alignment problem with HDR (problema di allineamento con HDR)
31 views
Skip to first unread message
massimo giannelli
Sep 1, 2018, 4:55:46 AM9/1/18
to PTGui Support
hi, I have a
problem, I installed the new software 11 and the spherical images in
some places are not well aligned, even using the template that I have
always used. I tried to align even old images that had come well but the problem remains. I tried to uninstall the update and reinstall the old software but the problem remains. If I do not use the HDR system, the problem is not there. I request your assistance. I attach photos with highlighted problems
I have always worked in mode A (aperture priority) to f.8 as indicated by Google Street View and I never encountered the problem.
John Houghton
Sep 1, 2018, 6:36:03 AM9/1/18
to PTGui Support
On Saturday, September 1, 2018 at 9:55:46 AM UTC+1, massimo giannelli wrote:
hi, I have a problem, I installed the new software 11 and the spherical images in some places are not well aligned, even using the template that I have always used. I tried to align even old images that had come well but the problem remains. I tried to uninstall the update and reinstall the old software but the problem remains. If I do not use the HDR system, the problem is not there. I request your assistance. I attach photos with highlighted problems
I have always worked in mode A (aperture priority) to f.8 as indicated by Google Street View and I never encountered the problem.
Massimo, You don't have to remove V11 in order to run V10, they can coexist in you install them in different folders. I have run a test with some of my old Sigma 8mm images and did not have any alignment problems like yours. Although probably not a factor here, have you seen the thread at https://groups.google.com/forum/#!topic/ptgui/z3WtJoOXzkg concerning distortions? It's definitely worth following the advice there.
The only thing I can think might be responsible is linking. When performing the HDR stitch, do you take the option to link the images or not? Have you tried unlinking them so that the images in each bracketed set are free to align with each other?
If that doesn't improve matters, please make available for download the set of images (jpg) and V11 project file so the fault can be easily reproduced.
John
Roy Hughes
Sep 2, 2018, 1:18:20 PM9/2/18
to PTGui Support
Hello
I have had the same issue both on fusion and HDR (although I do not use HDR on Apperture A priority often). I have raised on this forum. It seems to be a known problem.
I have been told by the guys here that ver 10 was never supposed to work on A with HDR or fusion with f8 (the Google setting on the 8mm Sigma) as you mention. As you say it did work fine for some reason but no longer works as well on version 11 even using the same template. I have done that test too several times in fact. It sometimes works on 11.
There is no cure I have found. You have to use ver 10 for your old pictures and yes I think you need to uninstall everything and use a separate folder to get 10 working. On future pictures try the M setting and set the initial exposure as you think looks OK. Then bracket. I have been told 3 images are enough at each 90 degree position. I would argue with that but you can get decent results with 3 exposures if the conditions are with you.
There are a lot of good things about ver 11 but it has lost some compatibility with the Google work flow pattern that 10 did well. And as John says look at unlinking the images. It can help.
Google no longer worry about the details they once did in terms of camera settings.
Roy
On Saturday, September 1, 2018 at 9:55:46 AM UTC+1, massimo giannelli wrote:
Erik Krause
Sep 2, 2018, 4:08:04 PM9/2/18
to pt...@googlegroups.com
Am 02.09.2018 um 19:18 schrieb Roy Hughes:
> I have been told by the guys here that ver 10 was*never supposed to work*
It also sometimes works on v10, and that is the case if A mode leads to
a constant base exposure by chance.
> There is no cure I have found.
There are at least two possibilities.
- You can fake the exposure by entering regular Exposure data on Image
Parameters tab, then link HDR. I won't recommend this, since it might
lead to unexpected effects in the overlap.
- You can merge the bracketed sets in an external program. I'd recommend
enfuse with EnfuseGUI for fusion since it's free and gives nice results.
Even better results gives SNS-HDR where you could use the light version
if you don't want to spend additional money. There is a small GUI for
it, otherwise it's command line.
> You have to use ver 10 for your old pictures
> and yes I think you need to uninstall everything and use a separate folder
> to get 10 working.*
No need to uninstall anything. Simply install the other version in a
different folder.
> On future pictures try the M setting* and set the
to fix ISO, too.
> There are a lot of good things about ver 11 but it has lost some
> compatibility with the Google work flow pattern that 10 did well. And as
> John says look at unlinking the images. It can help.
The google workflow pattern didn't work in v10 any better than in v11.
> Google no longer worry about the details they once did in terms of camera
> settings.
They even didn't worry when they gave wrong advice initially.
--
Erik Krause
http://www.erik-krause.de
> I have been told by the guys here that ver 10 was*never supposed to work*
> on A with HDR or fusion with f8 (the Google setting on the 8mm Sigma) as
> you mention. As you say it did work fine for some reason but no longer
> works as well on version 11 even using the same template. I have done that
> test too several times in fact. It*sometimes* works on 11.
> you mention. As you say it did work fine for some reason but no longer
> works as well on version 11 even using the same template. I have done that
It also sometimes works on v10, and that is the case if A mode leads to
a constant base exposure by chance.
