Stereo pair of a female lone star tick

Graham46 · Post by **Graham46** » Wed Jul 22, 2009 6:22 am

A post like Rik's HERE from the other day makes it easy to appreciate the added value of seeing in stereo. It makes it much easier to see the morphology of the insect, and can make certain areas leap right out that may have gone unnoticed before. Yesterday I shot a stereo pair of a female lone star tick, Amblyomma americanum. Give it a try and let me know if it works for you. Feel free to share some things you saw in stereo that you didnt when looking at the 2D image.

Both images shot with canon 65mm MP-E at 3x. ISO 100, 1/125 @ f/5.6. 34 and 32 images stacked in ZS Pmax and photoshopped.

AndrewC · Post by **AndrewC** » Wed Jul 22, 2009 7:01 am

Weird, that one snaps right in for me but I still can't get Rik's to merge

Andrew

rjlittlefield · Post by **rjlittlefield** » Wed Jul 22, 2009 8:16 am

Graham46 wrote:Give it a try and let me know if it works for you. Feel free to share some things you saw in stereo that you didnt when looking at the 2D image.

Very nice. I notice the concave underside of the body -- hadn't expected that and can't really tell without the stereo.

One nit: there are a couple of distracting reflections off the bases of the legs on the photo's right side.

It's impossible to avoid those with a shiny subject. They would look just the same in real life. The difference in real life is that you the viewer would move something (subject, head, lighting) just a little bit to tell for sure that it's a reflection and not something else.

For the best possible static stereo, you might think about cloning out the differences in the reflection. In this case, there are only a couple of small problem areas, so it would go pretty quick.

AndrewC wrote:Weird, that one snaps right in for me but I still can't get Rik's to merge

That is weird. They're exactly the same separation. My wasp mandible is coarser structured and has more depth, front-to-back. Those might be throwing you off.

But my guess would be that something about the composition is making you tilt your head just a little bit. That introduces vertical disparity and makes fusion practically impossible. The cure is to pick one easily identified feature, find that feature in each eye's view, then very carefully, slowly, and consciously tilt your head to allow fusing just that one specific feature. As soon as you get that one feature locked, everything else will pop in too.

I had to make this trick part of my routine practice years ago. As a matter of habit, I generally keep my head tilted a little to one side. For viewing the world, it's no problem. But stereo images won't lock up unless I specifically set my head level.

--Rik

Graham46 · Post by **Graham46** » Wed Jul 22, 2009 8:52 am

Rik,

Indeed there is an area significantly affected by reflections on the right side. These areas always appear to be "shimmering" and make them easy to find and fix. However, in most cases, I leave these areas the way they are. I find that attempting to clone in like parts leaves you with something much more noticable then shimmering: a seperate layer. Tony has suggested that this can be fixed by cloning in an area from one side to the other, then using photoshop's transform tool to skew the cloned area by the same 5 degrees as the original rotation. I have had no success with this method even though in theory it should work.

rjlittlefield · Post by **rjlittlefield** » Wed Jul 22, 2009 9:35 am

Graham,

Yeah, retouching stereo is tricky. If you're cloning from one side to the other, you have to make sure that details from the source land exactly on corresponding details at the destination. Get them just a little bit left or right, and the cloned part moves forward or backward. Clone a large area and the cloned part looks flat. I think this is what you're talking about when you say "a separate layer".

So, the technique that I use (when I bother to use it at all) is to clone with a small brush and to reset the source/destination coordinates at each detail I see.

One difficulty in applying Tony's suggestion is that the proper amount of correction depends not just on the 5 degrees of viewing angle between images, but also on the orientation of the surface with respect to the line of sight. The correction is cos(angle#1)/cos(angle#2), where the angles are deviations from perpendicular. If the surface happens to be tipped 2.5 degrees with respect to one eye and -2.5 degrees with respect to the other, then no correction is required. But if it's 15 degrees with respect to one and 20 with respect to the other, then some correction is required. The correction is typically not large, for example cos(15 degrees)/cos(20 degrees) = 1.028, so less than 3% correction. If you're cloning small areas, say 30 pixels wide, this translates to +- half a pixel, generally not noticeable. Hence my suggestion to clone with a small brush, frequently reset.

Let me know if you'd like to see a retouched version of this image.

--Rik

PS. Note to other readers... "Tony" is a professional colleague of Graham's. He's a member of this forum, but not a regular poster.

ChrisR · Post by **ChrisR** » Wed Jul 22, 2009 10:02 am

Give it a try and let me know if it works for you.

Certainly does. Much more interesting than the 2D versions. I'm a fan of 3D, and can't wait for multi-level 3D TV.
A year ago I could never see this type of 3D. I practised, and worked out about the alignment Rik referred to , and the business of the eye focus distance not being the same as the eye convergence distance. Now it takes half a second and I don't have to think about it. Keep at it, Andrew!

lauriek · Post by **lauriek** » Wed Jul 22, 2009 10:18 am

Chris, any tips for someone else who struggles with these? I can /never/ see these, my eyes seem to refuse to stay crossed when they refocus on the subject (I tend to use a finger in front of the subject to try to line everything up properly with both eyes) - very frustrating, I've /almost/ got to the point where I don't bother even trying any more, but still I persevere in hope!

rjlittlefield · Post by **rjlittlefield** » Wed Jul 22, 2009 10:56 am

Laurie, it sounds like you're getting messed up by the conflict between convergence and focus.

