Entrance pupil / no parallax point underwater

Hi

Am playing with doing some 360 degree panorama shots for virtual tour type shots of dive sites. I am having a pano head rigged up out of some stainless so I can take it underwater. (won't be the most elegant). The most important thing is to adjust the position of the camera so that the centre of rotation is about the entrance pupil or no parallax point. This is simple to do on land with a bit of trial and error.

My question is will this point remain in the same physical point on the lens or willl it change when the camera is in the water? (assuming focal length, focus point etc kept constant) That way I can adjust the head to the correct position on land, pop on the port and go shoot.

Any thoughts greatly appreciated.

there are some interesting (well, interesting for math types ) threads on WP on nodal points and such. I just did a quick search but can't find the one I am thinking of. Do a quick search on Nodal point or nodal point theory and many threads will come up about it. But.. I know there is one that has a whole bunch of info in it, i just can't find it...sorry

there are some interesting (well, interesting for math types ) threads on WP on nodal points and such. I just did a quick search but can't find the one I am thinking of. Do a quick search on Nodal point or nodal point theory and many threads will come up about it. But.. I know there is one that has a whole bunch of info in it, i just can't find it...sorry

Thanks Mike, will have a rummage...

Any thoughts greatly appreciated.

I think that even though you are underwater you would still rotate around the entrance pupil ... I don't see what would change that ...

If so, you should be able to position the housing in the bracket on land and use it underwater.

I think that even though you are underwater you would still rotate around the entrance pupil ... I don't see what would change that ...

 

If so, you should be able to position the housing in the bracket on land and use it underwater.

Nothing changes the fact that he should rotate around the entrance pupil.

However, the entrance pupil is defined as the image of the aperture as viewed from the front of the lens. If you add a dome port and water, that could change the image of the aperture, and move the entrance pupil. I'm not sure it does (I'm currently 3,000 miles from my optics texts and simulation programs), but I can certainly conceive of the possibility.

Nothing changes the fact that he should rotate around the entrance pupil.

 

However, the entrance pupil is defined as the image of the aperture as viewed from the front of the lens. If you add a dome port and water, that could change the image of the aperture, and move the entrance pupil. I'm not sure it does (I'm currently 3,000 miles from my optics texts and simulation programs), but I can certainly conceive of the possibility.

The basic geometry is simple ... the entrance pupil is the vertex of the refracted angle of view ... I'm not sure it moves, but now that I think about it, I'm not as sure it doesn't ... that's some tricky dome math I think.

My gut tells me it doesn't move ... but my gut is old and rotten at this point ... my math-figurin' days are over.

Also - if you are using a fisheye, the no parallax point is technically not the entrance pupil - which can change locations depending upon the angle you look for it from - I think the least parallax point for a fisheye is actually behind the pupil.

The basic geometry is simple ... the entrance pupil is the vertex of the refracted angle of view ... I'm not sure it moves, but now that I think about it, I'm not as sure it doesn't ... that's some tricky dome math I think.

 

My gut tells me it doesn't move ... but my gut is old and rotten at this point ... my math-figurin' days are over.

 

Also - if you are using a fisheye, the no parallax point is technically not the entrance pupil - which can change locations depending upon the angle you look for it from - I think the least parallax point for a fisheye is actually behind the pupil.

http://wetpixel.com/i.php/full/proper-dome...-and-placement/

This primer would lightly imply that the dome port does not alter the location of the pupil as it advocates using topside measurements of the pupil to correctly place the dome in the first place.

Wow, my head hurts now....

Spent many hours scouring the net and old posts on wetpixel. It would seem that there are many different names which are used both correctly and incorrectly depending on where you look. (nodal point, front entrance pupil, rear entrance pupil, blah blah blah.) Basically what you are looking for is the point you need to rotate around to eliminate parallax. Having found that the next question does slapping a port and a load of water infront of it make any difference to this point. I guess the only way is to go and try it.

I then looked at linear panoramas, ie taking photo's in a parallel plane to the object (eg a wreck). It would seem that most of the software does not like this as the viewpoint and perspective keeps changing. Assembling the image seems to be very manual especially if you have to de-fish each image first....

Then trying to get a guy to make up the mount (especially in Egypt!!!) aaaaaaagggggggghhhhhhh....

Time for a Sakara me thinks!!

I think that it's a job for super-Photoshop...

... not an engineer! The technical variability of shooting in the sea would, I guess, create more variability shot-to-shot than can be engineered out with a pano-head, and need to be handled in post-processing, anyway.

I could be wrong, of course...

Tim

I think that it's a job for super-Photoshop...

 

... not an engineer! The technical variability of shooting in the sea would, I guess, create more variability shot-to-shot than can be engineered out with a pano-head, and need to be handled in post-processing, anyway.

 

I could be wrong, of course...

 

Tim

 

For sure,

For a fixed viewpoint, the less change in viewpoint then the easier time the software has. A tripod and head would help here. A linear panorama seems to be a completely different animal with a huge amount of post processing. From what I have read to successfully do linear panorama you need to ensure that any object that falls outside the parallel plane of the sensor would require a seperate shot with that detail centred in the frame, so it not a case of taking a shot every metre or so but completely dependant on the subject. Read an article about a guy who took pics of a wall about 40m long. That was basically 2d and took 2 days of post processing. Looks awesome though. So add to that fish eye distortion, 3d objects so loads of perspective change, and probably a humping current then linear seems more hassle than its worth.

For fisheye lenses, as Jeremy rightly said there is a completely different effect happening here. The no parallax point does not exist as this moves with the angle of light entering the lens. Instead you have a least parallax point which falls behind the entrance pupil. This point minimises parallax errors. Usually not so much of a problem as less shots are being stitched. Therefore there will always be some parallax so the best we can do is minimise it. Any errors can then be brushed out after stitching.

Article on LPP Least Parralex point for fisheye lenses

Hopefully will have the bracket soon, enshala!!! And then will have a play in the pool.

A linear panorama

Use a rectilinear lens ... use good stitching software like AutoPanoPro ... force the EXIF to be a 1000mm lens ... should work.

Honestly, I'm not sure what all the hub-ub about finding the entrance pupil/nodal point is. All you need to do is set up your rig UW, with one object close to the lens and something else further away. As you rotate the camera, move the close object b/w the left and right sides of the frame. Look at the distant object to see if it moves relative to the close object. Slide the camera forwards or backwards on a nodal slide to compensate. Worth doing topside first and seeing if it correlates UW.

From the Really Right Stuff website:

The nodal point is best determined by experimentation: Use a "nodal slide," such as our 7.4-inch long MPR-CL II. The objective is to slide the camera along its lateral axis until the nodal point is positioned over the axis of rotation. Precise positioning is critical if your subject matter is close, but much less important if you're shooting distant landscape shots.

1. Find two vertical objects; one near, one far. Position your equipment so that these objects line up in the viewfinder.

2. Level your equipment.

3. Mount and center your camera on the MPR-CL II.

4. Start out by positioning the approximate center of your lens over the axis of rotation.

5. Now pan left.

6. If the rear object appears to shift to the left, then you are ahead of the nodal point. Slide the MPR-CL II forward and try again.

7. If the rear object appears to shift to the right, you are behind the nodal point. Slide the MPR-CL II back and try again.

8. When the optical center of the lens is directly over the axis of rotation, the rear object will not appear to move relative to the front object.

9. Record your results so that your setup can be re-created.

Cheers,

Chris