Use lens distorsion function in APP or not?  

This forum is dedicated to Autopano Pro / Giga bug reports and features discussion.
Please read the posting rules before starting a topic!
no avatar
marijonas
Member
 
Topic author
Posts: 22
Likes: 0 post
Liked in: 0 post
Joined: Wed Sep 22, 2010 2:51 pm
Info

Use lens distorsion function in APP or not?

by marijonas » Wed Sep 22, 2010 3:32 pm

Hi,

I make fullspherical panoramas with a Sigma 10-20mm/4-5.6 lens. This lens has nearly no distorsion but in the edges the distorsion is about 2-3%.

What parameters does APP take for the distorsion.

If I disable the lens distorsion I get higher RMS-numbers, but better results in the edges.

If I enable the lens distorsion I get lower RMS-numbers, but more steps in lines.

(see pictures below)

Can APP correct lens distorsion which is variing?

Thank you very much.

Richard





no avatar
mediavets
Moderator
 
Posts: 16415
Likes: 2 posts
Liked in: 130 posts
Joined: Wed Nov 14, 2007 2:12 pm
Location: Isleham, Cambridgeshire, UK.
Info

by mediavets » Wed Sep 22, 2010 4:15 pm

My guess is that the problem is either parallax, because your pano head is not set up correctly, or APP/APG placing control points on moving clouds.

Did you try editing control points?

Are you confident that your pano head is setup correctly?
Last edited by mediavets on Wed Sep 22, 2010 4:16 pm, edited 1 time in total.
Andrew Stephens
Many different Nodal Ninja and Agnos pano heads. Merlin/Panogear mount with Papywizard on Nokia Internet tablets.
Nikon D5100 and D40, Sigma 8mm f3.5 FE, Nikon 10.5mm FE, 35mm, 50mm, 18-55mm, 70-210mm. Promote control.

no avatar
mediavets
Moderator
 
Posts: 16415
Likes: 2 posts
Liked in: 130 posts
Joined: Wed Nov 14, 2007 2:12 pm
Location: Isleham, Cambridgeshire, UK.
Info

by mediavets » Wed Sep 22, 2010 4:19 pm

marijonas wrote:Hi,

I make fullspherical panoramas with a Sigma 10-20mm/4-5.6 lens. This lens has nearly no distorsion but in the edges the distorsion is about 2-3%.

Richard

Here's what the dpreview.com review of this lens has to say about distortion:
"Distortion is a little high, especially when compared to the other wideangle zooms we've tested recently. There's fairly strong (2.4%) barrel distortion at wideangle, with a rather unusual and complex pattern, localized mainly towards the corners of the frame (and therefore difficult to correct fully in software). This changes to pincushion distortion at longer focal lengths, peaking at -1.4% around 14mm."

http://www.dpreview.com/lensreviews/sigma_10-20_4-5p6_n15/page3.asp
Last edited by mediavets on Wed Sep 22, 2010 4:20 pm, edited 1 time in total.
Andrew Stephens
Many different Nodal Ninja and Agnos pano heads. Merlin/Panogear mount with Papywizard on Nokia Internet tablets.
Nikon D5100 and D40, Sigma 8mm f3.5 FE, Nikon 10.5mm FE, 35mm, 50mm, 18-55mm, 70-210mm. Promote control.

no avatar
marijonas
Member
 
Topic author
Posts: 22
Likes: 0 post
Liked in: 0 post
Joined: Wed Sep 22, 2010 2:51 pm
Info

by marijonas » Wed Sep 22, 2010 7:16 pm

mediavets wrote:My guess is that the problem is either parallax, because your pano head is not set up correctly, or APP/APG placing control points on moving clouds.

I´ve checked all contol points and deleted all points in the clouds.

Yes my panohead is not perfect adjusted because of using a monopod, but I´m in a radius of 1mm tolerance.
I know that this is much for my testing pano (distance 30cm). Stitching mistakes are usual. That is no(t my) question.

I´ve checked all contol points and deleted all points in the clouds.