> There is no cure I have found.
- You can fake the exposure by entering regular Exposure data on Image
Parameters tab, then link HDR. I won't recommend this, since it might
lead to unexpected effects in the overlap.
- You can merge the bracketed sets in an external program. I'd recommend
enfuse with EnfuseGUI for fusion since it's free and gives nice results.
Even better results gives SNS-HDR where you could use the light version
if you don't want to spend additional money. There is a small GUI for
it, otherwise it's command line.
> You have to use ver 10 for your old pictures
> and yes I think you need to uninstall everything and use a separate folder
No need to uninstall anything. Simply install the other version in a
different folder.
> On future pictures try the M setting* and set the
> initial exposure as you think looks OK. Then bracket. I have been told 3
> images are enough at each 90 degree position. I would argue with that but
> you can get decent results with 3 exposures if the conditions are with you.
Good advice. 3 brackets 2EV apart covers most situations. And remember
> images are enough at each 90 degree position. I would argue with that but
> you can get decent results with 3 exposures if the conditions are with you.
to fix ISO, too.
> There are a lot of good things about ver 11 but it has lost some
> compatibility with the Google work flow pattern that 10 did well. And as
> John says look at unlinking the images. It can help.
> Google no longer worry about the details they once did in terms of camera
> settings.
--
Erik Krause
http://www.erik-krause.de
Roy Hughes
Sep 2, 2018, 5:57:06 PM9/2/18
to PTGui Support
Hi
I am not the sort of person who argues for the sake of it. But some of the reply is not strictly accurate Erik.
1. I think most people involved in HDR would argue you need more than three exposures on some occasions depending on how extreme the conditions.
2. At one time during the development of the desktop editor Google did reject images not taken with their standard conditions. If after a shoot their editor failed to stitch you could take it to an outside stitching tool, say PTGUI. Having taken the shot with Google's conditions you were kinda stuck with them unless you adjust with software as you suggest or go back and reshoot. Not ideal commercial workflow in either case. This was how the program worked for a time for some of us in the uk. So giving good stitching with the A setting with fusion had some advantages with the workflow imposed upon us for a time at least. In fact when I trialed the software on 50 or so stitches PTGUI was the most reliable and simplest on the A setting.
3. With 10 fusion worked consistently with auto exposure (I am not claiming 100% of the time) but most of the time. In 11 when I revisit the same files stitched with 10 it is far less reliable on the A setting on 11. I have sent links to Dropbox with example stitches and pts files. The pts files when transferred from 10 to 11 lead to correct stitching whereas the pts file from 11 for the same group of images do not. I think that is still available on Dropbox. These can be checked and if I am incorrect and something is different which will allow the same behaviour on both platforms then please write to me and let me know. No solution was suggested to this other than use M and truly spaced exposure. I don't have a problem with this in most circumstances.
Again if this is incorrect and the pts files generated on 11 can match those on 10 I would be very happy to be wrong. But that does not currently seem to be the case. I have not used it enough on M to confirm a head to head comparison between 10 and 11. I have no intention of being awkward these are just my observations.
Erik Krause
Sep 2, 2018, 6:45:46 PM9/2/18
to pt...@googlegroups.com
Am 02.09.2018 um 23:57 schrieb Roy Hughes:
> I am not the sort of person who argues for the sake of it. But some
> of the reply is not strictly accurate Erik.
> 1. I think most people involved in HDR would argue you need more than
> three exposures on some occasions depending on how extreme the
> conditions.
I don't deny that. I wrote "most situations", which implies that there
> I am not the sort of person who argues for the sake of it. But some
> of the reply is not strictly accurate Erik.
> 1. I think most people involved in HDR would argue you need more than
> three exposures on some occasions depending on how extreme the
> conditions.
are some where it won't suffice. Of course it depends on the subject.
But keep in mind that extreme DR looks unnatural, even if tonemapped
very well. Shooting for CGI is a different case of course.
> 2. At one time during the development of the desktop editor Google
> did reject images not taken with their standard conditions. If after
> a shoot their editor failed to stitch you could take it to an outside
> stitching tool, say PTGUI. Having taken the shot with Google's
> conditions you were kinda stuck with them unless you adjust with
> software as you suggest or go back and reshoot. Not ideal commercial
> workflow in either case. This was how the program worked for a time
> for some of us in the uk. So giving good stitching with the A setting
> with fusion had some advantages with the workflow imposed upon us for
> a time at least. In fact when I trialed the software on 50 or so
> stitches PTGUI was the most reliable and simplest on the A setting.
suppose it was meant as a foolproof way for non-photographers. However,
even A setting could result in the same exposure pattern for all sets
(same as if shot in M mode) if the base exposure for any direction was
the same. So google should not have rejected those.
