Documentation help needed for R1 [Archive] - Nodal Ninja Forum

PDA

View Full Version : Documentation help needed for R1



clavi
12-12-2009, 07:01 AM
Having bought R1 for more than a month, I find the documentation absolutely miserable and really needs improvement.

In the camera settings page on your website, there is a decent list of lenses listed, but the settings are mostly empty. Being a manufacturer, I think it is essential to provide a complete set of documentation, not something mostly empty.

There is so much guesswork involved, and nobody is sure what is the correct setting even with common lenses such as Sigma, Nikon, or Canon. There is also no proper explanation on the purpose of tilt and roll too. E.g. what is nadir and zenith? (I had to google for nadir/zenith) How will tilting affect nadir and zenith? Should I keep my camera horizontal, vertical or roll it to 60 deg? What is the output difference between horizontal vs vertical vs roll?

What is lacking is a _complete_ settings table that would look something like this:
Lens make & model; Camera Make; Format (FF/APS); No. of shots; Tilt; Roll; Setting

Each line in the table should cater specific format, instead of lumping it together as found in your documentation "3 shots around (on full frame sensor or APS sensor with 60 deg roll)"? So does the 60 deg roll refer to APS sensor only or both APS and FF?

The R1 is a very well-made product with great finishing, but too bad I had a hard time using it.

nick fan
12-13-2009, 12:23 AM
Hi Clavi,

I apologize for your frustration. Documentation is our weak point. It has always be a work in progress.
However, if you need help for your specific setup. We can help you more easily. What is your camera and lens in use?

Panography is an immense subject, there a lot of excellent tutorials on the web
http://www.johnhpanos.com/tuts.htm
http://www.rosaurophotography.com/html/technical.html
http://www.panoguide.com/howto/


Nick

clavi
12-13-2009, 07:21 AM
Thanks for the web links. However, they are never specific to R1, probably something near to NN5. It is still essential to have your own content, specific to your own product/s.

I have a Nikon D700 with Sigma 8mm f/3.5, and I use EasyPano's Panoweaver to perform stitching. So far, I have tried out many settings but none seems perfect. In another forum, I have been advised to use 3 shots, 5 deg tilt at 24mm which is pretty close to perfection (still some "breaks" observed).

Hopefully, your documentation will be the first New Year resolution for year 2010.

nick fan
12-13-2009, 07:44 AM
Thanks for the web links. However, they are never specific to R1, probably something near to NN5. It is still essential to have your own content, specific to your own product/s.

I have a Nikon D700 with Sigma 8mm f/3.5, and I use EasyPano's Panoweaver to perform stitching. So far, I have tried out many settings but none seems perfect. In another forum, I have been advised to use 3 shots, 5 deg tilt at 24mm which is pretty close to perfection (still some "breaks" observed).

Hopefully, your documentation will be the first New Year resolution for year 2010.


R1 is a single row pano head, you can imagine it has one axis of adjustment set. You only need to adjust the to-and-fro axis. So follow the tutorials, ignore the adjustment for the lower horizontal rail.

Have you try the settings provided?
Sigma 8mm f/3.5 DG for Nikon
3 shots around (on full frame sensor, or APS sensor with 60? roll)
Tilt angle----NPP value
+7.5?---------0.50
+5?------------0.80
0?--------------1.50

The values are tested by myself. They should be at least accurate to 0.1cm. If you can't get good stiching, upload your source photos for us to try.

Nick

PS
I see you get the "recommended" setting here
http://www.easypano.com/forum/display_topic_threads.asp?ForumID=10&TopicID=4867&PagePosition=1
That is completely irrelevant. Smooth is a Canon guy. The settings for R1 is lens mount specific.

clavi
12-13-2009, 08:17 AM
Do I need to setup my cam body as portrait or landscape or roll 60 deg? What camera do you have?

Smooth was just trying to help, and he did state upfront he's on Canon. He has no obligation to help as he isn't the maker. In fact, he recommended your products and I trusted his recommendations. That's why I here today. :001_smile:

nick fan
12-13-2009, 08:28 AM
Do I need to setup my cam body as portrait or landscape or roll 60 deg? What camera do you have?

