PanoTools:
re: Re: Is the Manfrotto 300n accurate enough for template stitching? + tripod Question.
Stuart Milne 2006-Jun-04 20:56:21
Wow Hans, that is some good work there :-) I'll add some thoughts to the fire, though not all directed at you Hans, but more general interest for the list I hope.
One thing I'd like to add to what has been said though, is that I think looking at optimiser values is almost looking from the wrong direction. I'll try to explain.
As we have sticthed many panos, and played with the numbers, we have seen that dialing in similar numbers will have little effect on the optimisation error. E.g. sometimes for a pano that optimises to 0 60.1 120.1 179.8 etc can be overridden to 0 60 120 180 etc and there is almost no change in optimisation ave, or max error.
This is due to the accuracy of c/p placement, and opimisation routines (normally gradient descent IIRC, and are all done so to an arbitrary 'this is close enough' point in the maths!). Just becasue an optimiser gives you a number, does not mean it is the correct one.
As such, my preference for seeing if a panohead is accurate, is to try an image overlay (take shot, spin head, return to first postion and take similar shot). If these images all line up around the pano then you KNOW you have repeatable results, and it is then a case of getting the optimiser to give the correct numbers!
Also, although it is 'nice' to have numbers like 0 60.00 120.000 show up, it really is arbitrary what they are, so long as they are repeatable - I wouldn't care if a head shows 0 59 118 etc, so long as the repeatability was top notch (fwiw, all our heads are more accurate than that, but the repeatability is what we have mainly designed for :-)
The upshot to all this.......... is that I have no idea is the Manfrotto 300N is accurate enough having never seen one! (Though as it gets the thumbsup from Hans it may well be worth a shot :-)
#removed#
PS Hans, from numbers it looks like you are right, one was almost certainly shot on the wrong stop ;-)
--- In #removed#, "mshakeshaft" <msphoto@...> wrote:
>
> Is the Manfrotto 300n accurate enough for template stitching? At the
> moment I am using a Nodal Ninja 2 which is great, but the accuracy of
> the rotator is a bit hit and miss (I notice that it has been improved
> with the latest version). I was thinking about attaching the NN arm to
> the 300n rotator base.
There is not much difference in the accuracy if you use the 300N carefully.
To show this I made some statistics about both the 360P and the 300N
First the 360P template sets from their site.
By setting controlpoints and optimizing the Yaw I can get the actual Yaw.
360P - B&O D70+10,5mm
0-60.09-120.25--179.95--119.87--59.98
0-60.14-119.98-179.99--120.05--59.85
0-59.74-119.59-179.74--120.22--60.17
0-60.13-120.18--179.94--119.74--60.07
360P - Oxford D200+10,5mm
Template from the site.
0-60.14-120.32--179.63- -119.70- - 59.88
Panos real Yaw optimized
0-60.17-120.32- -179.74- -119.77- -59.96
0-60.75-121.02- -179.04- -119.04- -59.42
0-60.08-120.20- -179.56- -119.83- -59.95
0-60.06-120.29- -179.78- -119.81- -59.88
0-60.01-120.02- -179.96- -119.99- -60.09
As you can see they are very accurate but one in the D200 series is different.
My guess is that Matt had forgotten to start at Zero on this pano.
Actually I did some tests which showed that by setting the anchor on another image I
could find one which gave me the same Yaws as the other ones.
Now to the 300N.
First ones are random panoramas made with Canon 20D and 10,5mm during 6 months.
They are not even with the same bracket.
Not as exact as the 360P but enough for templates.
0-59.97-119.97-179.92--119.83--59.88
0-60.03-120.12--179.76--119.51--59.75
0-60.32-120.49--179.43--119.58--59.78
0-60.14-120.66--179.36--119.73--59.51
0-60.71-120.26--179.92--120.03--59.87
0-60.32-120.92--179.44--119.51--58.65
0-59.88-119.82-179.99--120.00--58.35
This is 4 panos made at the same time with same head on the rotator.
You can see 3 of them here.
http://www.panoramas.dk/dr/
0-45.01-89.93-134.94--179.82--134.7--89.80--45.02
0-45.01-89.83-134.97--179.88--135.05--90.13--44.98
0-45.10-90.09-135.17--179.79--135.00--89.96--44.97
0-45.10-90.07-134.90--179.88- -135.00--90.16--45.05
This is from a set of panoramas with Sigma 8mm which was batched with a template.
0-90.2--179.8--89.8
0-89.8--179.9--90.1
0-90.1--179.6--89.7
0-90.3--179.9--89.6
0-90.1--179.8--89.8
For template stitching the rotation is the least significant.
It needs to be within 0,5 degrees.
The pitch and rolls are much more important.
An error of 0.1 degree is enough to give stithing errors.
There may be some differences in the production series of the 300N.
I have 2 heads and have until now only used the first one but I recently discovered that the
second was not starting at Zero on all clickstop sets.
It is also possible that the head will be less distinct in the clicks by time.
It can be helped by getting a new clickstop screw.
But I have to say they are slow at Manfrotto. I had to wait 6 months on getting one of these
as a sparepart.
Also the heavier bracket you put on it the more difficult it is to feel the clickstops.
Hans Nyberg
www.panoramas.dk
2.000.000 visitors first 5 months of 2006.
SPONSORED LINKS
Multimedia authoring software Multimedia slide show software Multimedia software Multimedia presentation software
----------------------------------------
YAHOO! GROUPS LINKS
Visit your group "PanoTools" on the web.
To unsubscribe from this group, send an email to:
#removed#
Your use of Yahoo! Groups is subject to the Yahoo! Terms of Service.
----------------------------------------
[Non-text portions of this message have been removed]
------------------------ Yahoo! Groups Sponsor --------------------~-->
Home is just a click away. Make Yahoo! your home page now.
http://us.click.yahoo.com/DHchtC/3FxNAA/yQLSAA/.Cr1lB/TM
--------------------------------------------------------------------~->
<*> PanoTools.Info Wiki
http://wiki.panotools.org
<*> PanoTools.Info List Archives
http://lists.panotools.info/pipermail/ptml/
<*> PanoTools List Reader @ GMane:
http://news.gmane.org/gmane.comp.graphics.panotools
Yahoo! Groups Links
<*> To visit your group on the web, go to:
http://groups.yahoo.com/group/PanoTools/
<*> To unsubscribe from this group, send an email to:
#removed#
<*> Your use of Yahoo! Groups is subject to:
http://docs.yahoo.com/info/terms/