PDA

View Full Version : Bubble Trubble



newboy
04-14-2009, 03:58 AM
If I adjust my NN180 so that the bubble is dead centre in the reference circle of the spirit level, why does the bubble shift as I rotate my camera?

badders
04-14-2009, 05:42 AM
Because it does... Use it as a reference only for initial setup. As you're only shooting 1 x row of images with the NN180 it doesn't matter if the image isn't aligned to the horizon at time of shooting because you can fix this in your stitching software.

nick fan
04-14-2009, 05:45 AM
If I adjust my NN180 so that the bubble is dead centre in the reference circle of the spirit level, why does the bubble shift as I rotate my camera?


the bubble is very sensitive. when you rotate the camera, there can be very slight movement of tripod. This will cause the bubble to move. Also, the level is set accurate only to certain extent. There is no perfect level in this world.
The bubble is fine if it does not move out of the black circle. It will not affect your stitching.


Nick

Terrywoodenpic
04-14-2009, 07:33 AM
As Nick says the bubble is very sensitive.
when you set up, you area aligning the plane of rotation...(The flat surface in the rotator) to be level.
After that you must leave it alone. or you may give problems to your stitcher.

As you rotate the arm and camera, the weight distributed through the three legs of your tripod changes, even a minimum amount of flex is noticed by the sensitive bubble.
There is likely to be some minimally measurable flex in all tripod systems, ball heads, levellers and camera mounts.
However you will notice as you continue the rotation to where you started, the bubble will be level again. this shows the plane of rotation is constant even if the bubble moves.

newboy
04-14-2009, 07:48 AM
the bubble is very sensitive. when you rotate the camera, there can be very slight movement of tripod. This will cause the bubble to move. Also, the level is set accurate only to certain extent. There is no perfect level in this world.
The bubble is fine if it does not move out of the black circle. It will not affect your stitching.


Nick



So you wouldn't suggest I re-level at 180 degs?

nick fan
04-14-2009, 07:50 AM
So you wouldn't suggest I re-level at 180 degs?


no. as with any pano head, don't re-level it. Actually, bubble level in all pano heads move to some extent doing rotation. It is never a problem.


Nick

Terrywoodenpic
04-14-2009, 12:38 PM
If nick used some of the less accurate bubble levels in his manufacture. bubble movement may then, not even be noticeable.
However who wants a less accurate product?

realman10
10-15-2009, 09:33 AM
If nick used some of the less accurate bubble levels in his manufacture. bubble movement may then, not even be noticeable.
However who wants a less accurate product?


I'm sorry, but you are missing the point. The accuracy of a bubble level is not in bubble itself but IN THE WAY IT IS MOUNTED.
And it looks like many NN5, mine included, have a bubble which is not perfectly parallel with the rotation plan of the lower arm.
I'd like to ask Nick if there is a way to take the bubble off and then re-install it in a proper way.
My first nodal head was a Panosaurus, clearly inferior to the Ninja, but it had a detachable bubble (which kept falling onto the ground as I forgot to take it out), but at least I had been able to tune it very finely so that by turning the lower arm it would keep the level plane.
Nick, any advice? it's the only disappointment I had with the NN5.
Thank you

Terrywoodenpic
10-15-2009, 09:43 AM
I'm sorry, but you are missing the point. The accuracy of a bubble level is not in bubble itself but IN THE WAY IT IS MOUNTED.
And it looks like many NN5, mine included, have a bubble which is not perfectly parallel with the rotation plan of the lower arm.
I'd like to ask Nick if there is a way to take the bubble off and then re-install it in a proper way.
My first nodal head was a Panosaurus, clearly inferior to the Ninja, but it had a detachable bubble (which kept falling onto the ground as I forgot to take it out), but at least I had been able to tune it very finely so that by turning the lower arm it would keep the level plane.
Nick, any advice? it's the only disappointment I had with the NN5.
Thank you


I know exactly what you mean.
However You may not think so, but the bubble is more accurate than it even needs to be.
Provided it stays in the ring it will have no effect on the stitch as the plane of rotation will be identical in every shot.
The "plane" of rotation is always as accurate as the surface of the rotator.
(Provided that is, you do not adjust the head to level the bubble with each shot.)

hindenhaag
10-15-2009, 11:04 AM
Pay attention to the bubble once: in the beginning of your shots. Take your tour around and it will be at the same place. I never had stitching problems with levelling at the beginning.

To compare the results, i tested a round relevelling in between: i got a lot of stitching problems, had to add lot of controlpoints manually.

So this was the proof for me to leave it alone as was explaned above.