Smooth was just trying to help, and he did state upfront he's on Canon. He has no obligation to help as he isn't the maker. In fact, he recommended your products and I trusted his recommendations. That's why I here today. :001_smile:


Sigma 8mm is a circular fisheye on full frame sensor. No matter which angle your roll, you get same result. A circle is a indistinguishable whether it is in portrait/landscape/ or 60 deg rolled. :wink:

Nick

John Houghton
12-13-2009, 08:36 AM
Nick is right. Panoweaver should be able to cope with the camera at any angle, but if you still have stitching errors using the setings that Nick provided, do as he suggests and upload the set of jpeg images somewhere and we can try stitching with PTGui and see if there is anything amiss. If you don't have your own private web space, use a free service such as http://www.filedropper.com (http://www.filedropper.com).

John

clavi
12-15-2009, 07:26 AM
Ok, I tried using 5 deg tilt at 0.8cm, and there are still breaks when I use Panoweaver.
http://www.filedropper.com/dsc2790
http://www.filedropper.com/dsc2791
http://www.filedropper.com/dsc2792

I also tried 0 deg, but I will upload them tomorrow. Thanks.

John Houghton
12-15-2009, 09:34 AM
I had no trouble stitching with PTGui. This is the result:

http://www.johnhpanos.com/DSC_2790_Panorama.html

I'll maybe also try Panoweaver.

John

John Houghton
12-15-2009, 09:47 AM
I did notice a small glitch in the panorama on the top edge of the pews. This is the result of parallax, as shown here:

http://www.johnhpanos.com/DSC_2790_Panorama.gif

The head does need a little adjustment.

John

John Houghton
12-15-2009, 10:32 AM
I did try Panoweaver, but didn't manage to get a result as good as PTGui's. I added several extra control points (as I had to do in PTGui also) at the top and bottom but couldn't eliminate a number of small stitching errors. I'm not an experienced Panoweaver user though.

John

clavi
12-15-2009, 06:46 PM
John: How does parallax error occur in R1? How do we eliminate it?
Thanks.

nick fan
12-15-2009, 11:17 PM
Hi John,


Thanks for the stitch. I just tried it and got poorer result with a few breaks at the ceiling.

Hi clavi,

A fisheye lens has no fixed NPP. NPP changes with angle of view.
http://michel.thoby.free.fr/Fisheye_history_short/Beyond-the-pupil.html
We are just trying to get the least parallax. For a 3-shot single row pano, the seams at the horizon is 120 deg apart while the seam at the zenith is ~180 deg apart. This means that NPP for horizon and zenith are different and can't be removed simultaneously. You can reduce the parallax at the horizon at the expense of zenith and vice versa. The NPP data is optimized for the horizon where we are most interested in.

Your pano is a very challenging one. It has lots of structured lines at the ceiling with little distinctive features. This type of pano is best taken with a spherical pano head like NN3/5 which can take a separate zenith shot at NPP. With skills, a single row pano head with a lens having 180 deg vertical angle of view will be good for >95% panos. The <5% will need a spherical pano head.
If you use ptgui, you can try to take a zenith shot on R1 and add it to the project. You can also try a setting of 1.65 which will reduce parallax at the zenith.


Nick

John Houghton
12-15-2009, 11:49 PM
Parallax occurs because of the shift in position of the entrance pupil from shot to shot. As it shifts to the left or right, near objects will appear to move to the right or left, respectively, with respect to whatever's in the background. It's exactly like the effect seen when you hold up your finger and move your eye from side to side. To eliminate parallax, what you need to do is eliminate movement of the entrance pupil when you shift the camera from one shot to the next. This is very easy to do with your DSLR and fisheye, because you have the huge advantage of being able to see the entrance pupil merely by looking into the lens. You will need to hold the camera so that the eyepiece at the back of the camera faces a bright light or window. Light can then enter the eyepiece and be reflected down by the prism and forwards by the mirror through the aperture and lens. So what you see when you look into the lens is the bright spot of the aperture. So, the method of adjusting the panohead becomes very simple:

1. Position the head so that the camera is half way between two detent positions that will be used when shooting a panorama, and with the back of the camera facing a light or window so that the entrance pupil is visible as a bright spot.

2. Swing the camera clockwise to the position at the +60 degree detent position (i.e. half of the normal panorama shoot increment of 120 degrees).

