questions concerning panoramas, nodal points and alignment accuracy.

[SeanOly]

Hello.
I think I understand the principal in that it is desireable to have the camera and lens rotate about a point known as the "entrance pupil" or "nodal point".
In experimentation I have seen the effect of moving the camera and lens back and forth along the lens's axis relative to the axis of rotation i.e. when the camera and lens are 'panned', the introduction or removal etc. of a parallax shift between objects in the background and foreground.

Yesterday, just to see the effect of inaccuaracies in the alignment of the lens axis with the axis of rotation, rotating about a vertical axis only, once I had found this point for the combination I was experimenting with (Olympus E620, 18-180mm lens, f=100+a bit mm) I deliberately moved the camera and lens a couple of inches to the side i.e. perpendicular to the lens axis. I expected to see the introduction of a parallax error but in truth I didn't see any signifcant error introduced.

This leads to the question, what should I have seen?

Please note I did not take photos during this process and was looking at the live view on the screen on the back of the camera.

I am hoping to take a trip up the West Coast of Scotland and presume and hope I will have the opportunity to shoot many panoramas, am I being daft to consider using long focal length lenses, say up to 600mm for this?

Having seen proper panorama tripod attachments I am wondering would it not be possible to use a Gimbal Head as a panorama attachment?

[Manfred M]

The answer to your question is "it depends".

Parallax issues are going to potentially crop up in elements of the image that are relatively close to the camera and the impact of not rotating the camera around the no parallax point will show up in those parts of the image. The elements that are far away will not be impacted at all.

In this first shot, I had the camera set up and rotated around the no parallax point and the hand railing looks correct (straight). Had I merely rotated the camera, then this would have shown up as a curve.




This second shot is a hand-held pano where I rotated my whole body to get the shot. There is nothing that is close enough to be impacted by the parallax in the shot, so it looks fine.




The answer is it depends...

With respect to your other question, I have never used an ultra long lens for panos. The atmospheric effects (haze, local heating) will likely cause you some real issues. On a 4/3 camera a 600mm lens is FF equivalent of 1200mm, so simply keeping it level is going to be a challenge. I rarely do panos with anything longer than with a short to medium telephoto focal length. The first shot is at 24mm and the second at 70mm, both on a full-frame camera.

I don't shoot with a yoke, so don't know how well that will work. I use a very heavy duty tripod with an indexing head and long focus rail for many of my panos. The indexing head is not in the following shot, but the L bracket, tripod head I usually use as well as the long focus rail are. The ball head I use is rated at 22 kg / 50 lb, so there is little droop in the setup.

[george013]

Hello.
......

Yesterday, just to see the effect of inaccuaracies in the alignment of the lens axis with the axis of rotation, rotating about a vertical axis only, once I had found this point for the combination I was experimenting with (Olympus E620, 18-180mm lens, f=100+a bit mm) I deliberately moved the camera and lens a couple of inches to the side i.e. perpendicular to the lens axis. I expected to see the introduction of a parallax error but in truth I didn't see any signifcant error introduced.

This leads to the question, what should I have seen?

Create a frame where 2 vertical lines are in one line, by example a doorpost and a window post. If these are on the left side of your frame, then turn the camera so that those 2 lines are on the other side of the frame. If you rotated around the parallax point, they will be still in one line. If not, then they are not in one line.
How did you find that parallax point for your camera/lens?

It's not so difficult to make panos from the hand. As Manfred mentioned avoid long lines at a close distance.

I use Microsoft ICE for panos: free, easy, fast and good. https://www.microsoft.com/en-us/rese...posite-editor/

George

[SeanOly]

Manfred, thanks for the reply, I think I need more practise/experimentation.
George, thanks for the reply, I used the two vertical objects arrangement, in this case the screw atop a naked tripod column as the near object and a 'distant' lamp post as the distant object, though one or other was alway slighlty out of focus. That said, I made the initial alignment in the centre of my 'image' and compared that to the alignment at the sides of the image, your method, initially aligning to the side of the image, might give greater sensitivity. I have used ICE before but for merging map scans lol

[george013]

Manfred, thanks for the reply, I think I need more practise/experimentation.
George, thanks for the reply, I used the two vertical objects arrangement, in this case the screw atop a naked tripod column as the near object and a 'distant' lamp post as the distant object, though one or other was alway slighlty out of focus. That said, I made the initial alignment in the centre of my 'image' and compared that to the alignment at the sides of the image, your method, initially aligning to the side of the image, might give greater sensitivity. I have used ICE before but for merging map scans lol

Yes, you must use the maximal possible rotation of the camera, lens dependend. And correct de postion of the camera according the results. Is your tripod mount under your optical ax of the lens?

George

[Manfred M]

Sean - one other thing to watch for when determining the location of the non-parallax point for a zoom lens is that it can change, based on the focal length used. Check out various focal lengths of the lens that you are testing. The internal elements can shift around in a lens as you zoom in and out, so the entrance pupil, which defines the no parallax point.

