Re: [hugin-ptx] Digest for hugin-ptx@googlegroups.com - 1 update in 1 topic
26 views
Skip to first unread message
Kurt Hillig
Dec 10, 2016, 8:46:34 AM12/10/16
to hugi...@googlegroups.com
Although it can be challenging, the absolute best target for lens calibration is the night sky - I can guarantee that you won't have any parallax problems with the stars! As with a chessboard, the control point finder will fail miserably so you have to set them manually; but once you've got them the results will be pretty much perfect. I know, as I did this with a six-shot series with a 16mm lens and ended up with an RMS error under 1 pixel and a maximum error just over 1 pixel.
Sent from my PDP-11/93
- Printed image for stitch calibration - 1 Update
Bruno Postle <bruno...@googlemail.com>: Dec 09 10:24AM
Hugin cpfind looks for features that are unique, so a chessboard
really confuses it. Also for calibration parallax is a big problem, so
it is much better to use a scene where the features are a long
physical distance from the camera, i.e. go outside and hold the rig
high above the ground.
The fine-tune function in the Control Points tab uses different code
altogether, this is happiest with corners - when manually picking
control points you should always be looking for corners - in this case
chessboard-like features are ideal.
--
Bruno
On 8 December 2016 at 23:47, Ben Knill wrote:
You received this digest because you're subscribed to updates for this group. You can change your settings on the group membership page.
To unsubscribe from this group and stop receiving emails from it send an email to hugin-ptx+...@googlegroups.com.
Torsten Bronger
Dec 12, 2016, 2:51:00 AM12/12/16
to hugi...@googlegroups.com
Hallöchen!
Kurt Hillig writes:
> Although it can be challenging, the absolute best target for lens
> calibration is the night sky - I can guarantee that you won't have
> any parallax problems with the stars!
That's correct but I ended up with being limited by the polynomial
fit itself. Thus, I don't think you get more accuracy than with far
away structures, or with a least-parallax adapter. Moreover, you
should know the stars quite well, otherwise, it is a tedious search.
Tschö,
Torsten.
--
Torsten Bronger
Kurt Hillig writes:
> Although it can be challenging, the absolute best target for lens
> calibration is the night sky - I can guarantee that you won't have
> any parallax problems with the stars!
fit itself. Thus, I don't think you get more accuracy than with far
away structures, or with a least-parallax adapter. Moreover, you
should know the stars quite well, otherwise, it is a tedious search.
Tschö,
Torsten.
--
Torsten Bronger
Reply all
Reply to author
Forward
0 new messages