As aids, for parallel viewing try putting on a pair of reading glasses, or if you're nearsighted, take off your regular glasses. For crossed-eye viewing, try it with and without glasses.

Ultimately, what's needed is to break your eyes' lifelong habit of refocusing when they converge. It is just a habit, but it's a devilishly strong one. I still remember spending a couple of days of eyestrain doing the job, and it's been about 30 years now.

As a training tool, try starting 10 feet away from the monitor. You won't get a very strong stereo effect, but with the minimal convergence you should be able to fuse and focus OK. Then while keeping the image fused, slowly and smoothly move closer. This will gradually "stretch" the connection between convergence and focus. Stretch until it's uncomfortable, then let your eyes rest and do it again. Repeat as required.

--Rik

elf · Post by **elf** » Wed Jul 22, 2009 1:00 pm

What is the physical setup for shooting the stereo images?

How far does the camera need to move between left and right frames (assuming you're moving the camera)?

Is there an optimum distance based on magnification?

elf · Post by **elf** » Wed Jul 22, 2009 1:00 pm

Edit: Next time I'll try for three duplicate posts by clicking rapidly in non-responsive UI

What is the physical setup for shooting the stereo images?

How far does the camera need to move between left and right frames (assuming you're moving the camera)?

Is there an optimum distance based on magnification?

ChrisR · Post by **ChrisR** » Wed Jul 22, 2009 1:12 pm

One thing some people find very difficult to do is to hold a fixed amount of eye-crossedness.
Hold one arm out straight in front of you, with little finger and thumb raised above the other three knuckles. Cross your eyes and you'll see two hand images. Hold the eye-crossing with the thumb of one hand image against the little finger of the other.
See what happens when you tilt your head sideways - they become out of vertical alignment.
WHen you're happy to hold the crossed view, focus on the hairs on the back of one of the hands - holding the eye-cross. You'll find you can hold the eye-cross while looking over each hand image at will.

That's very similar to looking at the 3D pairs. You hold the eye-cross, which becomes automatic, then you can look around the middle of the 3 images.

That probably doesn't help at all....

ChrisR · Post by **ChrisR** » Wed Jul 22, 2009 1:41 pm

Another way of dealing with alignment:

When you first cross your eyes, the 3D pairs will probably go wild. like the top pic.
If you get your eyes flat-on to the screen, the angle will straighten, and then tilt your head left-right until the two line up vertically like the middle.
If you over-cross your eyes, I think it's easier. Relax them until the images are in the right place - it helps to concentrate on one spot in the image.
When you have that spot aligned, usually with the rest "not quite there", it's all too easy to "lose it" by making an effort to make your eyes behave. But I don't think there's anything extra you can consciously do. Just hold it for a few seconds, then look up and down that central image. Your brain has to do the rest!
Once it's "clicked in", you can move your head, look around the image, and it just stays there.

Planapo · Post by **Planapo** » Wed Jul 22, 2009 1:41 pm

Graham46 wrote:

Give it a try and let me know if it works for you. Feel free to share some things you saw in stereo that you didnt when looking at the 2D image.

Works great! In stereo the concavity of the underside abounds more , the coxae (bases of the legs) start to shimmer, and especially the 2nd and 3rd leg on the right side (seen from the animal) stick out nicely.

--Betty

Planapo · Post by **Planapo** » Wed Jul 22, 2009 1:48 pm

My recipe to get the stereo vision is just: I squint like mad when looking at it, then focus hard on the middle of the three images I get ... and voilà!

--Betty

rjlittlefield · Post by **rjlittlefield** » Wed Jul 22, 2009 3:01 pm

What is the physical setup for shooting the stereo images?

Lights, subject, camera. I believe that Graham rotates his subject while leaving the lights and camera fixed. I generally move the camera while leaving the subject and lights fixed. With very diffuse lighting, it doesn't make much difference. With directional lighting, I prefer to move the camera because otherwise shadow movement can introduce confusing differences (see HERE). But moving the camera allows to shoot only a limited range of angles, while rotating the subject allows to shoot a full circle.

How far does the camera need to move between left and right frames (assuming you're moving the camera)?

The amount is quite flexible. Values of 3-10 degrees work OK. Larger values produce an illusion of greater depth, but people are more tolerant of too little depth than too much.

Graham's use of 5 degrees between the lines of sight corresponds to viewing an object at about 27 inches. A separation of 8.5 degrees would correspond to about 16 inches. These are assuming a typical interpupillary distance of about 60 mm -- you can work the trig.

Is there an optimum distance based on magnification?

Short answer is no, it all depends on angle between the lines of sight.

Longer answer is yes, there are a couple of fine points that relate to magnification. But I'll skip those for now.

--Rik