I generally test max, min and middle of range when I test a lens and record the actual offset on a "cheat sheet" that I take along when shooting a pano.

[dem]

This leads to the question, what should I have seen?

Ghosting and distortion of straight lines in the foreground. To stich two or more shots they should look identical in the overlap area. If the overlap areas are not identical, the software will have hard time deciding how to stich this panorama.

... am I being daft to consider using long focal length lenses, say up to 600mm for this?

If you have a heavy perfectly leveled tripod and can control the orientation of the camera with accuracy of 1 degree or so, why not? Fire 10-20 shots, stich them together, job done.

How many megapixels do you want to end up with? I am guessing you don't want to shoot more than about 20 shots per panorama?

Multilayer panoramas (Brenizer method) are also possible but are more difficult to do.

https://en.m.wikipedia.org/wiki/Brenizer_Method

[george013]

A 10 year old pano, out of hand, 9 pictures I thought, 24mm, ICE.
I post it full size so you can check for faults.


George

[SeanOly]

George13 that pan is impressive.
With regard to "Is your tripod mount under your optical ax of the lens?", this is one of the reasons for the experiment of shifting the lens sideways, using the Gimbal Head the answer is, nominally no it is not. The discrepancy is about 6mm on the too-far-across side of things and to some extent varies with whichever Arca Swiss style plate is used. The latter is due to variations, between different plates, in the effective width of the dovetail, though I suspect this is less than 1mm. I was quite surprised by the former but maybe the use of a Gimbal Head for panoramas is abnormal and therefore not designed for.
Anyway I think I can just about correct the total discrepancy by using the cross slots in the plates to 'mount' the lens or camera rather than the lengthwise slots in those plates but I don't think this is an ideal solution and I have seen some Gimbal Heads where the discrepancy look to be too great to correct in this manner.
As an aside I actually also made a molding of epoxy putty around a lens's foot on one Arca Swiss plate so that that lens can be remounted accurately at any time. This worked quite well but means that that plate is dedicated to that lens and only that lens.
I realise I can also use a cross slide plate, as it looks to me like Manfred used in the photo of his camera, to correct for this discrepancy.
If I use a 'proper' panorama 'rig' e.g. a Manfrotto 303plus or 303SPH then it's an easier matter to align the axes but I think they, to some extent, still rely on judgement.

Manfred, thanks. BTW should have said I really like the city panoranma.
I was aware of the needs to map the 'nodal distance' for various focal lengths and have found a few charts for a limited range of lenses out in the www but data for Olympus lenses seems rare and the only site I have found with any is https://olypedia.de/index.php?title=Panorama. I was thinking of mapping them for each focal length marked on a given lens.
This raises two other questions,
1) What distance do you measure?
So far I am just recording numbers on scales on plates etc. corresponding to some notable feature, this is somewhat arbitrary and needs converting to some sort of consistent scheme. One chart I have seen, https://wiki.panotools.org/Entrance_Pupil_Database, quotes the distance of the entrance pupil fromthe lens's mounting face, this seems sensible to me as it is data for the lens itself but also obviously needs the position of the face relative to the "scale" to be determined. A different chart measures from the centre of the camera's mounting screw but this makes the reference length camera dependant and it seems that, between different camera models, there is some variation of the distance between the mounting screw's centre line and the lens mounting face. This leads to the next question.
2) Ignoring any variation of the latter distance, does the camera attached to the lens alter the position of the entrance pupil? For instance if I swapped the E620 for an E500? Or for say in a Canon rig an EOS 750D for an EOS 100D?

Dem thanks. I do have an indexing head that does seem to provide slop free increments. With regards to panorama size, the honest answer is I haven't a clue. The bigger the better I suppose but going by my experience of merging maps, the bigger the individual images and or the more images being merged the bigger the headache for the merging software. From memory trying to merge images in stages wasn't successful but I wouldn't guarantee that that recollection is correct

[Manfred M]

1) What distance do you measure?
So far I am just recording numbers on scales on plates etc. corresponding to some notable feature, this is somewhat arbitrary and needs converting to some sort of consistent scheme. One chart I have seen, https://wiki.panotools.org/Entrance_Pupil_Database, quotes the distance of the entrance pupil fromthe lens's mounting face, this seems sensible to me as it is data for the lens itself but also obviously needs the position of the face relative to the "scale" to be determined. A different chart measures from the centre of the camera's mounting screw but this makes the reference length camera dependant and it seems that, between different camera models, there is some variation of the distance between the mounting screw's centre line and the lens mounting face. This leads to the next question.

In my case, I use an approach that is specific to the equipment that I use. My long focus rail has markings that are 0.5 cm and my indexing head breaks things down to the mm, as shown in the picture below. I just read off those 2 scales. Right now it is set for 4.2 cm, which is more than close enough for this work.

The reason I do this, rather than measure from the camera's sensor marking is that my technique is really easy to use in the field, even though it is a relative versus an absolute measurement. Standing outside in sub-zero temperatures at night, I want something easy and fool-proof.