3. Check the position of the entrance pupil. If it has shifted to the left (as seen from the front), then it must be in front of the rotation axis. If it has shifted to the right, then it must be behind the rotation axis.

4. Now slide the camera forwards or backwards to bring the entrance pupil back to the original position when the camera was pointing straight ahead.

5. Now check that the entrance pupil is in the same position when the head is at yaw positions +60 and -60. There might be a tiny difference, but adjust the head to minimize this.

That's all there is to it. The one thing you need to have is a means of judging the position of the entrance pupil. You need to improvise a sighting device so that your line of sight is directed roughly along the optical axis of the lens. Use an empty ball point pen barrel positioned so that you see the entrance pupil centered in the hole at the end, or make something like a rifle sight using a strip of cardboard and a needle, as shown in http://www.johnhpanos.com/epcalib.htm.

There are other methods you can use, but this works very well and is simple and quick. (For rectilinear lenses, you need to stop down to f/22 and use the depth of field preview button to get a small spot).

John

John Houghton
12-16-2009, 12:05 AM
Thanks for the stitch. I just tried it and got poorer result with a few breaks at the ceiling.


Nick, I've uploaded my project file if you're interested: http://www.johnhpanos.com/DSC_2790_Project.zip.

John

clavi
12-16-2009, 06:16 AM
Nick: So that explains why I always get "breaks" in the ceiling and floor, because we can only optimize settings for horizontal view? And are u saying we can eliminate the ceiling and floor breaks by taking another 2 shots (one zenith and one nadir)? Thanks.

John: Well, with the Ultimate R1 and a known/common lens (namely Sigma 8mm f/3.5 Nikon mount), the given settings (by Nick) should already be free from parallax errors (only for horizontal view), isn't it? And in your article, you mentioned that the ideal should be 4 shots plus nadir and zenith (that's 6 shots in total). Could it be done with 3 shots plus nadir and zenith? Thanks.

nick fan
12-16-2009, 07:10 AM
Nick: So that explains why I always get "breaks" in the ceiling and floor, because we can only optimize settings for horizontal view? And are u saying we can eliminate the ceiling and floor breaks by taking another 2 shots (one zenith and one nadir)? Thanks.

If you take 4 shots around, plus zenith and nadir, then the seams are 90 deg to each other. In this way, the parallax is minimized in all seams.
if you take a separate zenith shot, you can get the complete ceiling in one shot, then the seams around zenith will only contain walls/ windows which are much easier to blend/ correct than the ceiling. Besides, you have more overlap around the zenith, it is easy to use alpha channel (for ptgui) to clone out area of misalignment.



John: Well, with the Ultimate R1 and a known/common lens (namely Sigma 8mm f/3.5 Nikon mount), the given settings (by Nick) should already be free from parallax errors (only for horizontal view), isn't it? And in your article, you mentioned that the ideal should be 4 shots plus nadir and zenith (that's 6 shots in total). Could it be done with 3 shots plus nadir and zenith? Thanks.


This settings are giving by some one who has just made a few panos in his complete life. :biggrin: The reason I made all NN adjustable is because I don't trust myself. You can make adjustment based on my initial value. To be proficient in panography, you need to learn how to do that yourself.
BTW, I just realize that if you install the lens ring according to the manual, you should rotate the camera to portrait mode instead of landscape. The opening of plastic insert of lens ring (for easy installation) may cause the lens axis to tilt a bit. It is best to point it down. This is my orientation when I got my NPP data.


Nick

John Houghton
12-16-2009, 08:52 AM
So that explains why I always get "breaks" in the ceiling and floor, because we can only optimize settings for horizontal view?
Not necessarily. There are no breaks in the ceiling and floor in my stitch of your images.

John

nick fan
12-16-2009, 09:12 AM
Not necessarily. There are no breaks in the ceiling and floor in my stitch of your images.

John



Hi John,

Do you think it is because you forced ptgui to change the lens parameters by adding control points to the zenith and nadir area?

Based on your experience, if you fix the lens parameters (say to what you got from sky map),
http://www.johnhpanos.com/starcal.htm
by just optimizing the fov, y, p, r, shifts, viewpoints, can you eliminate the misalignment?





Nick

John Houghton
12-16-2009, 10:08 AM
Do you think it is because you forced ptgui to change the lens parameters by adding control points to the zenith and nadir area?

