Home

Please support mflenses.com if you need any graphic related work order it from us, click on above banner to order!

SearchSearch MemberlistMemberlist RegisterRegister ProfileProfile Log in to check your private messagesLog in to check your private messages Log inLog in

Stitched panorama with Som Berthiot 4.5/90
View previous topic :: View next topic  


PostPosted: Wed Jan 13, 2021 9:07 pm    Post subject: Stitched panorama with Som Berthiot 4.5/90 Reply with quote

Finally some nice sun in the sky!

A more sophisticated way to use Som Berthiot 4.5/90: a couple of stitched images. Made of respectively circa 30 and 20 shots, taken at f5.6, handheld. Once processed (exposure balance) by the stitching software, further PP tweaks were applied, than resized.

Click on the images to see them in better resolution.

#1


#2


PostPosted: Wed Jan 13, 2021 11:01 pm    Post subject: Reply with quote

Very nice, I like a good stitched pano'.

Photomerge ?


PostPosted: Wed Jan 13, 2021 11:44 pm    Post subject: Reply with quote

Thank you, Lloydy. It's Hugin Panorama Editor, open source. I am pretty impressed with its capacities in simple mode. I just drag a number of shots in the window, do align then create, and it creates a well assembled pano (in most cases) from this mess.

There are several corrective options regarding the way the pano is stitched: rectilinear, cylindric, etc. It's interesting how vertical panos are made. In case of high architectural objects the software seems to try correcting the perspective distortion which makes the things look different than with eyes or in one-shot photo. Curious that the software creates this way another point of view.

Which stitcher do you use? Are you satisfied with its capacities?


PostPosted: Thu Jan 14, 2021 12:41 am    Post subject: Reply with quote

This was done in Adobe Photoshop / Photomerge, which sounds very similar to Hugin
This was two long exposures, maybe 1.5 sec's taken with the Tokina 17 / 3.5 and my excellent Sirui lightweight tripod. This is a vertical two image stitch, with a lot of Photoshop processing to get the perspective looking right. The angle of the upright piers of the viaduct is correct, they do taper in this way. But even if the Tokina is rectilinear, this stitch was a hell of a lot of work to get right. But it's a shot that I couldn't have got with a longer lens from a more distant viewpoint.

Sony A7II
Tokina RMC 17 / 3.5



PostPosted: Thu Jan 14, 2021 1:14 am    Post subject: Reply with quote

A very good example which gives a nice sense of volume dimension. Congrats for successful adjustments!

Here are two results after processing of the same image set. The shots are taken with another lens, Minolta Rokkor-HG 2.8/35, at f4, in 6 or 7 shots. The left one is processed with "cylindric" projection option in Hugin, the right with "rectilinear" one.



Both are rather mindblowing, as the software managed to "correct" the perspective distortion in a way that makes it look taken frontally, and not from below. Both are "wrong" from the point of view from which the images were taken, as well as from the point of view of the real proportions. Meanwhile computational intelligence of the software is rather impressive.


PostPosted: Thu Jan 14, 2021 6:41 am    Post subject: Reply with quote

alex ph wrote:
A very good example which gives a nice sense of volume dimension. Congrats for successful adjustments!

Here are two results after processing of the same image set. The shots are taken with another lens, Minolta Rokkor-HG 2.8/35, at f4, in 6 or 7 shots. The left one is processed with "cylindric" projection option in Hugin, the right with "rectilinear" one.

Both are rather mindblowing, as the software managed to "correct" the perspective distortion in a way that makes it look taken frontally, and not from below. Both are "wrong" from the point of view from which the images were taken, as well as from the point of view of the real proportions. Meanwhile computational intelligence of the software is rather impressive.


Like 1 small

Have you tried any of the CA correction workflow with that software package?


PostPosted: Thu Jan 14, 2021 2:47 pm    Post subject: Reply with quote

I don't see CA correction menu in the simple mode. It may be present in expert mode, but I haven't tested it yet.


PostPosted: Thu Jan 14, 2021 3:07 pm    Post subject: Reply with quote

alex ph wrote:
I don't see CA correction menu in the simple mode. It may be present in expert mode, but I haven't tested it yet.


