Settings with 5DMK2 & Canon 8-15 [Archive] - Nodal Ninja Forum

PDA

View Full Version : Settings with 5DMK2 & Canon 8-15



ericthoreau
09-02-2012, 12:27 AM
Hello,

When assembling my pictures, I'm still having problems with lines;

http://www.360vista.fr/temp/lequai.jpg

I'm using a NN5 with a RD16. My camera is a 5DMK2 with the Canon 8-15 lens.

At 8mm, here are my setting:
A:59
B:111
I take 4 pictures at -10° plus on of the zenith.

Do you have the same settings than me?

Regards

éric

hindenhaag
09-02-2012, 12:48 AM
Hi Eric,

Some basic info for canon 8-15mm from Michel. Adjust to upper rotator center. http://michel.thoby.free.fr/Canon_8-15mm/8-15mm_review.html#NoParallaxPoint

H in wiki database is 43.5mm, add 13mm for NN5 and try 56.5mm with Smooth method. Test till there is a round circle.

http://wiki.panotools.org/Entrance_Pupil_Database

http://www.easypano.com/forum/display_topic_threads.asp?ForumID=1&TopicID=4162

http://www.johnhpanos.com/epcalib.htm

In first case stitch the shots around without Zenith shots and in case you use PTGui Pro, go down to max distance >= 2, but you can go down possibly to 1.5 etc. by "del worst Control points" and than manually correct the rest. Finished "save as Template". Use "apply template" for next project.

Adding Zenith shot you should click "viewpoint" in "advanced, optimizer" and optimize. Use "heavy and lens shift correction" for fisheye optimization.

Feel free to ask. In first case I would check your LRS and URS by the methods in the links.

In the meantime you can send the set of pics via www.ge.tt so we can check it.

Regards,
Heinz

John Houghton
09-02-2012, 05:24 AM
éric, The appearance of the panohead at the nadir in your panorama indicates gross misalignment of the head. For setting "B": I find it is around 104. The "A" setting is dependent on how the camera is mounted and I have a Manfrotto QR plate so my figure is no use to you. You have the link to my tutorial epcalib.htm, and with that you can set up your head quite nicely in ten minutes or so with out taking any photos or stitching. But why are you shooting this subject at 8mm? This wastes pixels in the black area surrounding the image circle. 12mm would still enable you to shoot 4 around, still with almost maximum vertical fov.
.
John

ericthoreau
09-02-2012, 08:00 AM
Thank you both for your replies;
I've tried to be as thorough as possible.
For the A value, I've set the camera upside down (images attached).
For the B value, I've used John's method with a piece of tape on my window.
NN5 horizontal arm was set to -10° and focal used is 8mm (easier for me).
Now, here's the part I do not understand; since I'm using my camera at 8mm, I'll be taking 4 shoots (one every 90°).
When taking my 3 shots in order to find the nodal point, my mark (the tape on my window) is centered with my camera. If I turn my camera 90°, I won't see the mark.
So, in order to have my mark still visible in the viewfinder, I turn my camera by only 60°. Obviously, my way of doing this is perfectible...
Bottom line, for the B value, I now get 106,5.

éric

PS: sorry for my english, little rusty ;)

ericthoreau
09-02-2012, 08:56 AM
new test:
left at -73°
center at 0°
right at 87°

John Houghton
09-02-2012, 10:13 AM
Now, here's the part I do not understand; since I'm using my camera at 8mm, I'll be taking 4 shoots (one every 90°).
When taking my 3 shots in order to find the nodal point, my mark (the tape on my window) is centered with my camera. If I turn my camera 90°, I won't see the mark.
When shooting panos at 4 shots around, to find the no parallax point you position the panohead midway between two 90° detents and centre the tape in the image frame. Then take two shots at -45° and +45°, i.e. rotating your camera as you would when shotting a panorama. The tape will then fall in the middle of the overlap between the two shots. If your rotator has an option for 8 shots around at 45°, then you can simply centre the tape at one detent position and then take two shots at the detents on either side.
.
John

hindenhaag
09-02-2012, 11:06 AM
Eric,
Sorry to say, but your LRS is not centered using -90° shot. We do recommend Smooth's method not for nothing, the easypano link. Using this method gives you a higher precision, because by stitching your pics around PTGui will correct misalignment of sensor in the camera and differences in lens production as well. SO to get a precise LRS = lower rails setting, taking your time once to get your settings right will prevent you spending hours to get your panos right with wrong settings.