Why get the picture worse, if I enable distorsion correction. This was the only change between this two pictures.

I guess, APP only can correct non changing (simple) distortions. Is this correct?
Maybe a topic for next version of APP.

no avatar
mediavets
Moderator
 
Posts: 16415
Likes: 2 posts
Liked in: 130 posts
Joined: Wed Nov 14, 2007 2:12 pm
Location: Isleham, Cambridgeshire, UK.
Info

by mediavets » Wed Sep 22, 2010 7:33 pm

marijonas wrote:I know that this is much for my testing pano (distance 30cm). Stitching mistakes are usual. That is no(t my) question.

Oh... I thought your concern was how to get a good stitch.

But it seems it's not...

I have noticed that others have had problems getting a good stitch using that Sigma 10-20mm wide angle rectilinear lens. Perhaps there are specific issue with it and APP/APG?

Have you tried correcting lens distortion with DxO before stitching?
http://www.dxo.com/uk/photo
Last edited by mediavets on Wed Sep 22, 2010 7:39 pm, edited 1 time in total.
Andrew Stephens
Many different Nodal Ninja and Agnos pano heads. Merlin/Panogear mount with Papywizard on Nokia Internet tablets.
Nikon D5100 and D40, Sigma 8mm f3.5 FE, Nikon 10.5mm FE, 35mm, 50mm, 18-55mm, 70-210mm. Promote control.

no avatar
marijonas
Member
 
Topic author
Posts: 22
Likes: 0 post
Liked in: 0 post
Joined: Wed Sep 22, 2010 2:51 pm
Info

by marijonas » Wed Sep 22, 2010 9:09 pm

Yes I´ve read about this specific issue with this Sigma lens and APP,
and I´ve seen, that other stitching software solve this difficulty a little bit better.

So my question aimes deep in technology of stitching algorithm.

My suggestion is, that APP only uses one spherical parameter to correct distorsion.
May be a topic for next release.

In the case of Sigma 10-20, which has a very good correction of distorsion in a great centered field,
it´s better to disable the distorsion correction, if the overlapp is big enought and the blender is in centered area. In case of low overlapp it might be better to enable the distorsion correction.

But I haven´t verified.

DxO lens correction I havn´t tried, because of APP tipp not to correct distorsion before.

I´ve tested distorsion correction with CaptureNX2 with my Nikkor 18-70mm/3.5-4.5 before stitching, but the results got a little bit worse.

no avatar
mediavets
Moderator
 
Posts: 16415
Likes: 2 posts
Liked in: 130 posts
Joined: Wed Nov 14, 2007 2:12 pm
Location: Isleham, Cambridgeshire, UK.
Info

by mediavets » Wed Sep 22, 2010 9:20 pm

I feel that the only valid test scenario would be a pano scene which provides adequate features to enable the stitcher to place sufficient good control points to achieve a good stitch.

It would be equally important to shoot from a stable tripod and with a properly calibrated pano head. Otherwise there will be too many confounding factors to determine/diferentiate effects associated with lens distortion.

Alternatively do trial with images corrected using DxO, which claims to be able to handle complex distortions, and see whether you get a better result than when using the same image set without DxO correction.

I can't comment on how distortion correction is designed in APP/APG only the developers can describe that.
Andrew Stephens
Many different Nodal Ninja and Agnos pano heads. Merlin/Panogear mount with Papywizard on Nokia Internet tablets.
Nikon D5100 and D40, Sigma 8mm f3.5 FE, Nikon 10.5mm FE, 35mm, 50mm, 18-55mm, 70-210mm. Promote control.

no avatar
marijonas
Member
 
Topic author
Posts: 22
Likes: 0 post
Liked in: 0 post
Joined: Wed Sep 22, 2010 2:51 pm
Info

by marijonas » Wed Sep 22, 2010 9:30 pm

Hello Andrew,

thank you very much for your helpful comments. I will test distorsion settings on other locations.

Yes the bigger issue, in my case, is my uncalibrated panohead.

Richard


Who is online

Users browsing this forum: No registered users and 4 guests