Using align_image_stack for focus stacking

[dgjohnston]

Hi, this is a link to the test subdirectory I’m using to get align_image_stack to process a few files for focus stacking. Don’t mind the laundry, these were just some quick test shots I made.

https://ln5.sync.com/dl/1535e65f0/fxrj8uqv-yuw77g77-wf4gv4sy-m4sa6k47

If you look at the jpg images you’ll see that the files are well aligned from the start (I had a tripod with a remote trigger); but I’m testing the batch file I created (AlignFocusStack.command). After running this command the first tif image is not aligned to the other four images at all and you can see the ghosting in the output1.tif file that this creates.

In the DGJVine.pto file you’ll see that there are 9 control points between image 4 and image 0 but they are all in the lower left.

These are somewhat blurry images in some areas but that is the intent of focus stacking.

I did try adding the -x, -y, -z options to align_image_stack and tried setting —core down as low as 0.5 and as high as 0.95 with any success.

I was able to use the Hugin GUI to add control points between image 4 and image 0, and between image 0 and image 1 in the blurry areas.

Any suggestions on how I might improve on the ability of align_image_stack in this situation?

Don J.

[Bruno Postle]

align_image_stack works by identifying useful features (corners) in one photo, and then trying to find a similar collection of pixels in the next photo. It may be that it just isn't very good at catching focus variation. Also, and I haven't checked this, it may be matching all subsequent images with the first image - have you tried testing just pairs of consecutive images to see if you get better results?

cpfind uses a feature classifier approach, it might give get better results than align_image_stack. Though I haven't checked this either.

-- 

Bruno

[T. Modes]

bruno...@gmail.com schrieb am Samstag, 30. April 2022 um 09:36:55 UTC+2:

align_image_stack works by identifying useful features (corners) in one photo, and then trying to find a similar collection of pixels in the next photo. It may be that it just isn't very good at catching focus variation. Also, and I haven't checked this, it may be matching all subsequent images with the first image - have you tried testing just pairs of consecutive images to see if you get better results?

align_image_stack works by default only on pairs of consecutive images (this can be changed by switch --align-to-first).

For focus stacking the images should be supplied sorted by focus distance on the command line and you should add the switch --use-given-order. By default align_image_stack sorts the images by exposure values and compares so only images with least exposure difference. But for focus stacking this is not the best approach - in this case disable the sorting with --use-given-order (then align_image_stack uses the order as given on the command line). Now align_image_stack should only compare images with the small focus distance changes.

Thomas

[dgjohnston]

Thomas and Bruno.  Thanks for the feedback.

I had tried the use_given_order but without thinking about putting the images in order of front to back focus.

I’m hitting the road tomorrow so I’ll be able to try some real stacking with landscape images and I’ll try to ensure that I have a bit of overlapping in-focus sections from one image to the next

Don

--
A list of frequently asked questions is available at: http://wiki.panotools.org/Hugin_FAQ
---
You received this message because you are subscribed to the Google Groups "hugin and other free panoramic software" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hugin-ptx+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/hugin-ptx/e868047f-32e7-41f9-a397-9bdec5e7c5cfn%40googlegroups.com.