to put the entire scene beyond the stereoscopic window on a large screen.
Mathematics first, then tests.

 

[threed123]

My take, and I do not have much experience with hyperstereos of distant objects is to get much farther than the 115 mm you are using. I think the limit will be with the bracketing and fixtures needed to attach same to the drone. Without a diagram, it is hard for me to visualize. You need a lot more spacing.

How much depends subjectively on the effect you are after. Me I like hypers to be real hypers, ignoring any distortion of size ...it is after all an illusion. I could be wrong. Let us know what you find. In other words, report back and maybe reproduce a frame or two... Aloha,gs

I can reach 220mm - 250mm with 3d printed rig and some gopro mounts. Actually thats the hardest part. I will post a video sample, but it needs to be seen on a big screen to get the right impression.

I have a 160" screen and will take a look at any videos you have. It will be difficult to post a video here, though because of size. You could try dropbox or similar with a link to it.

For a flight sim, you shouldn't have close up objects anyway (or you'd be in crash dive...;o). But that brings up the question of how do you simulate the cockpit? Again, I'd be happy to help review whatever you have on my screen. What format will your video be in? SBS? T/B? Frame-packed? Others have stated that for realism, you need 4K video, but that would involved using two stack 4K projectors, one for each eye, polarized. I think 1080p is good enough, though. How big of a screen are you trying to get to? You'd need a powerful 3D light-canon projector for a large theater environment, and they are not cheap.

What 3D editing program do you intend to use? Syncing up Hero 6 video is not going to be straightforward. You will need to create a sync point, which can be done by using a clicker and hold it near the cameras just before the flight to create a noise spike. The frames will go off slightly by probably a half or full frame as the flight continues. This can cause eyestrain or a weird 3D effect, so you will need to resync the frames throughout the video. Maybe have a portable beeper on board to create sync points. You'll need a good editing program with a frame-level stereo alignment feature, of course--such as Edius.

Also, post your request in this forum: https://www.avsforum.com/forum/192-3d-source-components/ There is a ton of experience using gopros and other cameras there for 3D.

 

[Turbguy1]

As I said, YMMV...

The 1/30th rule works just fine to keep the closest 15 meter distant object AT the window's edges, while keeping infinity (and everything else) behind the window.

As for controlling on-screen deviation, that requires testing with multiple viewing subjects (the audience).

That said, 1/50th will "work"...

--
Wayne
See more at:
http://www.flickr.com/photos/turbguy/

 

[3D Gunner]

As I said, YMMV...

The 1/30th rule works just fine to keep the closest 15 meter distant object AT the window's edges, while keeping infinity (and everything else) behind the window.

As for controlling on-screen deviation, that requires testing with multiple viewing subjects (the audience).

That said, 1/50th will "work"...

The 1/30th rule don't work for subjects between 15 m (at stereoscopic window plane) and others at very long distances, on large screens, because will generate immense positive parallax for the distant objects.

................................................

I tested the attached Red-Cyan image on 180"/457 cm (diagonal)/~400 cm wide screen (4K projector), where the farthest area produces a positive parallax of ~21.5 cm(!), but I can see (and my wife confirmed) all the contents of the image without any problems, an impressive depth.
The interaxial distance is in the ratio of ~1/100, from the foreground wall (which is at ~68 m distance from the camera).
For the project to fit within the "rules", an interaxial distance of 25 cm would be OK, with the statement that in the case of aerial images, very distant elements are inherent, and the 3D effect will be much diminished vs in the attached image. I would go on a more aggressive solution, 30-35 cm interaxial, even if the stereoscopic window could be violated and a positive parallax greater than recommended should be accepted for objects situated at long distance.

A HMD system can easily handle very deep images, on the fact that it is not a major problem if the stereoscopic window, which is located at the limit of the peripheral vision, is violated.
Can check with the attached image on a smartphone with high resolution screen (Sony XZ2 Premium with 4k screen is recommended) and an HMD.

 

[rablaw 3d]

As I said, YMMV...

The 1/30th rule works just fine to keep the closest 15 meter distant object AT the window's edges, while keeping infinity (and everything else) behind the window.

As for controlling on-screen deviation, that requires testing with multiple viewing subjects (the audience).

That said, 1/50th will "work"...

The 1/30th rule don't work for subjects between 15 m (at stereoscopic window plane) and others at very long distances, on large screens, because will generate immense positive parallax for the distant objects.

................................................

I tested the attached Red-Cyan image on 180"/457 cm (diagonal)/~400 cm wide screen (4K projector), where the farthest area produces a positive parallax of ~21.5 cm(!), but I can see (and my wife confirmed) all the contents of the image without any problems, an impressive depth.
The interaxial distance is in the ratio of ~1/100, from the foreground wall (which is at ~68 m distance from the camera).
For the project to fit within the "rules", an interaxial distance of 25 cm would be OK, with the statement that in the case of aerial images, very distant elements are inherent, and the 3D effect will be much diminished vs in the attached image. I would go on a more aggressive solution, 30-35 cm interaxial, even if the stereoscopic window could be violated and a positive parallax greater than recommended should be accepted for objects situated at long distance.

A HMD system can easily handle very deep images, on the fact that it is not a major problem if the stereoscopic window, which is located at the limit of the peripheral vision, is violated.
Can check with the attached image on a smartphone with high resolution screen (Sony XZ2 Premium with 4k screen is recommended) and an HMD.

Thanks for helping

For 30-35 cm interaxial I will ned to build a wing for the drone ;-)

Anyway I will deal with the violation in post ( by masking)

What interaxial did you use for the attached image?

 

[3D Gunner]

About 65 cm.
So, 35 cm interaxial will significantly reduce the depth for the entire scene from same position.
But if you will have objects closer to the drone, at 15 m by example, you will gain big depth at closer distances (between 15 and 100 meters) as initially mentioned.
Tests can be done with pictures, no need videos in the first stage.

The attached picture is a crop from a larger scene, made with 35 cm distance between cameras. I isolated an area where you can see objects from ~8-10 m to the cameras, up to ~100 m in the farthest point. In this configuration, objects situated at 90 m from the cameras generate ~9 cm positive parallax on 4 m width screen, and ~13,5 cm positive parallax on 6 m width screen. More distant objects would generate more positive parallax.

Objects at ~10 m, ~15 m and ~90 m are marked in the picture, for reference.

 

[relaxman]

What 3 axis gimbal do you have for 20-30cm base distance?

 

[rablaw 3d]

What 3 axis gimbal do you have for 20-30cm base distance?

I had to attach DSLR gimbals for this project. and still not quite working.

 

[rablaw 3d]

Hello,

I started testing few days ago.. and this is a video sample.


Please watch it (anyone who has big screen) and tell me what you think.


The quality is not the best and excuse the vibration in the video (still need to adjust the rubber damper) .


The distance between the cameras is 28cm.


You can download the video from youtube:

 

[rablaw 3d]

Here is dropbox link:

Loading…

www.dropbox.com