 |
|
|
|
|
|
 |
| ||||||||
 |
 
|
|
|
|
|
|
|
|
|
| ||||||||
|
|
| 
| User list | You are not logged in.
Pages: 1
Hi, I am trying to create a pano with 2 bracketed sets of images. The brackets are actually from the same capture - but the exposure has been modified and then exported from Aperture as two different versions to be combined in Autopano Giga.
I have followed the instructions carefully on how to do this but the trouble is that in the resulting psd file, the light version of the stitched pano is positioned and stitched slightly differently from the dark version in some places. In the screen shot of the render I have attached here, you can see the two layers laid over each other and how they do not line up.
I have tried altering the control point detection settings from "detect links in : one stack level" to "all all stack levels" as well as experimenting with many other settings but it makes no difference.
In the screenshot of the "layers" attached, you can see how the Yaw, Pitch, Roll, Fov and Computed Focal length changes slightly between corresponding shots in the two different stacks. Presumably these should be the same if both stacks are being computed identically. It seems like both stacks are being computed separately but I can't see how to change this.
I have tried to show screenshots of all my current settings.
I have tried this in the new 2.6.0 beta 4 version of Autopano Pro Giga and also 2.5.2 (it makes no difference).
Many thanks for your help, Michael.
Last edited by tigerchick (2011-10-31 15:39:02)
Offline
On detection panel > Links : Try "For a stack" > "Use hard links" instead of detect control points
Offline
Hi there, thanks for your replies. I have tried using "hard links" instead and sadly there is no improvement in the rendered image - however, the layers show each corresponding image to have the same yaw / roll etc figures now. See images.
I have previously experimented with the "detect links in: all stack levels" option and it did not work either.
NB. The screenshot of the render shown is just a small part of the whole image to clearly illustrate the misalignment.
Last edited by tigerchick (2011-10-31 16:40:06)
Offline
I sometimes also have such results but it doesn't come from a misuse of the software, but from where the control points are located. Unfortunately, it happens the control points are on clouds and clouds are moving between shots ( even bracketed shots ). This can gives such issues.
Offline
It's true that some control points are in clouds but it is odd that this should matter as I was under the impression that the software should treat each stack in exactly the same way. If it was indeed treating each stack in the same way then they would both be incorrectly stitched (or correctly stitched) but either way they would align perfectly.
The image attached shows which images link to which using the "hard links" option combined with "detect links in: one stack level". You will see that all of those in the lighter bracket (the file names that do not say Version 2" do not have any links at all. Presumably this is because they are theoretically taking their control points and therefore links from the other stack.
As I said before, the 2 brackets from each image are actually taken from the same capture so clouds will not be moving between the two bracket versions - they are identical except for post-process software exposure.
Offline
Any members of the development team like to comment on this?
Offline
I'm from the development team ... in fact 
( Isn't my logo big enough ? )
So, same image, but under developped, over developped or something similar, right ?
In this case, hard link or not won't change anything. We have to look for the answer somewhere else. Could you provide us the 2 images or more image where this zone is located ?
Did you create stacks in the group of course ? ( http://www.autopano.net/wiki-en/action/ … ing_Stacks )
Offline
Hi Alexandre - sorry, I didn't notice your logo - just that it says you are an "absolute beginner"!! 
"So, same image, but under developped, over developped or something similar, right ? "
Yes that's correct.
"Could you provide us the 2 images or more image where this zone is located ?"
Sure - it happens in several places on the image so maybe best if I send you the whole thing. Can I pm you a link to download it?
"Did you create stacks in the group of course ?"
Yes.
Many thanks, Michael
Offline
As an alternative approach you might want to fuse the images outside autopano, with for example enfuseGUI. Should solve the missallignment.
Offline
Hi Hans,
it's a good idea and one I may have to revert to in the end - but I really wanted to get this working in Autopano first. Processing each set of bracketed images in external software presents its own problems as it can be difficult to get precisely the same treatment on each set.
Michael
Offline
tigerchick wrote:
Hi Hans,
it's a good idea and one I may have to revert to in the end - but I really wanted to get this working in Autopano first. Processing each set of bracketed images in external software presents its own problems as it can be difficult to get precisely the same treatment on each set.
Michael
Well if it does not work like it should, it should work like it does.....a saying over here in dutch, explaining practical thinking and often followed by `....said the farmer and he placed his wife in front of the plough...`
enfuseGUI took me minutes to get it working and haven´t fused in autopano ever since.....
Offline
We have a ftp dedicated for such transfer. See the topic here : http://www.kolor.com/forum/t766-ftp-server
Offline
Hi Alexandre, I've uploaded the files to the ftp box.
Michael
Offline
HansKeesom wrote:
enfuseGUI took me minutes to get it working and haven´t fused in autopano ever since.....
The problem is it won't do alignment of images. So with any handheld / no-panohead shots it does not work. In a perfect world I'd take 10 images freeform/handheld, then another 10 with different exposure and AUTOpano will AUTOmatically do the stitching, blending and fusing. Alas, we don't live in a perfect world - that's why I STILL haven't found a way to properly stitch handheld panoramas with multiple exposures or great variance of exposure.
Offline
Bo,
Indeed why do things the easy way with a tripod ;-)
Offline
Pages: 1