settings for photomosaics using iPhone 13 (mini)?

45 views
Skip to first unread message

Fotis K

unread,
Apr 2, 2022, 1:26:54 PM4/2/22
to hugin and other free panoramic software
I've been using hugin to successfully stitch 100mpx mosaic reconstructions of larger artworks using my mirrorless camera and a 50mm lens but I can't seem to be able to use my iphone photos for it; On loading the photos it auto-selects a 5.1mm focal length with a 5.098 multiplier - switching to rectilinear projection messes it up even more. 

Does anybody know the actual values for the lens or what might be wrong?

Screenshot 2022-04-02 at 7.39.10 PM.png


dkloi

unread,
Apr 4, 2022, 5:17:37 PM4/4/22
to hugin and other free panoramic software
Might be better to use a panohead to take a 360 cylindrical pano with lots of overlap, create many spread out control points, optimise, and then save the lens parameters. This will give an accurate horizontal field of view, (equivalent) focal length, and the distortion values.

When you say, switching to rectilinear projection, you mean the output projection, not the lens mapping (should be a rectilinear lens anyway)? Looks like it's a problem with the field of view of the output causing the rectilinear projection to stretch the periphery. Some of your images are badly places causing this, it would seem.

Fotis K

unread,
Apr 4, 2022, 5:47:23 PM4/4/22
to hugin and other free panoramic software
thanks dkloi! that's an interesting idea, thought I don't have a panohead to do the experiment!
I'd assume since iphone 13 is a quite popular smartphone there would be  already information on this, but now that you mention it, I wonder if there is some simple way for me to calculate the values for hfov/focal length/distortion once and for all! 
As for rectilinear, I indeed was referring to the output projection and trully they are being grossly misplaced! 

dkloi

unread,
Apr 13, 2022, 3:17:51 PM4/13/22
to hugin and other free panoramic software
If you can find a scene with objects far away to place control points, you could use a normal head with a planning base to take the 360 cylindrical photo, this should reduce the effect of parallel on the stitch and optimisation.

It looked as if you had a few bad control points leading to misplaced images and the distorted output. Try manually placing them instead, a few per overlapping pair, just to get a reasonable alignment at first. Refine by adding more control points.

Reply all
Reply to author
Forward
0 new messages