I encouraged the optimizer to evaluate the lens distortion correction parameters as accurately as possible by making good the deficiencies in the control point generation. PTGui often fails to find matching features at the zenith and nadir because of the extreme distortion, but the optimizer needs points all along the seam if it is to align the images well.


Based on your experience, if you fix the lens parameters (say to what you got from sky map),
http://www.johnhpanos.com/starcal.htm
by just optimizing the fov, y, p, r, shifts, viewpoints, can you eliminate the misalignment?

Well I tried that, using my template for the 40D. The stitch didn't look any worse. I couldn't find any stitching errors, but the optimization report wasn't quite as good. Optimization will inevitably be affected by parallax errors, depending on precisely where control points happen to be assigned. The ceiling is probably relatively far away, so is less affected by parallax.

John

nick fan
12-16-2009, 07:14 PM
Hi Clavi,

As John indicated DSC_2791 and DSC_2792 reveal parallax at the pew,
http://www.johnhpanos.com/DSC_2790_Panorama.gif

The pattern indicate that the entrance pupil (NPP) is in front of the rotation axis. So a larger value >0.8cm should be used. But this can also be caused by tilting of lens axis in the lens ring. So try this
1. repeat the shots with camera in portrait orientation with rail setting at 0.8cm
2. repeat the shots at 0.85cm, 0.9cm, 0.95cm and 1cm

Bring a piece of paper, jot down which sequence uses with setting.

Try them in panoweaver. upload a zip of each sequence for me to try in ptgui.


Nick

clavi
12-17-2009, 04:22 PM
This settings are giving by some one who has just made a few panos in his complete life. :biggrin: The reason I made all NN adjustable is because I don't trust myself. You can make adjustment based on my initial value. To be proficient in panography, you need to learn how to do that yourself.

BTW, I just realize that if you install the lens ring according to the manual, you should rotate the camera to portrait mode instead of landscape. The opening of plastic insert of lens ring (for easy installation) may cause the lens axis to tilt a bit. It is best to point it down. This is my orientation when I got my NPP data.

Nick


As your end-user, I would have expected you had tested your settings thoroughly with experts. (for that matter, I ain't the expert :001_smile:) So we end-users having bought your product have absolute faith that you will supply us with the correct settings right from the start. Would you buy a device (printer, dvd-writer, etc) knowing that the manufacturer did not supply users with a proper driver? I don't think so. :001_smile:

Anyway, it does not matter whether it is portrait or landscape mode, there are still "breaks" detected in either modes. But I will take note of your advice.

clavi
12-17-2009, 04:29 PM
Not necessarily. There are no breaks in the ceiling and floor in my stitch of your images.

John



Whether you encouraged PTgui optimizer in a certain setting is one thing, from a technical discussion, each R1 setting will either be biased towards the horizontal view -OR- nadir/zenith view?? Isn't that what Michel Thoby is trying to convey?

DemonDuck
12-17-2009, 09:13 PM
As your end-user, I would have expected you had tested your settings thoroughly with experts. (for that matter, I ain't the expert :001_smile:) So we end-users having bought your product have absolute faith that you will supply us with the correct settings right from the start. Would you buy a device (printer, dvd-writer, etc) knowing that the manufacturer did not supply users with a proper driver? I don't think so. :001_smile:

Anyway, it does not matter whether it is portrait or landscape mode, there are still "breaks" detected in either modes. But I will take note of your advice.


If Bill and Nick had to spend the time to calibrate and catalog each of their pano heads for every camera/lens combination, there would have to be a price increase.

It's not hard to find the right settings and orientation for your own particular set up yourself. And it is a useful study to learn the technique. It takes less time than the time you've spent writing back to this list for help. And it's even fun and gives you a sense that you are in control of your own domain. And if you make a mistake, you'll learn from it.

And you will also learn the limits of you stitcher software. Really, if exact settings were provided so that all you had to do was dial in the numbers, you would be losing a lot of valuable experience. Give it a try. Enjoy the process as well as the results.....

nick fan
12-17-2009, 09:33 PM
As your end-user, I would have expected you had tested your settings thoroughly with experts. (for that matter, I ain't the expert :001_smile:)

no, we never advertise this. As far as I know, there is only one company which claims its pano head is ready for use out of box. But even with this kind of "calibrated" pano head, many a user are still having stitching problem in some panos.


