 |
|
|
|
|
|
 |
| ||||||||
 |
 
|
|
|
|
|
|
|
|
|
| ||||||||
|
|
| 
| User list | Rules | You are not logged in.
tried to do a fullspherical pano, same sources as always
but the detection fails..
- results are unusable
tried on APP 2.0.9:
- results are unusable
tried APP 2.0.6:
- good results without problems
there must be an issue in that part...
Offline
Yup, I too think FE stitching is broken in 2.5.0 A1 as reported here:
http://www.autopano.net/forum/t9223-apg … -image-set
And here's another example.
Using this FE image set - low res. shot with D40 and Nikkor 10.5mmFE:
http://www.three60views.org.uk/image_se … 105-01.zip
First screenshot APG 2.5.0 using Ransac model - Similarity.
Second screenshot APG 2.5.0 using Ransac model - 3D full. Two of the images are omitted altogther.
Thirsd screenshot APG 2.0.9.
OS is Windows XP Pro with SP3, and using Nvidia GPU.
Last edited by mediavets (2010-06-25 19:18:24)
Offline
Unfortunately you must count me in too.
Tried to stich big pano, bracketed with 60 shots. I had to abort the program as in the middle of the recognizing it stopped working.
Afterwards I trierd to stitch another one with only 23 pictures. I had different results. The fist was squirked all over. Once more new, I had a strange 3 row effect (top middle nad bottom, each separated) which I could not reproduce any more. Next try I had a rather good, but instead of rms 2.44 I only had 7.6 and less links.
Manually adding links with the cp editor was a mess. Only connecting the pictures ended up with CPs out of the picture (yes rme from .27 to .68), I had to manually CP fields (old) and even manually adding CP wporked fine´. But when I started a new optimization the picture was even more distorted.
Something is wrong, but I need a further investigation.
Best regards
Georg
Offline
I agree : that's issue 125
Offline