Take a postcard or any other hard paper, cut out a circle of 10 cm with a 8mm hole in center or use a cd as john recommended and place this below your rotator directly on the tripod. Install your equipment. Then you can use your old LRS setting by shooting down as a pre set up. Set aperture to f/16mm, and pitch down the upper rail till you see a part of the border of the round disk. Focus to the corner. Now take your shots around. Import to PTGui. Align images, go to advanced and optimize with "heavy and lens shift". "Del worst control points" and try to get max dist <2. Set "panorama settings" to 360 x 180. Set output to .mov file. Create pano. Have a look at it in Quick time player. Move down to Nadir. The card should form a round circle. If you see teeth standing out, check the direction they are cutting to. They cut to the right, move to the right by 1 mm, cutting to the left move to the left by one mm. To be sure you should have a set of three: 49mm cut to the right, 50mm without a tooth, 51mm cuts to the left. Then LRS is 50mm.

Panoheads once you found your LRS and URS move around and use the horizontal rail moved up and down - pitch - by keeping the camera lens combination both in LRS and URS in NPP. So to find your URS = upper rail setting shoot at 0°. This will keep vertical lines vertical. Shoot straight forward and a second one after moving clockwise for example. In case the tape in relation to the outside reference point follows your movement, move forward by 1mm and check again. In case it moves counterclockwise, you have to move backwards. You can do this in live-view zooming in till you think reference stands still. Then shoot the two shots. Import in Photoshop and ad pics side by side. Then use "move tool" and place the second shot having been moved to the right exactly on top of the first shot. To do this use 60% opalescence for the pic you are moving to adapt the top to the background. By clicking the "eye" for the layered pic on and of, you can check if the reference point moves.

Use URS setting of John in first case to get something you can work with to check LRS and URS.

You are not the first one trying to work around recommendations to find the settings. But we give our advices based on a lot of mistakes we did on our own and we had to work around them with a lot of time on our own in the beginning. You can always follow John's advices. He is one of the most experienced users in Panos you can find on several forums. Most of us like me have learned a lot of him and his great basic knowledge and experience.

Success for your tests. Feel free to ask,
Heinz

ericthoreau
09-02-2012, 12:05 PM
thank you John for your advices.
Here's 3 shots at 45°.
Would you consider the result as good enough for 360°?

John Houghton
09-02-2012, 03:36 PM
éric, If you look at the left hand side of the tape: as you pan the camera from left to right, the tape appears to move to the right. This means that the entrance pupil of the lens is moving right to left and so must be behind the rotation axis. Hence the camera needs to be moved forwards a little. I would try moving forwards by 1mm and trying again. Note: you are only interested in the area of the image around the tape, so zoom in - we don't need to see the whole image.
.
John

ericthoreau
09-03-2012, 12:11 PM
thank you Heinz & John for your time.

I've been conducting some tests trying to follow your advices.
This afternoon, I've been sliding the vertical rail on the Lower Horizontal Rail (file A.jpg).
The results are visible on value_a.jpg file.
Not sure what to think you it.

éric

hindenhaag
09-03-2012, 10:10 PM
For me, with all settings the teeth cut to the left. This means you have to move to the left step by step. So please try 56.5mm 56mm etc. till you find the teeth cut to the right and you find the a round circle in between.

638637636

48mm cuts to the right, 49mm is ok, 50mm cuts to the left.