Nick

nick fan
12-17-2009, 09:42 PM
If Bill and Nick had to spend the time to calibrate and catalog each of their pano heads for every camera/lens combination, there would have to be a price increase.

It's not hard to find the right settings and orientation for your own particular set up yourself. And it is a useful study to learn the technique. It takes less time than the time you've spent writing back to this list for help. And it's even fun and gives you a sense that you are in control of your own domain. And if you make a mistake, you'll learn from it.

And you will also learn the limits of you stitcher software. Really, if exact settings were provided so that all you had to do was dial in the numbers, you would be losing a lot of valuable experience. Give it a try. Enjoy the process as well as the results.....




Exactly. if one is making pano for fun, then learning is a very important and rewarding process. I can't remember how many tutorials I read before I attempted to develop NN. If one is making pano for money, then you can have a sense of security (after mastering it) because it is not as simple as buying equipment and software, setting up a few knobs, dials and pushing a few buttons. You don't want to invest your time and effort in something with no entry barrier.

That said, we may spend time to calibrate the panos head and provide more accurate data for popular camera and lens in the future. But we have much more important thing to do at the moment---product development. :wink:


Nick

John Houghton
12-17-2009, 11:24 PM
Would you buy a device (printer, dvd-writer, etc) knowing that the manufacturer did not supply users with a proper driver?
When you buy a photo printer from the likes of Epson, Canon etc, you are supplied with a colour profile that can be used with colour managed applications such as Photoshop to give reasonable colour rendition in prints. For optimum results, you need to calibrate the printer and make a custom profile, which is what I do for my Epson 2100. It is the user's responsibility to do this. The printer manufacturer does not provide individual calibration data for each printer.


Whether you encouraged PTgui optimizer in a certain setting is one thing, from a technical discussion, each R1 setting will either be biased towards the horizontal view -OR- nadir/zenith view?? Isn't that what Michel Thoby is trying to convey?
My tutorials clearly state that the no-parallax point setting is necessarily a compromise in the case of fisheye lenses. Individual users will have their own preferences as to how the head is set up, depending on the type of scenes being photographed and the particualr workflow. For example, small stitching errors at the nadir will be of little consequence if that area is to be covered by a separately taken nadir shot or a logo disc. The settings suggested by Nick should give good results but should be regarded as a starting point. It takes only 10 minutes to set up the head how you want it. Stitchers and blenders are capable of coping with minor inaccuracies in the head setup, particularly in the hands of an experienced user.

John

clavi
12-18-2009, 05:35 AM
If Bill and Nick had to spend the time to calibrate and catalog each of their pano heads for every camera/lens combination, there would have to be a price increase.

It's not hard to find the right settings and orientation for your own particular set up yourself. And it is a useful study to learn the technique. It takes less time than the time you've spent writing back to this list for help. And it's even fun and gives you a sense that you are in control of your own domain. And if you make a mistake, you'll learn from it.

And you will also learn the limits of you stitcher software. Really, if exact settings were provided so that all you had to do was dial in the numbers, you would be losing a lot of valuable experience. Give it a try. Enjoy the process as well as the results.....


Cataloging is part of product documentation. The camera settings page on this website has been left blank for unknown reasons. A prospective end-user would have asked "what happened to the R&D of this company?".

Make no mistake, I have tried so many settings over the past month, but I find it unfruitful as I seem to be trying my luck on each settings. With limited documentation on this product, I felt helpless. I personally hate products that had limited documentations. When I bought my Nikon camera, I was very pleased that it comes with a "thick" manual. It was awesome!

But the R1? It only came with a miserable piece of paper which I could have printed out of my computer. Could I say anymore?

clavi
12-18-2009, 05:47 AM
Exactly. if one is making pano for fun, then learning is a very important and rewarding process. I can't remember how many tutorials I read before I attempted to develop NN. If one is making pano for money, then you can have a sense of security (after mastering it) because it is not as simple as buying equipment and software, setting up a few knobs, dials and pushing a few buttons. You don't want to invest your time and effort in something with no entry barrier.

