360x180° Panorama Tutorial - Pt.6: Finding the no-parallax point (aka nodal point)


Uploaded by elfloz on 24.09.2011

Transcript:
Hello and welcome to another tutorial video. Today i'm gonna talk about
how to adjust or calibrate your panoramic tripod head.
In the earlier videos I showed you that you don't necessarily
need a tripod, you can just use the "string technique",
but sometimes it is useful to use a tripod.
But if you do use a tripod you absolutely have to use a
panoramic tripod head --- otherwise there's no point, because you
always want to rotate your camera
about the no-parallax point of the lens.
So, where do we begin? Well, I've started to assemble this "Nodal Ninja 3" this case.
Most of the panoramic tripod heads come with a base plate
that you install on the base of the camera, and you put that into the tripod head.
Now, when install the base plate, you always want to make sure you
install it in the same way. For instance,
the base plate of the "Nodal Ninja 3" has little "lip" at the front
and at some stage I decided that I'll always have the lip at the front,
and this is how I however always install it (because the plate is not symmetrical).
Another thing you alwyas want to look out for is that
the plate is really installed parallel, absolutely parallel to the camera,
as good as you possibly can.
Now, once you've installed that,
you can . . .
. . . bring that into the tripod. The next step then is going to be
to adjust that vertical arm along the horizontal axis.
What you want to do is get the camera centered exactly
above that rotation axis.
If you have a camera with a live-view feature this is very straightforward,
because what you do is you just rotate
the camera vertically down like that
(you should see this in the little inset now)
Then as the next thing you would turn on the live-view,
zoom all the way in on the live-view and then make sure is that
this point here in the middle is in the middle off your image.
If you don't have a live-view, what you would do is
just slide the arm back and forth,
making sure you tighten it in (because, as you can see, as I tighten
the screw, it changes a little bit).
So make sure you always check it with the screw tightened:
You take an image, zoom all the way in
then you look whether it is aligned or not
--- and you'd shift it accordingly.
In this case I've installed the stopper
so that I can find that point quickly when I re-assemble it.
But you'd also would want to write down the numbers
as sometimes you change stopper,
or if you have different cameras, which require different settings,
you'd need to change that as well,
so make sure you install the stopper and also write down those values.
Once I've done that, the next thing - this is the more time-consuming part -
is going to be to adjust the
no-parallax point along the upper rail. For that I'll bring the camera around.
Okay, so the next step, as I said, is to adjust the camera on that
upper rail, moving back and forth. I forgot to say earlier is that
you want to make sure that the tripod head is level (it has a little bubble level here
that you want to get centered), because otherwise
you'll get some funny off-axis effects.
Now, in order to adjust the camera along the vertical rail I have a
little tool installed here, a highly scientific skewer stick.
I have aligned that with some big, pointy object in background,
this beautiful clock tower down there.
What I would do now is to try to keep
those two things aligned, as I move the camera around. They'd want to be in the
same line because if I want to stitch those images later on
I'd want the skewer relative to the clock tower in the exactly same position.
If you manage to adjust the tripod head correctly, it'll be
very, very easy for the stitching software later on
to put these two images together.
Now, there's another important thing, and it took me a little while to
figure this one out or to believe it (I thought, well, it can't be
that big an effect, but it is a big effect!)
The no-parallax point of a fisheye lens shifts back and forth as you move away
to the side, away from the optical axis. This makes big difference, because
It depends on *where* you are going to stitch your images.
This is an 8mm fisheye lens on a crop camera (1.5x crop)
which means I will be taking 4 images around, i.e. 90° apart.
Now that means, for the lens,
the two images together will be overlapping
at the 45° mark, so 45° to the left and 45° to the right
of the center of the lens.
So it's at that point (45° to the left and right), that's where I want to get
the parallax absolutely perfect.
Some people adjust their lenses where they go to the very extreme side of the lens
--- you don't want to do this. You want to do it at
45° to the left and 45° to the right.
Now if this was, say, a 10mm or 12mm fisheye lens,
(whatever) and you'd need 6 images around the horizon,
you would have to take your images 60° apart, and in that case you want to
reduce the parallax at 30° to the left and right.
