 |
|
|
|
|
|
 |
| ||||||||
 |
 
|
|
|
|
|
|
|
|
|
| ||||||||
|
|
| 
| User list | Rules | You are not logged in.
APG 2.5 RC - Windows 32
A/ After having performed the optimisation of a 24 image spherical panorama, I got a satisfactory result with an RMS error of 1.86. I saved the pano file, performed a complete rendering and was also very happy with the final result.
B/ Now, as there was a remaining ghost in the panorama, I modified the ghost related image in gimp (create alpha channel; then erase related pixel in image) and saved it as a .png file.
C/ I edited the .pano file with wordpad to replace the name of the modified imagexx.tif by imagexx.png
D/ Then I restarted APG and loaded the modified .pano file. The issue is that, at pano loading, APG optimises the panorama ! So, the good result achieved in step A/ above desappeared and I got a very bad rms, unable to recover the good score even after several trials.
E/ I tried again with imagexx.png set back to imagexx.tif (in order to come back to the originally saved .pano file); but, anyhow, APG feels it necessary to optimise the panorama; and the achieved RMS is again much worse than in A/. It is also "funny" to note that at each new loading of the .pano file a different RMS is calculated (2.52, 3.85, 4.47, 3.55, see screen capture).
>> I have not dreamed of a good RMS: the .pano file records the figure of 1.86 that I had reached just before saving !
>> May I say that I feel very angry with this behaviour ! APG definitively shall not re-optimise the panorama at loading !!! Otherwise it is useless to spend hours optimising a panorama and at the end have all this work ruined in a few seconds ! Furthermore, additional ghost removal as may be sometime necessary will not be possible anymore if optimisation is restarted at each pano loading.
Please, please, please : make a quick correction of this bug ! Otherwise I do not see how I can continue to work !!! I could live with and overcome the bugs in beta1 and beta2; but this issue is really at the heart of my process with APG and I do not see how I could overcome it....
Anyhow, happy new year to the Kolor team, very efficient and very professional, I must say.. despite my anger of the day !
Last edited by hermer-blr (2010-12-26 19:13:04)
Offline
Offline
I agree too,
Had to go back to beta 2,
RC is not usable this way.
Reinier
Offline
RMS is not a linear measurement of panorama quality. I often stitch projects with RMS of 7-9 and more and most of my handheld panos have no problems with RMS of 3-5. So I see no problem with the projects shown in the screenshot - have you rendered them?
Also, being angry during the holidays is not good for your mental health - take a deep breath, do not use beta software if it does not suit you and have some mulled wine!
Happy holidays!
Offline
Thanks for your wise recommandations, that I will obviously strictly obey, drinking some champagne wine to your health and to the health of the Kolor team (they deserve it).
I share your view that RMS is just a number and that a panorama may be good with a bad RMS. But, when a .pano is saved, it has to reopen exactly the same : otherwise, efficient ghost hunting will be impossible.
As I shoot my panoramas handheld, it takes some time to adjust the control points and to omtimize the stitching. When this is done and when I am happy with the resul, I save the .pano and start the ghost hunting (if any ghosts either to completely remove or to resurrect) process: input image modification in gimp and creation of .png files; then editing the .pano to replace the names of the modified files. At this stage, when going back to APG, I do not want any further optimization to be performed before rendering. I only want the panorama to be rendered with the modified images and the ghosts fully removed or fully resurrected.
So, chocolate and whisky are the solution until a software upgrade is performed...
Happy new year !!!
Offline
Confirmed bug of RC : issue 479 ( reopening a .pano changes RMS ).
Offline