It appears to be a workflow with Hugin being the first part.

http://hugin.sourceforge.net/tutorials/tca/en.shtml


PostPosted: Thu Jan 14, 2021 3:09 pm    Post subject: Reply with quote

A sample image presented in the info page looks really good. Certainly worth trying. Thank you for the hint!


PostPosted: Thu Jan 14, 2021 3:23 pm    Post subject: Reply with quote

I gave another glance to the software interface. It evolved in the current version (2020) as compared to the one present in the tutorial (2009). In simple mode you have no access to CA correction via Settings menu.


PostPosted: Thu Jan 14, 2021 3:59 pm    Post subject: Reply with quote

alex ph wrote:
I gave another glance to the software interface. It evolved in the current version (2020) as compared to the one present in the tutorial (2009). In simple mode you have no access to CA correction via Settings menu.


I had not tried the approach either. Just curious.

Here is the version mentioned in the steps.

https://sourceforge.net/projects/hugin/files/hugin/hugin-0.7.0/


PostPosted: Thu Jan 14, 2021 4:24 pm    Post subject: Reply with quote

Like 1 small


PostPosted: Fri Jan 15, 2021 2:35 am    Post subject: Reply with quote

alex ph wrote:
Like 1 small


It's all command line work. There is a folder in the newest copy of this app with some scripted tools. They are called tca_correct and fulla.

I tried a couple images and didn't see much difference. They might have too much color to be a good reference. Once I produce a good reference it should work better. Looks like some trial and error though.

Run this: /Applications/Hugin/Hugin.app/Contents/MacOS/tca_correct -o abcv DSC03576.TIF

Output: -r -0.0003765:-0.0000647:0.0005017:0.9997212 -b -0.0009326:0.0021264:-0.0015615:1.0003248

Run this: /Applications/Hugin/Hugin.app/Contents/MacOS/fulla -r -0.0003765:-0.0000647:0.0005017:0.9997212 -b -0.0009326:0.0021264:-0.0015615:1.0003248 DSC03576.TIF

Output: DSC03576_Corr.TIF


PostPosted: Sat Jan 16, 2021 4:26 pm    Post subject: Reply with quote

Blazer0ne wrote:

Output: -r -0.0003765:-0.0000647:0.0005017:0.9997212 -b -0.0009326:0.0021264:-0.0015615:1.0003248

Run this: /Applications/Hugin/Hugin.app/Contents/MacOS/fulla -r -0.0003765:-0.0000647:0.0005017:0.9997212 -b -0.0009326:0.0021264:-0.0015615:1.0003248 DSC03576.TIF


That looks a bit too much techie for me. Crying or Very sad I rely on your evidence that you did not see a huge difference. But this might also be because of the software correcting CA by default?


PostPosted: Sat Jan 16, 2021 4:35 pm    Post subject: Reply with quote

Sometimes, especially with complex architecture objects, the software does not manage to match points from different shots. In those cases the resulting stitch comes pretty curious. Always with Som Berthiot 4.5/90:



PostPosted: Sat Jan 16, 2021 4:43 pm    Post subject: Reply with quote

alex ph wrote:
Blazer0ne wrote:

Output: -r -0.0003765:-0.0000647:0.0005017:0.9997212 -b -0.0009326:0.0021264:-0.0015615:1.0003248

Run this: /Applications/Hugin/Hugin.app/Contents/MacOS/fulla -r -0.0003765:-0.0000647:0.0005017:0.9997212 -b -0.0009326:0.0021264:-0.0015615:1.0003248 DSC03576.TIF


That looks a bit too much techie for me. Crying or Very sad I rely on your evidence that you did not see a huge difference. But this might also be because of the software correcting CA by default?


I think the idea is that you can use this tool to build a CA profile a for any lens at a given focal length, focus or aperture. That profile can be applied to any new photo. The quality of the profile is dependent on the source image for generating the map. Though I am sure with this tool there is more in it than what I have seen.


PostPosted: Sat Jan 16, 2021 4:47 pm    Post subject: Reply with quote

alex ph wrote:
Sometimes, especially with complex architecture objects, the software does not manage to match points from different shots. In those cases the resulting stitch comes pretty curious. Always with Som Berthiot 4.5/90:



Van Gogh's Inception


PostPosted: Sat Jan 16, 2021 4:49 pm    Post subject: Reply with quote

Thank you for the explanation!


