Printable View
If I understand correctly, you take two images using the foreground in focus and the background in another then blend them together? If this is correct, what would be the depth of field of the final composite? And is this simple to do in say Photoshop CS2/CS3?
Actually, the whole concept of depth of field falls apart when describing the DoF of the composite. There are instead two (or more) focal planes-- each of which is critically sharp at that distance and becomes progressively less sharp further in front of or behind that distance. This is ok if your photo has distinct/layered subject planes, however images with a continuous gradation from near to far would show odd sharpness patterns when combining just two images. Several images would create a wavy focus look, unless these focal planes are so closely spaced that they are within their adjacent image's DoF.
This is pretty easy to do in Photoshop, except for the fact that the field of view changes depending on the distance of focus (this is a subtle effect, but enough to create misalignment). Something not many are aware of actually. Just use layers and only pass through the sharpest parts of each layer (using masks).
How does one actually do this in practice? I am struggling with this because, as you say, I find that the field of view changes when I re-focus.
The main complication with this technique, as you mention, is that the field of view changes depending on focus distance. Therefore you have a few options:
First, use custom software which takes this into account and combines the images automatically (such as Helicon Focus or other software used in macro/microscopic shots).
Second, you could only use this technique on scenes where there are distinct focal planes in non-intertwined layers (ie, there is no layer overlap). This works well if there is a smooth and near texture-less region in between the layers--which is largely unaffected if there was slight misalignment due to changing fov. Also note that the fov does not change too much if the focusing distances are further away (and/or not as far apart).
Finally, you could combine these manually (very time consuming), by enlarging one image say 200% and the other (narrower fov one) say 205%. You would have to adjust the percentage difference until these are just right by stacking them in layers in Photoshop and using the difference combine (this quickly shows where they are out of alignment and by how much). The reason I say to enlarge both photos instead of just enlarging the smaller fov image by 105%, is because it allows you to achieve sub-pixel alignment. Once done you can downsize and then treat these as normal.
There's no perfect solution really...
I'm actually experimenting how to combine multiple images to increase depth of field.
One option that seems to work fine is to first load the images onto a stack in phostoshop setting the flag "align images". Then export the aligned images and combine them with tufuse. Tufuse is a new software that allows to combine multiple images to extend dynamic range and/or depth of field.
Another option that I'm investigating is to use the free software enfuse in combination with the program align_image_stack. This should allow to automate the whole process of alignment and combination of multiple shots. Enfuse and align_image_stack are both included in the latest releases of hugin.
I also tried the free program CombineZm, but I didn't like the results very much.
Ciao,
Fabrizio
Photoshop CS4 will apparently have the ability to build focus stacks built in, plus the ability to stitch & edit 360 panoramas, & more - check out this video starting at about 20 min
https://admin.connectpro.acrobat.com...308/p40652716/
sam
- - - -
Sam Rohn :: Location Scout :: 360 Degree Panoramas :: New York City ::