When you don't believe it, test it on your own with releveling, and i think you will spend a nice evening.
:wink:

Heinz

realman10
10-16-2009, 02:46 AM
Hi, let me make myself clearer:

1. I am not saying that you need to re-level the head as you turn it around, that would be totally wrong. Since we know the level is offset, by re-leveling you only make sure that it is always off-level and always in a different off-level state. Not very good right?
2. The fact that when you complete the 360¬? turn it comes back to the right level proves nothing, just that the head and tripod did not move, the very least we can expect
3. The problem is not in the bubble but in the incorrect way the bubble is mounted that is not perfectly parallel to the rotating plan of the lower arm.
4. Everyone obviously chooses his own quality standards.

I'm getting the uncomfortable feeling that I am being seen as a "provocateur", which I am not. I am a paying customer trying to help improve a good but not perfect product.
The only real answer I got so far is that this is not a serious issue and we should accept the small error, let's call it tolerance for convenience.
For those who are interested I've found a workaround: find the real neutral position of the bubble, which would be probably when the bubble is somewhere on the border of the black ring and then mark that position with a tape. Next time just set your rig so that the bubble is again in the marked position. Now if you turn the arm around, the bubble will stay in the same position, in other words it will not move around. Now the head is properly leveled.

I still believe that a properly calibrated bubble is a pre-requisite for any quality precision tool, no matter how critical the item may be.

Wim.Koornneef
10-16-2009, 03:00 AM
Hello Forum,

Many people using all kind of panoheads from different manufacturers are complaining about there bubble level.
The issue is not that the bubble level itself is inaccurate but that the precision of the leveling when it is mounted (often glued) is not 100% but at most 99.99%
This very small off set is normal as long as the bubble keeps inside the black circle of the bubble level and can be ignored.
However even the smallest off set of the ideal level mounting position is causing a moving bubble when the panohead is rotating with complaints as a result.

An off set is not as bad as it sounds but if you want to have a "precise" bubble level you can "adjust" every bubble level yourself, not by removing the bubble level and then remounting it but by finding the real center of the bubble level and then marking or memorizing this point.
This is how you do it:
1) mount your panohead on a sturdy tripod and place the tripod on a hard en solid surface,
2) level the panohead so that the bubble is in the centre of the black marked circle of the bubble level,
3) rotate the panohead 180 degree, for sure the bubble will move and perhaps even will go a little outside the black circle !
4) adjust the legs of the tripod, or use a panohead leveler if mounted, to level the bubble "halfway" the original center position and the new position, remember this "halfway" position, think about the position like it was a clockwork and memorize the "time" of the bubble,
5) rotate the panohead 90 degree, the bubble will move again,
6) level the bubble again "halfway" the last (memorized) position and the new position and remember the new "halfway" position,
7) repeat step 4 to 6 and eventually you will find a point where the bubble is not moving,
:001_cool: congratulations, you have found the real center of the bubble level and all you now have to do is to memorize this position (again think about it like it was a clockwork) or you put a small dot with a black permanent marker pen in the center of the bubble,
9) when using the bubble level the next time you don't level the panohead by using the black marker circle but instead you level to the real center of the bubble level and you will see that now the movement of the bubble will be very small or even absent :-)

Success,
Wim

Terrywoodenpic
10-16-2009, 03:55 AM
Wim.Koornneef's way seems to be ideal for any one who is anal about this.
However you can make perfect spherical pans with the head in any orientation, It simply is not an issue for stitchers.

I am sure Nick would say the that neither perfect bubbles nor perfectly set bubbles are economic.
Scientific and engineering instruments usually come with adjustable bubbles that you can set yourself. Those with fixed bubbles will state the tolerance. ( usually quite large.)

I would have thought. a bubble with a spherical base could be made, at a reasonable price, that could be self set and locked in place with a top cap. If this were to be true Nick could sell it at a premium price as an alternative.

However I am very happy with my bubble as it is.

The bubble on a tripod does not usually rotate, so we are unaware how inaccurate they always are.

realman10
10-16-2009, 04:32 AM
A big thank-you to the mods.
Problem solved, I'm also happy I gave a little contribution.
A prompt response like this glues a customer forever, I know you are mods and not employees but the community is here!

Mike

nick fan
10-16-2009, 09:25 AM
Wim.Koornneef's way seems to be ideal for any one who is anal about this.
However you can make perfect spherical pans with the head in any orientation, It simply is not an issue for stitchers.

