atomicmak
Posts: 173
Location: ahmedabad, India
Registered: 7 Dec 2006
|
need help finding nodal point of new lense
Posted: 9 Apr 2008 at 13:02 GMT
|
|
i bought new Nikon 17-55mm f/2.8G ED-IF AF-S DX
now i would want to have its nodal point settings for my nn3 which i run hard to find.
can anyone help me on what would be nn3 upper and lower rail for this lense ?
i am in need of taking still panoramics and for that i bought this lense. i do have 12x24 nikon lense too
suggest me plz.
mak
|
|
alert moderator
|
|
gus
Posts: 381
Location: United Kingdom
Registered: 19 Jun 2007
|
|
Matt Rogers
[360 Precision]
Posts: 209
Location: Oxford UK, United Kingdom
Registered: 16 Jun 2005
|
Re: need help finding nodal point of new lense
Posted: 9 Apr 2008 at 16:04 GMT
|
|
That's a great guide but the funny thing is you can clearly see misaligned sections in the images with green arrows. Particularly in the vertical image samples.
Matt
|
|
alert moderator
|
|
Robert Piontek
Posts: 97
Location: Germany
Registered: 26 Jul 2007
|
Re: need help finding nodal point of new lense
Posted: 10 Apr 2008 at 7:27 GMT
|
|
|
I had an idea for aligning the other day. I have used the scotch tape on window technique. It occurred to me that it would be much easier to tape a cloth or paper tape measure across the window. Haven't tried it yet, though.
|
|
alert moderator
|
|
John Houghton
Posts: 2259
Location: Hitchin, United Kingdom
Registered: 17 Jan 2005
|
Re: need help finding nodal point of new lense
Posted: 10 Apr 2008 at 13:54 GMT
|
|
Robert Piontek said: It occurred to me that it would be much easier to tape a cloth or paper tape measure across the window.
Robert, I've done that. You can measure the parallax shift in mm with the camera at two positions on the upper arm of the pano head - the positions being either side of the NPP, preferably. E.g. at the 9cm and 10cm scale positions, say. You can draw a simple straight line graph to work out the zero shift position (where the line crosses the axis). Or you can work it out by arithmetic if your maths is up to it! It's quicker than shifting the camera backwards and forwards and taking photos. However, I now generally prefer the even quicker way of directly observing the movement of the entrance pupil from the front of the lens.
John
|
|
alert moderator
|
|
Rosauro
Posts: 233
Location: Toronto, Canada
Registered: 15 Dec 2006
|
Re: need help finding nodal point of new lense
Posted: 17 Apr 2008 at 5:33 GMT
|
|
Matt Rogers said: That's a great guide but the funny thing is you can clearly see misaligned sections in the images with green arrows. Particularly in the vertical image samples.
Gee Matt, funny how the image still stitched well.
You've been on the negative lately against people just trying to help. You ok?
|
|
alert moderator
|
|
Matt Rogers
[360 Precision]
Posts: 209
Location: Oxford UK, United Kingdom
Registered: 16 Jun 2005
|
Re: need help finding nodal point of new lense
Posted: 17 Apr 2008 at 8:13 GMT
|
|
Yeah, I'm fine. Just a bit worn down having to answer the same tech support queries over and over from people have learnt the wrong info online, that's all. Plus also having to support PTGui, an application that we neither sell nor have the opportunity to sell. It's really a catch-22 as people need to be able to use the software correctly to maximise their usage of our products. And it's a bit rough telling them to bugger off after investing in our system. There are obvious solutions to these problems but it's just a matter of finding the time to implement them.
Whether an image will stitch or not more often than not has nothing to do with the correct entrance pupil. I once stitched a set of panoramas where the customer had actually put the camera on the head facing the wrong direction. The entrance pupil must have been out by 180-200mm. But if you set manual control points only, the images will still stitch. The final panos did require some touch up in photoshop to remove slight ghosting in the overlap regions.
Most cases of poor stitching in PTGUI come from an incorrect optimisation procedure where people fail to delete control points with large maximum distances and then re-optimise. If you have a set of well calibrated lens parameters and your maximum control point distances are less than 1px per 10,000px then you won't have stitching problems. Part of the problem is PTGUI reports the values are "very good" when in-fact they're not really "very good", I don't stitch an image or take a template as final unless it says "too good to be true". I'm not sure where this optimisation rating system came from but it's a pretty poor indication of your final result.
On average I can get emailed 5-10 PTGui files with image sets to look at per month. I'm often shocked to see what people think are accurate control points. I try to explain that if your maximum CP distances are 3 pixels then you can expect to see stitching errors around 3 pixels in width. This isn't always the case but can be taken as a very general rule of thumb. With an Adjuste for example at the supplied settings you should be aiming to have your maximum control point distance around 0.6-0.7 pixels to ensure consistently accurate results.
