Set on Fire

Play of lights on Epsomite crystals. Specimen is about 4.5 cm wide. (cross-eyes view)

Interesting view.  Feels rather flat though.  Which interpupillary distance did you use?

Cheers.

Ron

About 3 cm at ~35cm distance from camera sensor.
At me is not flat at all.

Tourlou wrote:

Interesting view. Feels rather flat though. Which interpupillary distance did you use?

Cheers.

Ron

Plenty of depth for my eyes.

Three-dimensional perception is built into the brain, and the ability for three-dimensional separation is different in different people.

More than 10 years ago I prepared a three-dimensional test for a medical company, which among other things also identifies three-dimensional separation ability - the minimum difference in the disposition of details in images corresponding to the two eyes, measured in minutes of arc needed to perceive the depth disposition of objects.🙂

With your specs, I was able to reproduce, at scale, your pictures for my interpupillary distance.  Seen on the big screen, it looked a lot better.

That's what I call a lot of depth...

In any situation it is about compromise.
- There is the basic rule of thumb with the "standard specification" of 1/30.
"The 1/30 Rule refers to that the Stereo Base or separation between lenses (from center to center) should be no more than 1/30 the distance of the NEAREST object to the camera lenses." This is good for life-size objects viewed on medium to large screens (3D TVs to cinema screens).
- For normal size PC monitors a larger stereo base will increase the perceived depth.
- For same stereo base, the larger the 3D picture will increase the depth.
- For small things, a larger than 1/30 stereo base is recommended (1/25-1/15).
- For macro work, even larger stereo base is recommended (1/15-1/10).

So, your picture with "my specs" (1/12) is very deep and have a window violation. Without window violation, it have "out of normal depth" and can be viewed comfortable on a 3D screen up to 49"/125cm wide. At ~49" wide, the depth will be from the screen plane to infinity.
With screen violation can be viewed on any screen size, but not comfortable and not by everyone (because the subject have a limited depth in reality).

My image can be viewed on screens up to ~100"/250cm wide as it is (without window violation), because it have a small depth and no background. To be properly seen on my 134"/340cm wide projector screen, will need to accept some window violation for the crystal at the base.

So, my "specs" for close-up/macro shots are not good for life sized 3D subjects.
Starting some time ago, I prepare my 3D pictures to be viewed on VR devices, which are equivalent to medium/large screens, so on small screens can be perceived as to have no great depth.🙂

no window violation

Trimming the edges to prevent window violation makes for a neat result!  Thanks.  I'll be less lazy next time!

Where did you take all this information relative to screen width?  I'd like to read about it.  I built 3D feedback systems for allowing remote operation of large and complex industrial machines.  My experience doesn't match with your data.  We developed that system over a 2 years period.  I read a lot about visual perception, played a lot with interpupillary distances, vertical plane alignment/disalignment, angles of view of the operator to the TV.  We had one operator who had a narrower that usual interpupillary distance.  A camera center distance wider than 100mm with the optics we had would make him feel dizzy when watching the 47 inches 3D TV.  Having him to sit further from the TV had the same effect as getting the cameras closer together.  So, according to my observations, it has more to do with the angle of view than with the absolute size of the picture.  We tried with kids to have narrower faces too.  To get them interested, we strapped the cameras on a RC truck and put the batteries and broadcasting system on a toy trailer we pinned behind the RC truck.  We had a lot of fun with this one!

Regards.

Ron

It is recommended that the depth of the image (3D) to be within the comfortable limits of eye convergence for each viewer. That is, the closest element in the image should not force the eye to look too close, and the farthest element should not force the eye to look away. Parallel looking is equivalent to looking at the furthest possible point, the Moon for example, or the stars.
Some people can look slightly divergent without undue effort, but this should not be abused.

For the maximum recommended screen sizes I have taken into account the situation where the maximum gap for the most distant elements in the image does not exceed ~7 cm, i.e. I have taken as reference an interpupillary distance of ~7 cm. Medium and large screens are viewed from a reasonably large distance so that slight deviations from the calculations can be tolerated.
So the stereoscopic window is the plane of the screen, which is easy to view being located at a reasonable distance, and the farthest element requires a parallel view for comfortable viewing.
The calculations I made were based on the position of the closest elements placed in the plane of the screen, so they do not violate the stereoscopic window.

You can actually measure the width of the image observed on the screen (for one eye!) adjusted so that the stereoscopic window is not crossed and you can also measure the distance between the furthest points in the image, for the two eyes.

In the case of the image taken by you, corrected so as not to violate the stereoscopic window, displayed on the screen at a width of ~50cm (for one eye), the offset between the furthest objects is ~2.8cm. For an inter-eye distance of 7cm, results 7:2.8=2.5. So the 50cm wide image can be enlarged to ~50x2.5=~125cm wide, so that the offset of the farthest points reaches 7cm=the interpupillary distance taken as an example, resulting in a parallel view.

The entire depth range can be brought forward by breaking the stereoscopic window (this procedure is used in cinema projections), and the image may contain elements far in front of the stereoscopic window, which may not be intersected because of their location, etc...

I work for some projects involving VR, in which case it has to be calculated in advance how all the elements of the image should be placed in depth, so that the viewer's gaze is not forced, regardless of personal interpupillary distance.🙂

Well, we're on the same page!  Thanks.  Now it makes more sense to me.

Regards.

Ron