The indexing head allows me to centre it on my Arca-Swiss mount and level the whole apparatus using the ball head. I lock everything down and then index my shots by rotating the indexing head.

2) Ignoring any variation of the latter distance, does the camera attached to the lens alter the position of the entrance pupil? For instance if I swapped the E620 for an E500? Or for say in a Canon rig an EOS 750D for an EOS 100D?

Any particular make of cameras has a specific flange to sensor distance, so as long as the cameras take the same lens, the data should be consistent. I think there could be a problem with Olympus / Panasonic 4/3 versus micro 4/3, but anything else should be constant. Same comment with Nikon, Canon (although I think that their full-frame and APS-C mounts are different). Nikon is consistent between FF and APS-C.

[george013]

Sean,

I'm not sure if I understood you, but the sizes you've to take care of are listed in your link https://wiki.panotools.org/Entrance_Pupil_Database. It's L1+L2.

George

[george013]

Sean,

I'm not sure if I understood you, but the sizes you've to take care of are listed in your link https://wiki.panotools.org/Entrance_Pupil_Database. It's L1+L2.

George

Not much feed back

George

[Manfred M]

Not much feed back

George

That is quite common on this site. People pop up and ask a question and are often never heard from again, so we don't know if the responses were helpful or not.

[SeanOly]

Sometimes people can't get on the web etc.
Yes, the responses have been useful and I am trying to get out for a play today.

Manfred, with regards to "Any particular make of cameras has a specific flange to sensor distance, so as long as the cameras take the same lens, the data should be consistent." I thought that would be the case unless there was something really weird about this "entrance pupil" but, since I had seen mention of the camera being involved, I thought it best to check.

George, ta I saw that but was wondering what people use out in the field.

[Manfred M]

Manfred, with regards to "Any particular make of cameras has a specific flange to sensor distance, so as long as the cameras take the same lens, the data should be consistent." I thought that would be the case unless there was something really weird about this "entrance pupil" but, since I had seen mention of the camera being involved, I thought it best to check.

The flange to sensor distance has to be fixed for any group of camera bodies that take specific lens mount lenses, so that will not vary. Third party lens makers have to take this into account when they design and manufacture a lens that will be used across multiple camera makes.

The entrance pupil is a property of the lens and the reason we have to check multiple focal lengths for a zoom lens is that the entrance pupil can shift as the lens elements physically shift in the lens body when we zoom in and out.

[William W]

. . . I am hoping to take a trip up the West Coast of Scotland and presume and hope I will have the opportunity to shoot many panoramas, am I being daft to consider using long focal length lenses, say up to 600mm for this? . . .

It seems to me a silly idea.

Without wanting to divert the conversation into a debate upon precise definitions, let’s just take for granted that a ‘panorama’ assumes that the final image is a ‘wide view’.

Let’s next assume to agree that a ‘standard view’ or ‘normal view’ is what you get when you use a 25mm lens on your Olympus E620 – ergo a ‘panorama’ seeks to make a picture of a wider view than that.

Let’s say, for the sake of this conversation you want your ‘panorama’ to make twice as wide a view as that of a ‘standard lens’ - i.e. the view of a 12.5mm Lens. Using a 600mm Lens on your Olympus E620, I think (mental arithmetic only) that you’d need about 50 shots, stitched together just to make a similar horizontal span. (i.e that is negating any Vertical Span that you might want to make).

***

OBVIOUSLY - if by 'panorama' you are NOT referring to the expanse of view of the Final Image, but 'panorama' is used to describes the technique of STITCHING images together, then that may be a different matter.

WW

[xpatUSA]

For some light relief, here's a stitch-up of today's porch repair-in-progress:

Two shots with the Sigma DP1 Merrill 19mm fixed lens compact, taken without much care (did use a tripod, though).



Blind guess-work as to the pivot axis but it came out quite straight seeing as the cam was angled up a bit. Used 'Hugin' default everything ...

Gusting wind blew some of my porch roofing off: a good thing really because it revealed wet rot, so I decided to replace the whole fascia.

[William W]

Aside - Nice stitch-work. Do you knit too?

I do admit that I tend to make (most of my) 'panos' handheld and with guessometry: it's a lack of patience with landscape work thing.

From another Forum, I have chats with a bloke from Texas. There's wild (unusual) weather there, so he tells me. He's still complaining about the odd cold spells.

Nice to see the detail applied to painting the conduit black at the appropriate place.

Wet rot is a pain. I have it in the fence a bloody long fence all around the house.

Happy Easter, etc.

[Manfred M]

I had a wet rot problem that got fixed last year. Some of the fiberglass insulation looked a bit like what we see in this shot; compressed by the water that got in. We ended up having to change our roof drainage to ensure that the problem did not come back.

Nothing that throwing money at it couldn't fix, but I'd rather spend it on something that is more fun.

Yes, I've done a lot of hand stitched panos too. Right now I have one that I cannot stitch together with the usual software because there aren't enough edges for the software to figure out where the individual shots overlap. Salt pan and sky look kind of the same.