I am sure Nick would say the that neither perfect bubbles nor perfectly set bubbles are economic.
Scientific and engineering instruments usually come with adjustable bubbles that you can set yourself. Those with fixed bubbles will state the tolerance. ( usually quite large.)

I would have thought. a bubble with a spherical base could be made, at a reasonable price, that could be self set and locked in place with a top cap. If this were to be true Nick could sell it at a premium price as an alternative.

However I am very happy with my bubble as it is.

The bubble on a tripod does not usually rotate, so we are unaware how inaccurate they always are.


Is there a real need to get perfect level? The fact that other manufacturers are accepting this "problem" allows us to justify this "problem". :-)
We try to set the level as accurate as possible. We mount the lower rail on a heavy metal block which is leveled using instrumental precision bubble which is calibrated EVERY time before use. The bubble is glued on the lower rail using very thin layer of epoxy. Before the epoxy is cured we set the bubble so that it is centered. We keep checking the bubble till the epoxy is cured. Of course, human is prone to error. and every degree of error will be doubled when the lower rail is rotated by 180 deg. The bubble in NN5 has a sensitivity of 34' (0.57deg) per 2 mm movement. To set a tolerance of 1mm bubble movement, we need 34'/4 (remember 180 deg rotation doubles the movement)=0.14 deg of accuracy which is pretty high for my OWN standard.
You know what, even the instrument grade bubble will shift every time before use. Recalibration is needed. I guess it has something to do with thermal expansion. But then instrument grade bubble is much more sensitive.

We can make the bubble adjustable, say by making a bubble holder which is mounted to the lower rail by 3 screws. By adjusting these 3 screws, the level can be adjusted. Then I know 99.7% users will not use this feature and 99% users will have problem in calibrating the bubble. I am not sure how many people will justify the added cost. and not to speak the danger of failing to restore acceptable leveling after attempt of calibration. I can almost hear their complaint. "Damn, stupid Nick! Why not set and glued the bubble as other manufacturers do."

Nick

Terrywoodenpic
10-16-2009, 02:27 PM
Nick

you are 100% right.... but this is something that will continue to haunt every one who fits bubble levels.
If you went to the trouble of offering an adjustable bubble, people would not want to pay the price, and even fewer would use it.

If they actually understood how their software functioned they would not worry at all about bubbles.

bigwade
10-16-2009, 03:13 PM
Hi Folks,
I'm a kind of level/bubble freak... :-)
I like to collect all kind of those tiny acryl bubbles.

After reading this topic, I could not resist :-)
In our kitchen we have a glass induction plate for cooking.
Guess what ? It is in Level !!

So I put all the NN's I have on the same place at the plate without a camera.
NN3/NN5/NN180/NN R1/R10
The R 10 has an original ext. NN bubble, the ring bubble.
The R1 it's own + an original ext. NN bubble.
I rotated all NN's in 4 steps about 90 degr.
At ALL NN's, the bubble stayed in the black ring EVERY 90 degr.
The R1 AND R10 are the absolute winners, followed by NN180 and NN3.
The NN5 was at "last" position, bad ? NOT AT ALL !! it was just a tiny, tiny difference.
So Nick Don't worry ! It's not the rig ! All are more or less spot on.
No need for an adjustable bubble.

Now what can cause a running bubble if it's not the NN ?
A) a not so sturdy floor like carpet, grass or sand.
B) a not so sturdy centre-column of the tripod
C) both !

About "A" that's clear I guess.
With a camera hanging out of balance on a NN3/5 one leg of the tripod has more force to handle.
About "B" With a camera hanging out of balance on a NN3/5 the force on the column is huge and with a not so well guided column you get flex and hey.........the running bubble.
Fanotec can make a BATCH HEAD but without a sturdy, well guided centre column-tripod it is not going to work.
Good tripods I know, Gitzo/Feisol
Problem solved.

Last point;
I tried a NN3 with 5D2 on a carbon monopod mounted upside down on a Heavy but small Gitzo.
After 1 rotation (90degr) I had to wait before it was not moving again.
Tried the same with the R1 and that was much more balanced because the camera is centered ±ON the rotation point and it will only hang backwards a bit.
It's all a matter of weight and place of the weight, direct on top or forces via L-bow

grtz
Frank

bigwade
10-16-2009, 04:54 PM
Ah, forgot one thing..
D) the flex of the rotator...... :-)

The R-D4/8 etc series are much better than the standard NN3/NN5 rotators, fact.

realman10
10-17-2009, 03:46 AM
If Bigwade's theory were right I could never find a "right" position for the bubble, but this is not the case: I have found the right mark on the bubble which is close to the black circle and when the bubble is located there, it will stay there during the a full (slow) rotation. I have a good tripod but nothing exotic. Obviously rigidity is not an issue in my case at least.