That said, we may spend time to calibrate the panos head and provide more accurate data for popular camera and lens in the future. But we have much more important thing to do at the moment---product development. :wink:


Nick


Yes, if it is for fun, then it does not matter. But this is a world where time matters. For basic settings, it has to be provided with absolute confidence. For advanced settings, you can certainly leave it to gurus to play around with the settings. That is where photography creativity comes in. But if basic work cannot be fulfiled, then that's bad.

I think the most important thing now may not be product development, but rather product documentation. U can have the best product packed with best features, but if documentation lacks, then it will only be a white elephant.

I meant well even if I sounded critical, so that you can improve your product holistically (not just feature-wise). A bad customer is one that keeps quiet about the flaws and dumps your product secretly. Hence you don't even know what/where/how to improve. I hope you understand where I am coming from.

nick fan
12-18-2009, 08:36 AM
Yes, if it is for fun, then it does not matter. But this is a world where time matters. For basic settings, it has to be provided with absolute confidence. For advanced settings, you can certainly leave it to gurus to play around with the settings. That is where photography creativity comes in. But if basic work cannot be fulfiled, then that's bad.

I think the most important thing now may not be product development, but rather product documentation. U can have the best product packed with best features, but if documentation lacks, then it will only be a white elephant.

I meant well even if I sounded critical, so that you can improve your product holistically (not just feature-wise). A bad customer is one that keeps quiet about the flaws and dumps your product secretly. Hence you don't even know what/where/how to improve. I hope you understand where I am coming from.


Hi Clavi,

There is no denying that we need to complete the documentations and videos etc ASAP. Luckily, panoramic hardware is not moving as quickly as electronics and softwares. We still have time to catch up. :biggrin:

BTW, do you have the rotator handle?
http://store.nodalninja.com/product_p/r-h.htm

it is needed for best performance.

Nick

John Houghton
12-18-2009, 09:20 AM
Make no mistake, I have tried so many settings over the past month, but I find it unfruitful as I seem to be trying my luck on each settings.

The purpose of the setup of the panohead is to eliminate parallax. You say you have tried many settings, apparently with no success. So how are you validating the setup - i.e. how are you checking that parallax has or has not been corrected adequately?

John

Terrywoodenpic
12-19-2009, 02:51 AM
People make a huge number of assumptions when making panoramas, using brackets and any of the available software.

Probably the major source of errors comes in picking and placing control points. This is even more true when using fisheye lenses which do not have a single position for the Entry pupil.

The better you get at making these choices, the better your pans turn out. Even then one some times need to resort to photoshop to make minor corrections.

The NN range of Heads are excellent but like all others can only help so much, they are only a starting point.

It is only As photographers progress along their own particular learning curve, that they understand that there never was a magic bullet that gives answers to every question.

The nearest one can hope for, is to follow in detail John Houghton's advice and tutorials. There is no short cut for experience.

clavi
12-19-2009, 04:32 AM
The purpose of the setup of the panohead is to eliminate parallax. You say you have tried many settings, apparently with no success. So how are you validating the setup - i.e. how are you checking that parallax has or has not been corrected adequately?

John


Previously did not check for parallax. Now yes, following what you advised in your tutorial. Thanks.

clavi
12-21-2009, 03:10 AM
Nick, I am abit lost to why there are so many different settings for 2-shots, 3-shots, 4-shots at 5 degrees tilt (or any particular tilt angle)? Shouldn't it be one setting per tilt per lens?

I am using John's rifle tip method to eliminate the movement of the entrance pupil. Thanks.

nick fan
12-21-2009, 04:46 AM
Nick, I am abit lost to why there are so many different settings for 2-shots, 3-shots, 4-shots at 5 degrees tilt (or any particular tilt angle)? Shouldn't it be one setting per tilt per lens?

I am using John's rifle tip method to eliminate the movement of the entrance pupil. Thanks.


Because entrance pupil moves with viewing angle. That is what Michel what to say
http://michel.thoby.free.fr/Fisheye_history_short/Beyond-the-pupil.html


Nick

TheOldFart
01-18-2010, 12:43 AM
Thanks for all of the valuable info. A wealth of information and it makes me realize once again that you can't simply pull a product out of the box :'( and expect it to work perfectly without some fine tuning. Trail and error & experience learnt from it is the answer! :biggrin:

Cheers
Tony