So in this case here (this is the zero-position now)
I will be turning the camera --- 45° in one direction
--- back to neutral --- and 45° in the other direction.
Now, look very closely at the skewer relative to the church tower.
What I'll do now is I'll bring the camera all the way back on the rail.
That means, the camera . . .
. . . is too far, it is *behind* the no-parallax point.
Now look closely how the skewer,
--- that is the "close object" --- is moving relative to far object.
You can see it is moving in
the *same* direction as I rotate. So now I'm rotating the camera this way,
and you'll again see the close object (skewer) shift in the *same* direction.
So it's shifting in the *same* direction as I am rotating.
That happens when you're *too far back* relative to the no-parallax point.
Now I'll go all the way to the *front*, so that I'm closer
to the object, that is *in front* of the no-parallax point
(relative to the rotation axis)
Again, look closely what's happening to the skewer:
As you can see, it is moving . . .
Okay, right, I think I'm right on the no-parallax point here . . .
Hm, I can't go any closer . . .
Well, what you would see if you are
*beyond* the no-parallax point,
you would see that the close object
is moving in the *opposite* direction as I'm turning
I came up with this really great way of remembering this:
If the close object (we're *always* talking about the close object)
is moving in the *opposite* direction
relative to the background object (and how you are turning)
so Opposite --- then you want to move Outwards, right?
"O.M.O." --- if it's moving Opposite Move Outwards.
Now this is getting really hard to judge
on the display here on the back of my camera . . .
So what you would do next is
you would take photos and then zoom all the way in and compare them.
Ok? So I will turn off the video on this camera and
I'll start taking still images, where I'll zoom all the way in.
Ok, so I'll turn all the way to the left . . .
(that's 45° on here)
take an image . . .
rotate to 45° of the center in the other direction . . .
and take another image.
Now I'll zoom all the way in . . .
(I'll show you my screen on the video)
Here I have the *right* image,
that's right of the clock tower,
and the skewer is in a certain position relative to the clock tower.
Then I'll switch to the previous, "left", image
(so the church tower is now all the way on the right)
again zoomed on the way in
to judge what happened:
As I turned the camera *left* you notice the skewer moved
over to the *right*
--- so that's *opposite*, right?
The skewer moved in the *opposite* direction as I moved the camera.
In that case: I have to go outwards a little bit.
So I'll do the same exercise again.
I'll go back to the initial position on the left . . .
taking one image to the left of the tower . . .
rotate 90° . . .
and one to the right of the tower.
Again, I zoom all the way into the image.
Look closely now. So again, we are to the right of the tower . . .
and the skewer is in a certain position.
Switching over to previous image, so I moved the camera over to the *left*.
and the skewer appears to have moved to the *right* again.
So *opposite* --- hence I need to go *outwards*.
So I'll move a little bit back further.
Again, same trick: Take one image to the left . . .
90° to the right . . .
zoom all the way in . . .
On my preview-screen back here
I notice that the skewer is now aligned with the side of the clock tower . . .
I know "rotate" the camera towards the *left* and . . .
what happened relative to the clock tower?
(I'm scrolling around here . . .) Aha!
So this time the skewer has moved to the *left* as well.
That means . . .
If *opposite*, you have to move *out*.
But if it's the moving in the *same* direction as the camera turned
I now have to move inwards a little bit.
There we go, a little bit more . . .
And same exercise again.
I turn to the left, take one image . . .
rotate over . . .
take a second one . . .
Zoom all the way in on the preview-screen back here . . .
Look, there's a tiny little bit of that clock tower just
to the right of skewer. Now look at the next image . . .
having virtually turned my camera to the left... what happened to the skewer?
The skewer... well it's almost there, but I'll leave it at that. You get the idea.
Just remember if the close object is moving in the *opposite* direction
as you are turning the camera, then you want to move outwards.
So with this back and forth
you start to find that point,
and once you've aligned it perfectly
don't forget to bring the stopper up
to the position that you have the camera in now
and then you also want to take note
--- or even take a photo with your cell phone ---
of the marking where you have your camera.
At that point you have found the no-parallax point and you're done.
So thank you very much for your attention,
please to visit my website www.pano.ie
and shoot me an email if you have any questions, alright?
Bye.