[ptx] pano12.dll ver. 2.7.0.11 and hugin - strange results

[Mike Runge]

Hi experts,

there seems to be something strange with the combination pano12.dll ver.
2.7.0.11 / hugin_rc5.2.

I'm currently working with hugin_0.5rc2 and pano12.dll ver. 2.7.0.10.
I have a project with 4 cropped circular fisheye images (Pentax istD with
Sigma 8mm) that's sufficiently optimized using 29 t0 controlpointpairs
and one t1 pair.

If I open that project and just optimize "Everything" (using pano12.dll
ver. 2.7.0.10) nothing changes (like expected).

Here's a part of the projectfile:

Pano12.dll ver. 2.7.0.10
===========================================================================================================
# PTOptimizer script, written by hugin

p f2 w6000 h3000 v360  n"TIFF_m c:NONE"
m g1 i0 f0

# image lines
i w3008 h2008 f2 a-0.051089 b0.101489 c-0.127637 d15.6368 e-26.4724 g0 p0
r90.0496 t0 v177.209 y-24.2077  C129,2911,-413,2369 u10 n"n0001.tif"

i w3008 h2008 f2 a=0 b=0 c=0 d=0 e=0 g0 p0.0268594 r90.8236 t0 v=0
y65.7509  C129,2911,-413,2369 u10 n"n0002.tif"
i w3008 h2008 f2 a=0 b=0 c=0 d=0 e=0 g0 p-0.673028 r90.7771 t0 v=0
y156.114  C129,2911,-413,2369 u10 n"n0003.tif"
i w3008 h2008 f2 a=0 b=0 c=0 d=0 e=0 g0 p-0.718714 r90.0651 t0 v=0
y-113.745  C129,2911,-413,2369 u10 n"n0004.tif"

# specify variables that should be optimized
v a0 b0 c0 d0 e0 r0 v0
v p1 r1 y1
v p2 r2 y2
v p3 r3 y3
v
===========================================================================================================

The problem starts here :-(
If I load exactly this file into hugin using pano12.dll ver. 2.7.0.11, I
get different values (and horrible stitching results):

Pano12.dll ver. 2.7.0.11 - first optimize
===========================================================================================================
# PTOptimizer script, written by hugin

p f2 w6000 h3000 v360  n"TIFF_m c:NONE"
m g1 i0 f0

# image lines
i w3008 h2008 f2 a-0.174953 b0.242337 c-0.205942 d-0.480338 e-0.466305 g0
p0 r90.0537 t0 v172.921 y-24.2077  C113,2895,-387,2395 u10 n"n0001.tif"

i w3008 h2008 f2 a=0 b=0 c=0 d=0 e=0 g0 p0.0297414 r90.8171 t0 v=0
y65.7503  C113,2895,-387,2395 u10 n"n0002.tif"
i w3008 h2008 f2 a=0 b=0 c=0 d=0 e=0 g0 p-0.660744 r90.7736 t0 v=0
y156.114  C113,2895,-387,2395 u10 n"n0003.tif"
i w3008 h2008 f2 a=0 b=0 c=0 d=0 e=0 g0 p-0.708861 r90.0712 t0 v=0
y-113.745  C113,2895,-387,2395 u10 n"n0004.tif"

# specify variables that should be optimized
v a0 b0 c0 d0 e0 r0 v0
v p1 r1 y1
v p2 r2 y2
v p3 r3 y3
v
===========================================================================================================

Easy to see, that the values for v, d and e changed a lot. All other lens
parameters changed as well and the positioning parameters also changed a
bit.

If I hit optimize "Everything" again, I get again different results!

Pano12.dll ver. 2.7.0.11 - second optimize
===========================================================================================================
# PTOptimizer script, written by hugin

p f2 w6000 h3000 v360  n"TIFF_m c:NONE"
m g1 i0 f0

# image lines
i w3008 h2008 f2 a-0.174917 b0.242259 c-0.205888 d15.5197 e-26.4663 g0 p0
r90.0536 t0 v172.921 y-24.2077  C129,2911,-413,2369 u10 n"n0001.tif"

i w3008 h2008 f2 a=0 b=0 c=0 d=0 e=0 g0 p0.0297517 r90.8171 t0 v=0
y65.7503  C129,2911,-413,2369 u10 n"n0002.tif"
i w3008 h2008 f2 a=0 b=0 c=0 d=0 e=0 g0 p-0.660749 r90.7736 t0 v=0
y156.114  C129,2911,-413,2369 u10 n"n0003.tif"
i w3008 h2008 f2 a=0 b=0 c=0 d=0 e=0 g0 p-0.708877 r90.0712 t0 v=0
y-113.745  C129,2911,-413,2369 u10 n"n0004.tif"

# specify variables that should be optimized
v a0 b0 c0 d0 e0 r0 v0
v p1 r1 y1
v p2 r2 y2
v p3 r3 y3
v
===========================================================================================================
Now d and e are more similar to the results using pano12.dll ver.
2.7.0.10, but the other lens and positioning parameters are still
different and the

Stitching result is still a mess.

I verified this behavior by using another project (without any
vertical/horizontal controllpoints). Same results:
- different values, d, e tend to 0, v decreases.
- pressing optimize again will flip d and e back to similar values than
with 2.7.0.10, v stays smaller than needed.
- stitching results are reaaly bad (unsuable).

I know that it would be better to define more verticals. However, is
behavior is significantly different from pano12.dll ver. 2.7.0.10. Is it
a pano12.dll bug, a hugin bug or am I doing something stupid?

I attached the full projectfile wich works fine with pano12.dll ver.
2.7.0.10.

Thanks for your help,
mike
-------------- next part --------------
A non-text attachment was scrubbed...
Name: project_27010.pto
Type: application/octet-stream
Size: 1830 bytes
Desc: not available
Url : https://www.email-lists.org/pipermail/ptx/attachments/20051028/cce67c36/project_27010.obj