But I'm happy now and I agree with Nick, making an adjustable bubble will create more problems than it solves.

As for the other friend who wrote that we would not bother this bubble issue if we understood how the software works I choose the Gandhi way: peace and love to you!

Wim.Koornneef
10-17-2009, 06:06 AM
Hello Mike,

As Terry pointed out a moving bubble isn't an issue when you know how to use your stitcher,
As Nick pointed out the precision of the bubble level of NodalNinja products is pretty good,
As Frank pointed out there are more reasons why a bubble can move,
As I pointed out if a bubble movement really bothers you then you can find the real center of the bubble level to make it precise.

IMO all is true ;-)

In this thread is not mentioned yet when a precise bubble level is needed and when not.

* When stitching panos without setting horizontal or vertical CP's, or other ways of leveling in the software, then a precise bubble level is needed to get a fine leveled horizon.

First some math to get an idea about the impact of an error of a bubble level off set.
A bubble level off set of 0.5 degree will cause a variation of 2*0.5=1 degree over 180 degree (=half the width of a pano).
The impact of this on the horzion of a pano of 7200x3600 px will be a slope of 3600/180<>20px at the horizon.
When the view angle of the pano player is 75 degree then the slope of the horizon in the players window will be 75/360*20<>4 px.
IMO a slope of 4 px of the horizon is hardly noticable and so acceptable.

For larger panos the slope can be larger but in practice this is hardly the case.
See for example this 9600x4800 px pano that is shot at the shores of a large lake with a carefully leveled R1, despite the fact that there are no horizontal CP's set, you will hardly see any sloping of the horizon:
http://www.dmmdh.nl/panos/ouddorp_harbor_18092009/image_002.html

* In all other situations, so when the pano is leveled with the software, the precision of the bubble level, or the lack of precision, is not of any impact on the quality of the pano as long as the proper level is manually set in the software.
See for example this 8400x4200 px handheld pole pano:
http://www.dmmdh.nl/panos/pole_pano_harbor_ouddorp_29052009/image_014.html

Conclusion: both examples proves that you don't need a "perfect" bubble level with zero tolerance to get a good leveled pano.

BTW, I can confirm the results of Frank's tests, the bubble levels of the NN products he mentioned are all within the precision needed for batch stitching(!).
(!) With all mentioned NN products you can get good results with a EasyStitch method. This method is based on a calibrated lens template, no horizontal or vertikal CP's are placed and only the lens shift and image params are optimized in an automated batch stitch process.

I hope this helps to take away any confusion or misunderstanding.

Wim

bigwade
10-17-2009, 10:40 AM
If Bigwade's theory were right I could never find a "right" position for the bubble, but this is not the case: I have found the right mark on the bubble which is close to the black circle and when the bubble is located there, it will stay there during the a full (slow) rotation. I have a good tripod but nothing exotic. Obviously rigidity is not an issue in my case at least.


Ok?©, if you have a sturdy tripod on a sturdy floor, rotating, the bubble is at the same place you are correct.
If you have problems with that ask Bill or your local reseller to replace your underrail.
BTW it's not theory but just what I see with a good and not so good tripod.
Even with the "not so good tripod" or monopod on a tripod with all the bubble-walking I see, I get perfect final results.
grtz
Frank

realman10
10-17-2009, 01:52 PM
Slightly OT: Wim I also use something similar to your EasyStitching, but the shift correction is part of my lens calibration data (which I freeze for one specific focusing distance, 1m @f/10 basically the hyper-focal distance). Normally I don't optimize the shift parameter for each project.
I mostly use a D3 with a rectilinear 14mm Sigma which has excellent resolution and geometric stability (but is weak on flare and vignetting).
I just load the images, load the template and the panorama is 99% ready BEFORE any optimization and CP work. Since I received my NN5 things have improved further and now every project is a breeze. I even did a high resolution job with a 50mm lens and again it was a piece of cake...Thank you Ninja...and Nick!

Wim.Koornneef
10-18-2009, 04:18 AM
Hello Mike,

In my EasyStitch method I also optimize for the lens shift params because on my cameras (Canon) there is a very little movement possible of the lenses in the bajonet mounts.
Although not much, approx. 0.2mm in all directions, this small movement can cause a shift of the projection of the image on the sensor of approx. 30 px (depending on the size and resolution of the sensor it can be more or less).
When it occurs and the template is not optimzied for lens shift the small movement will cause visible errors.

Wim