Issue: NIKON D200 photos fail to import properly

[virtualrepublic]

Hello,

I am trial testing at the moment and did some successful tests with iPhone 6 shots.

I am importing photos. They appear in the Model window as cameras and the alignment works fine.

Now I shot with a NIKON D200 and exported the pictures with Lightroom incl. Metadata. When I import the pictures to Photoscan they don's show up as cameras in the Model window and they are not checked in Reference. I am not able to check them. The camera alignment fails badly. Just three pictures show up wrong aligned.

Any ideas what's wrong with thoses pictures compared to the iPhone ones?


Thanks in advance for your help and answers.

Michael *:-)

[Alexey Pasumansky]

Hello Michael,

Maybe you can send us the complete problematic dataset? Also for small objects you should probably use the image frame space more effectively, although providing sufficient focal depth and image sharpness is also required.

[virtualrepublic]

Hello,

thank you very much for your reply. I sent a ZIP with two images where you can check the problem discribed.

Cheers
Michael

[Alexey Pasumansky]

Hello Michael,

I don't think we'll be able to understand, why the images don't align, looking on one photo from each set

[virtualrepublic]

Hi,

I will send the set. But when you load the iPhone JPG and the NIKON JPG you will see the main problem. The NIKON one will not show up as a camera and can't be selected in Reference. This is in my opinion the reason why the alignment will anyway fail because there is some lack of information by the picture Metadata maybe?

[virtualrepublic]

Hi,

I sent a ZIP with all 48 pics by WeTransfer to support@agisoft.com.

Thanks for your help. I am curious what the reason is *:-)


Cheers
Michael

[Alexey Pasumansky]

Hello Michael,

Thanks, will check the set.

The difference that you are speaking about is related to the presence of the coordinate information in the image EXIF. uch data is loaded automatically to the Reference pane. If there's no such information it shouldn't affect the alignment results.

[Alexey Pasumansky]

Hello Michael,

Looks like you are rotating the object in front of static camera, but not using masks to cover the background. So the alignment issues are caused by high number of matching points found on the static background, that's why many cameras are "stacked" to the same position after the alignment process.
Additionally I suggest to take more images when to object is rotated by the shortest side to provide better overlap.

And the recommendations regarding the focal depth and image sharpness are still valid, since the quality of the images in 100% zoom is not so good due to noise and blur.

[virtualrepublic]

Hello,

thanks for your support.

I shot another range with a clean plate of the background for masking. Works perfect.

The only issue I had: when I exported the RAW files to TIFF they did not work. Finally an export to JPEG gives a result. I will post it here when finished.

Cheers
Michael

[Alexey Pasumansky]

Hello Michael,

Strange that TIFFs do not work. Probably EXIF data was discarded during conversion?

[virtualrepublic]

Hi,

no idea. Did not check.

Finally I could work out a result with 160 shots. I am impressed. But there are still issues to solve. The shoestrings are broken. I think it might be blurry parts in the pictures due to a still too high aperture. I will do another test with 16-22 at 4 seconds exposure.

Check this out:
https://vimeo.com/145555682

When I have finished more tests we will buy the standard version.

Cheers
Michael