> 3. With 10 fusion worked consistently with auto exposure (I am not
> claiming 100% of the time) but most of the time. In 11 when I revisit
> the same files stitched with 10 it is far less reliable on the A
> setting on 11.
apparently changed the control point generator. The precondition of
having all images shot with the same exposure pattern in order to be
recognized as bracketed sets has not changed.
Roy Hughes
Sep 3, 2018, 2:35:56 AM9/3/18
to PTGui Support
Hi Erik
1. Agreed. I am being pedantic, HDR people tell me 3 is nowhere near enough, but it's not a debate I am interested in. Unless it is for as you say CGI or art or marketing want it for some crazy overlay they're building then I too avoid the extremes of HDR.
2. Agreed M is better than A but it doesn't help Massimo's problem. It is a hang up from the old Google days, but whether Google were "right or wrong", and I can't say because I don't know how they engineered their software, in a sense doesn't matter. You had to play by their rules. And this was the route some of us were forced to tread some of the time. So if I wanted to go back and restitch something shot in that period I would be using content generated using their specified conditions. So if 10 and 11 handled the data in the same way that might be handy for me and other Google content generators from way back. But no one is under any obligation to make this the case. The pts files do interchange from 10 into 11. Then it will create the original stitch and a good similarity to the fusion settings that worked in 10. If I forgot to save the pts then I have to go back to 10.
3. This debate appeared in another thread. You may know far better than I the cause. This IS a puzzle. I have been told this before, and subsequently tested this myself. If you grossly reduce the number of points in an A generated file using fusion everything seems to get better, both with the stitch and the handling of "exposure" in the output image file. And using the pts file generated in 10 and using that in 11 correctly creates the stitch and fusion output appears as it did in 10 using 11. However this is a perceived quality rather than enumerated. Consequently I have not analysed exactly why it performs less well in a fully objective manner. The practicality is that it is 'apparently' inferior. I use the word apparently advisedly. There are so many variables here it's difficult to test but also impractical for one photographer to do so. For example I have always dedicated the same bodies to the shoots, are their characteristics such when pushed to extremes that they produce significantly different outputs in PTGUI to other models? So what I see as typical would in fact be atypical for most users? I can't answer that. But since the software is NOT designed to handle Google's old A setting it's probably not worth pursuing.
1. Agreed. I am being pedantic, HDR people tell me 3 is nowhere near enough, but it's not a debate I am interested in. Unless it is for as you say CGI or art or marketing want it for some crazy overlay they're building then I too avoid the extremes of HDR.
2. Agreed M is better than A but it doesn't help Massimo's problem. It is a hang up from the old Google days, but whether Google were "right or wrong", and I can't say because I don't know how they engineered their software, in a sense doesn't matter. You had to play by their rules. And this was the route some of us were forced to tread some of the time. So if I wanted to go back and restitch something shot in that period I would be using content generated using their specified conditions. So if 10 and 11 handled the data in the same way that might be handy for me and other Google content generators from way back. But no one is under any obligation to make this the case. The pts files do interchange from 10 into 11. Then it will create the original stitch and a good similarity to the fusion settings that worked in 10. If I forgot to save the pts then I have to go back to 10.
3. This debate appeared in another thread. You may know far better than I the cause. This IS a puzzle. I have been told this before, and subsequently tested this myself. If you grossly reduce the number of points in an A generated file using fusion everything seems to get better, both with the stitch and the handling of "exposure" in the output image file. And using the pts file generated in 10 and using that in 11 correctly creates the stitch and fusion output appears as it did in 10 using 11. However this is a perceived quality rather than enumerated. Consequently I have not analysed exactly why it performs less well in a fully objective manner. The practicality is that it is 'apparently' inferior. I use the word apparently advisedly. There are so many variables here it's difficult to test but also impractical for one photographer to do so. For example I have always dedicated the same bodies to the shoots, are their characteristics such when pushed to extremes that they produce significantly different outputs in PTGUI to other models? So what I see as typical would in fact be atypical for most users? I can't answer that. But since the software is NOT designed to handle Google's old A setting it's probably not worth pursuing.
PTGui Support
Sep 3, 2018, 5:28:22 AM9/3/18
to pt...@googlegroups.com
The thing is, PTGui was not designed for stitching HDR panoramas taken
in Av mode. It's not supposed to work. And it's not only v11: similar
problems were regularly reported for v10 too, all caused by shooting HDR
in Av mode. So using PTGui 10 might be the solution for you but it
certainly won't be the magic solution for all.
If it works you've been lucky; if it doesn't work there's nothing I can
do apart from telling you to shoot in M mode.
Kind regards,
Joost Nieuwenhuijse
www.ptgui.com
in Av mode. It's not supposed to work. And it's not only v11: similar
problems were regularly reported for v10 too, all caused by shooting HDR
in Av mode. So using PTGui 10 might be the solution for you but it
certainly won't be the magic solution for all.
If it works you've been lucky; if it doesn't work there's nothing I can
do apart from telling you to shoot in M mode.
Kind regards,
Joost Nieuwenhuijse
www.ptgui.com
Reply all
Reply to author
Forward
0 new messages