In the case of your images if PTGUI does set control points in the misaligned areas and you optimise correctly they'll get deleted anyway. What's far more important than the correct entrance pupil is having a consistent entrance pupil, whether it's correct or not doesn't always matter. If you have a pano head that flexes even slightly then you can end up having the entrance pupil translate in the 3D space the head rotates in. This translation of the entrance pupil presents much larger problems then simply having it set incorrectly but the same for each photo.
Before some people spend as much as they do on our heads they do a lot of research and obviously ask a lot of questions. More often than not they comment that they can find and set the correct entrance pupil with a cheaper head just as easily. This misses one of the main points of our products though and that's the repeatability of each shot between sets of panoramas. It's this precision and repeatability that's crucial in minimising ones workflow. Plus the ability to accurately re-set the head to within 0.1mm of previously used settings.
Matt
|
|
alert moderator
|
|
Matt Rogers
[360 Precision]
Posts: 209
Location: Oxford UK, United Kingdom
Registered: 16 Jun 2005
|
Re: need help finding nodal point of new lense
Posted: 17 Apr 2008 at 8:26 GMT
|
|
You won't find them as the entrance pupil for this lens will be out of range by quite a large margin on the NN3. On a NN5 you could probably use the lens at 50-55mm only.
Matt
|
|
alert moderator
|
|
John Houghton
Posts: 2259
Location: Hitchin, United Kingdom
Registered: 17 Jan 2005
|
Re: need help finding nodal point of new lense
Posted: 17 Apr 2008 at 16:20 GMT
|
|
Matt, Your quality standards are admirable. Most of us on this forum, alas, don't own equipment of the calibre of your products. I haven't so much as touched a 360Precision head, but since you thoughtfully provide sets of images on your web site and even a PTGui template, it's at least possible to try one's hand at stitching to see if sub pixel accuracy is achievable. I therefore downloaded a couple of sets of Oxford Market images (4 & 5). Since accurate template stitching is claimed, that is what I tried first.
I was disappointed with the results, (following the simple instructions provided), since both stitched panoramas showed all too visible stitching errors. Here are a couple of samples from set 5 at full size:
These fall far short of your quality standards and I feel there must be a simple explanation. They certainly wouldn't impress potential customers visiting your site - at least I hope they wouldn't. The instructions say "Control points are not welcome here". Maybe not, but the results are rather better with than without them.
John
|
|
alert moderator
|
|
Hans Nyberg
Posts: 803
Location: Denmark
Registered: 28 Aug 2005
|
Re: need help finding nodal point of new lense
Posted: 17 Apr 2008 at 17:14 GMT
|
|
Well
Obviously this is not the correct template.
The PTMac template is correct and stitches perfect.
Hans
|
|
alert moderator
|
|
Matt Rogers
[360 Precision]
Posts: 209
Location: Oxford UK, United Kingdom
Registered: 16 Jun 2005
|
Re: need help finding nodal point of new lense
Posted: 17 Apr 2008 at 18:28 GMT
|
|
Thanks for pointing this out. For some reason when we moved the website over to the dedicated server (2-3 weeks ago) an old test version of the PTGUI template was copied across.
The problem is the d and e offset values are specific to the image dimensions. The version of the PTGUI template that you downloaded was created for the test images at 1472px x 2200px whereas the final image sets that you download are 2592 x 3872.
As Hans pointed out the PTMac template is the correct one, if you give me an hour I'll upload the correct PTGUI template.
Matt
|
|
alert moderator
|
|
Nick Fan
[NodalNinja]
Posts: 276
Location: Hong Kong
Registered: 26 May 2006
|
Re: need help finding nodal point of new lense
Posted: 18 Apr 2008 at 2:37 GMT
|
|
Hi Matt,
a business idea for you. why not sell a bundle of you 360P absolute and calibrated DSLRS with lens for some popular combinations? This will help the newbies a lot. With pre-made template, they will only be responsible for a few push on the shutter button. And it will save you lots of trouble in supporting products you don't sell. what do you think?
Nick
|
|
alert moderator
|
|
Matt Rogers
[360 Precision]
Posts: 209
Location: Oxford UK, United Kingdom
Registered: 16 Jun 2005
|
Re: need help finding nodal point of new lense
Posted: 18 Apr 2008 at 6:24 GMT
|
|
Hi Nick,
Well we do this on demand for select customers. The main problem becomes one of tech support and cost. Initially we did offer a free template service but this quickly got out of hand. If an individual customer spends 3-4 hours setting up, shooting and making a template that's no big deal. But if we have to spend that time ourselves setting up systems and making templates it quickly gets out of hand. I'd rather try and stick to the core business of developing and selling pano-heads.
We have 4-5 new products coming out this year which takes up most of our time. Maybe when the new products are established in the market we can take another look at selling full systems.
Matt
|
|
alert moderator
|
|
back to threads list |
this thread is closed