PostPosted: Sat Jan 16, 2021 4:50 pm    Post subject: Reply with quote

Blazer0ne wrote:

Van Gogh's Inception


Laugh 1


PostPosted: Sun Jan 17, 2021 8:12 am    Post subject: Reply with quote

Blazer0ne wrote:
alex ph wrote:
Sometimes, especially with complex architecture objects, the software does not manage to match points from different shots. In those cases the resulting stitch comes pretty curious. Always with Som Berthiot 4.5/90:



Van Gogh's Inception

Laugh 1 Thank You Dog Happy Dog


PostPosted: Sat Jan 23, 2021 10:44 am    Post subject: Reply with quote

poilu wrote:

Laugh 1 Thank You Dog Happy Dog


Happy Cat


Last edited by alex ph on Sat Jan 23, 2021 11:07 am; edited 1 time in total


PostPosted: Sat Jan 23, 2021 11:06 am    Post subject: Reply with quote

I tried another free stitching software which is Image Compose Editor. It is made by Microsoft, in a very easy MS spirit. It has only simple mode, almost no correction to implement on the user side. But the job it makes is really good. And what is even more impressive, the software offers an option to fill in the gaps and irregularities left by the shot series you produced.

Here is an example, the lens is Som Berthiot again.

If you look at the pano in general you may hardly find some anomalies.

#1


If you give a closer attention to corners, you may notice a branch suspended in the air in the top left. But not much more than that. Here is the original stitch, before applying the Auto complete function.

#2


As you may see, the lacking parts of the pano are pretty extensive, and the AI work is well done. This is a bit delusive how the algorithm recreates the reality that lacks - with possible fragments - and staples it in a photo form destined to certify the reality.

In some other cases the auto completion gives more banal or pathetic results, just repeating the closest corner pattern with fill-in function. The kind you may see in the top right corner. In general it seems thus to make a poorer work with small details and arbitrary geometry (trees, irregular buildings, patched background) and works better where the geometry is more regular, made of clear lines and large forms.


PostPosted: Sun Jan 24, 2021 8:19 am    Post subject: Reply with quote

alex ph wrote:
I tried another free stitching software which is Image Compose Editor. It is made by Microsoft, in a very easy MS spirit. It has only simple mode, almost no correction to implement on the user side. But the job it makes is really good. And what is even more impressive, the software offers an option to fill in the gaps and irregularities left by the shot series you produced.

Here is an example, the lens is Som Berthiot again.

If you look at the pano in general you may hardly find some anomalies.

#1


If you give a closer attention to corners, you may notice a branch suspended in the air in the top left. But not much more than that. Here is the original stitch, before applying the Auto complete function.

#2


As you may see, the lacking parts of the pano are pretty extensive, and the AI work is well done. This is a bit delusive how the algorithm recreates the reality that lacks - with possible fragments - and staples it in a photo form destined to certify the reality.

In some other cases the auto completion gives more banal or pathetic results, just repeating the closest corner pattern with fill-in function. The kind you may see in the top right corner. In general it seems thus to make a poorer work with small details and arbitrary geometry (trees, irregular buildings, patched background) and works better where the geometry is more regular, made of clear lines and large forms.


Well, just the same than the popular content aware filling of Adobe


PostPosted: Sun Jan 24, 2021 9:49 pm    Post subject: Reply with quote

Thanks for the hint. I haven't used Adobe since a decade or so, I was unaware it was made an built-in option. Does it work in the same way in Photoshop? I mean just filling-in the lacking parts with repetitive pattern, in difficult cases.


PostPosted: Sun Jan 24, 2021 10:07 pm    Post subject: Reply with quote

alex ph wrote:
Thanks for the hint. I haven't used Adobe since a decade or so, I was unaware it was made an built-in option. Does it work in the same way in Photoshop? I mean just filling-in the lacking parts with repetitive pattern, in difficult cases.


Yes, it does. You just have to make a selection and then cancel its content: You'll be asked to choose among a color filling, or the "content aware" filling. Of course it fails often, and someone took the chance to make funny videos that you can find under "content aware fails" keywords; here some of them: https://www.youtube.com/results?search_query=content+aware+fails