Heinz

ericthoreau
09-04-2012, 02:30 AM
getting also there...
this morning, tried from 55 to 56.5.
I guess the right setting must be between 55.5 & 56.
Still can't have a perfect circle :(

éric

hindenhaag
09-04-2012, 03:19 AM
At the moment I should choose for 55.5mm. Now re check your upper rail setting, get it right and re check LRS. Good luck eric, you are on the way.

Heinz.

ericthoreau
09-04-2012, 12:49 PM
At the moment I should choose for 55.5mm. Now re check your upper rail setting, get it right and re check LRS. Good luck eric, you are on the way.

Heinz.

I think I got the upper arm right!
but now, the LRS (setup at 55.5) has to be recheck... :((
Is this normal or am I doing something wrong?

ericthoreau
09-04-2012, 01:57 PM
I'm getting lost here...
6 different settings for the LRS & still not a good result.
Around 57 seems almost good but when I try +.5 or -.5, it get worst.
Fustrating.

hindenhaag
09-04-2012, 10:21 PM
Sorry Eric,

Sort of misunderstanding. Normally you do the Smooth test once till you found your LRS setting. But in the beginning when you do not know LRS and URS, you might get misalignments because both settings might be a little bit out of NPP. When I found my LRS for my camera and change the lens I try to find my URS and of course do not touch my LRS. Plus you do not have to check LRS with different pitches because you have to shoot 2 rows for example. Pitching up and down with the horizontal arm will keep the NPP because you move in NPP.

Could you please send the solo pics of your URS with http://ge.tt/ so I can check them in PS? I feel the reference is still moving a tiny little bit. You see a sort Knob in the top of the white line on the left side, which you can not see in the centered shot.

643

Heinz

John Houghton
09-05-2012, 08:22 AM
éric, 57 seems to be about right. However, can I suggest you do the following test:

With the camera angled straight down at -90 degrees, take two shots - rotating the head by 180 degrees between them.
Stitch the images in PTGui, control points only on the floor, output format: rectilinear 90x90 degrees; PSD with individual layers.
Open the output file in Photoshop and reduce the opacity of the upper layer to 50%. You should see something like:

http://www.johnhpanos.com/nad-cali.jpg

Ideally, you don't want to see any double image of the centre screw. If there are two images visible, measure the horizontal displacement with the ruler tool and convert that to mm by dragging the ruler over to the scale marked on the NN5 arm. Half that figure is the correction needed to the LRS setting to eliminate the misalignment.

The displacement in the vertical direction is due to the mounting of the camera onto the upper arm being slightly off. It can be corrected by twisting the camera a bit (if the mount permits this). Judge the amount by looking through the viewfinder to assess the position of the centre focusing spot relative to the screw head.
.
John

ericthoreau
09-05-2012, 02:02 PM
I've posted the 3 images on my website;
Here's the link:
http://360vista.fr/temp/
Thank you for your time.

eric

John Houghton
09-05-2012, 03:12 PM
Eric, That looks to be as good as you are ever likely to get it:
.

http://www.johnhpanos.com/eric.gif
.
John

hindenhaag
09-05-2012, 11:47 PM
Hi eric,

Cp-2 plates have a slight movement which is needed to get them into the rail. Once close to be fixed try to feel if there is a"movement". With CP-2 plates I always try to set my camera "pushed forward downwards". That way the inside of the baseplate will stop at the lower front and the upper back corners inside the rail and I always get the same situation. May be you can "heal" the vertical that way.

Thx for posting and enjoy your equipment.

Heinz

ericthoreau
09-06-2012, 01:07 PM
again, thank you both for your help;
I'm off for a couple of days. I'll go back to the NPP asap & will let you know.

eric

ericthoreau
09-20-2012, 12:33 AM
I'm back.
So far, A=57 & B=105,5
With these settings, I'm tried a new way to create panorama.
I've been using a monopod (Manfrotto 628B). Not too stable but it does the job.
Here's the result:
http://www.360vista.fr/drupal/ame/nyc/central01.html

One question.
If I try to optimize my settings, should I start working on A or B value?

Again thank you.

éric

hindenhaag
09-20-2012, 02:42 AM
Hi Eric,

Start with A = Lower Rails setting.

ericthoreau
04-01-2013, 07:54 AM
hello,

I'm back since i still have some broken lines when creating a pano.
Quick question: i'm still using a 8-15 Canon @ 8mm.
I take 4 shots @ -10° & one of the zenith.
Should it be better to take 4 shots @ 0°, then one of the zenith & one of the nadir?
Which method do you use to erase the tripod?
Kind regards

eric

hindenhaag
04-01-2013, 09:37 AM
Hi Eric,

When you are not shooting from a pole, I would take the shots around at 0° and take and extra Zenith and Nadir1 and Nadir2 with Nadir Adapter. Then you can mask out your tripod in PTGuiPro with Nadir1 and Nadir2.

825827829830831832

Heinz

John Houghton
04-01-2013, 10:18 AM
eric, Do you have a particular reason for shooting at 8mm? 12mm would avoid wasting so many pixels and give better quality, and you still only need 4 shots around. Looking at the images you have posted, you don't have a nadir adapter, so if the camera is tilted @ -10° you should be able to take one nadir shot to cover the tripod area plus a zenith shot to cover the hole there.
.
John

ericthoreau
04-01-2013, 11:45 AM
Hi Eric,

When you are not shooting from a pole, I would take the shots around at 0° and take and extra Zenith and Nadir1 and Nadir2 with Nadir Adapter. Then you can mask out your tripod in PTGuiPro with Nadir1 and Nadir2.

Heinz

Hello Heinz,

thank you for your quick response; I'm not using a pole but a standard Manfrotto tripod (sometimes, I'm also using a monopod).
I've recently purchased a pano software but not PTGuiPro; I'm now using Autopano Giga 3. Do you know if this software has this feature?

I've tried this technic and I'm struggling with the fisheye effect.

http://www.360vista.fr/fish.jpg

Kind regards

John Houghton
04-01-2013, 01:15 PM
Eric, you need to set up your panohead much more accurately than this when shooting in such a confined space with things so close to the camera. The appearance of those tripod legs suggest that the adjustment is a long way from optimum, resulting in rather large parallax effects.

John

hindenhaag
04-01-2013, 02:13 PM
Eric,

Yes, I have Autopano Giga but do not use it much. I am a PTGuiPro Fan. Except for Gigapixel Panos.

Yes it has masking function: http://www.autopano.net/wiki-en/action/view/Autopano_Giga_-_Panorama_Editor_-_Masking_tool.

Fisheye effect will be handled by the software.

Shoot @zoom 12mm as John advised you. This will help you not to waist pixels. He has this combination as well. I am a Nikonian.:th_wink:

Heinz

ericthoreau
04-03-2013, 04:19 PM
So, I've shot @ 12mm; 4 shots @ -10° & one of the zenith.
A: 55
B: 105,5

Here's a shot of the center of my NN5: http://ericthoreau.free.fr/temp/IMG_3735_center.jpg (the red dot is the center of the image)

After stitching with AutoPano 3, my RMS started at 4,01 & after improvement, I've ended at 2,61
http://ericthoreau.free.fr/temp/12mm-5shots-B105.5_down_web.jpg

The pano is here: http://ericthoreau.free.fr/temp/12mm-5shots-B105.5.mov (made with pano2vr)

I've blurred out some parts of my messy hotel room ;)

Do you believe my end result could be better?

éric

ericthoreau
04-03-2013, 05:41 PM
so, I've setup my head & took 3 shots @ 0° every 45°.

My original setup @ 12mm: http://ericthoreau.free.fr/temp/105,5.jpg

After working around @ 12mm: http://ericthoreau.free.fr/temp/102.jpg

The rail B: http://ericthoreau.free.fr/temp/102_railB.jpg

ericthoreau
04-03-2013, 05:44 PM
New pano with new settings
5 shots: 4 @ -10° + 1 zenith

Center: http://ericthoreau.free.fr/temp/IMG_3789_center.jpg

Nadir: http://ericthoreau.free.fr/temp/12mm-5shots-B102_down_web.jpg

The pano in mov: http://ericthoreau.free.fr/temp/12mm-5shots-B102_out.mov (note that I've cleaned my room ;)

hindenhaag
04-03-2013, 10:19 PM
Eric,

Just use smooth's method because just focussing down does not help you: http://www.easypano.com/forum/display_topic_threads.asp?ForumID=1&TopicID=4162

833834835

Heinz

ericthoreau
04-05-2013, 06:26 AM
first test for A rail with these values: 57; 58; 60 & 62

http://ericthoreau.free.fr/temp/rail_A_57-62.jpg

with this, I'm not sure where to go...

John Houghton
04-05-2013, 06:43 AM
Eric, Could you please upload jpeg versions of the camera images used for the 58mm setting, say. It's possible there may be stitching issues.
.
John

ericthoreau
04-05-2013, 07:07 AM
done
5 pictures here http://ericthoreau.free.fr/temp/

& added 55 & 65 just for the sake of it

http://ericthoreau.free.fr/temp/rail_A_55-65.jpg

John Houghton
04-05-2013, 08:16 AM
Eric, My suspicions proved to be correct. The stitch of the 58mm images is fine and indicates a satisfactory setting:

http://www.johnhpanos.com/erics-nadir.jpg

While there is nothing wring with Smooth's method, it does demand some experience with using PTGui and I would not recommend it for novices, as misleading results can be produced.
.
John

ericthoreau
04-05-2013, 09:31 AM
John & Heinz, thank you for your help;

so, to sum up:

A: 58

For rail B, should I start again or is the value founded previously (102) correct?

eric

John Houghton
04-05-2013, 09:58 AM
Eric, I suggest you check for parallax at the 102 setting. Include something close to the camera (50cm) with something a long way away in the background that has sufficient detail such that you can see whether the near object changes position from one shot to the next. Take two shots exactly as you would for an actual panorama and set the near object in the middle of the overlap of the two images. So rotate the camera from -45° to +45° The near object can be anything you like - a pole or stick or a piece of sticky tape on a window that looks out onto a distant view. The object should be in the centre of the frame when the head is at 0°. You can use live view and magnify the image to judge the parallax shift visually, or you can take the shots and compare them in Photoshop, or stitch the shots to a layered PSD file and compare layers in Photoshop by switching the top layer on and off.

John.

ericthoreau
04-06-2013, 04:03 AM
Hello John,

here are my result for B rail; i've used a tape on my window at 34cm of the lens.

103 looks pretty good.

http://ericthoreau.free.fr/temp/B=100.jpg

http://ericthoreau.free.fr/temp/B=102.jpg

http://ericthoreau.free.fr/temp/B=103.jpg

http://ericthoreau.free.fr/temp/B=104.jpg

http://ericthoreau.free.fr/temp/B=105.jpg

John Houghton
04-06-2013, 05:17 AM
Eric, I think 102mm looks better. They are easier to judge when you remove the fisheye distortion by stitching the pair of images and cropping out the area of interest as a rectilinear PSD layered file. (Put control points only on the background). Switching the upper layer on/off then reveals the parallax shift clearly. I did this as best I could using your crops:

http://www.johnhpanos.com/102parallax.gif

That looks very good to me.

John

ericthoreau
04-07-2013, 02:32 AM
thank you John;

the final sum up:
With a Canon 5Dmk2, using a Canon 8-15 @ 12mm,
A=58
B=102
4 shots at -10° + 1 zenith
or
4 shots at 0° + 1 zenith + 1 nadir (and another one with the tripod slightly on the side)

eric