Ad
Ad
Ad
Pages: [1]   Bottom of Page
Print
Author Topic: Expectations on focus stacking results  (Read 2634 times)
torger
Sr. Member
****
Online Online

Posts: 1477


« on: May 28, 2012, 03:09:38 AM »
ReplyReply

It seems common these days that medium format landscape photographers do focus stacking to overcome depth of field problems. With the D800 it starts to become popular with DSLRs too. I prefer solving the problem with tilts when possible (the one-shot image is more worth to me), but it would not be wrong to have stacking in the arsenal.

I've had good results with stacking in the past, but it has been on quite easy scenes that could be solved with tilt too. Now when I have a tech camera I don't need to stack in those cases. So I made this forest scene for a test, those are often difficult to solve with tilt. The attached example is a 8 image stack of 90mm f/16 on a 33 megapixel back. I was careful when doing the stack so there is some DOF overlap between all images.

As usual I want everything to be perfect (not healthy!), but I noted that there is ghosting issues when something has moved slightly and blur around foreground objects as shown in the attached crops. ZereneStacker has been used. I have a very low tolerance on composite image artifacts, so I find this result a bit disappointing.

I wonder what expectations one should have on stacking results (is it typical to have some artifacts left that can be found by expert eyes when nosing the print, or can they be made 100% artifact free?), and how much manual retouching that is typically required etc, and what the best software for landscape work is. Perhaps one must avoid situations when ghosting issues could occur (slowly moving water with stuff floating around in it, branches moving in the wind etc)? The blur around foreground subjects is not ghosting though, but seems to be that the stacker fails to compensate for the blur widening of the foreground subjects which occurs when the background is in focus.

Any feedback appreciated.
Logged
henrikfoto
Sr. Member
****
Offline Offline

Posts: 708


« Reply #1 on: May 28, 2012, 03:35:01 AM »
ReplyReply

I have used focus-stacking mostly for macro.
I like Hilicon focus best. I found that the bigger the stack is the more chances for
strange effects. Often there is only one or two frames that causes the problems.

Whan you are outside with a little movements by the wether I think there will be some
problems like you have seen no matter what you do. But the bigger the stack, the more problems.
I would rather do 5 shots in f11 than 10 shots in f5,6.

Henrik
Logged
BernardLanguillier
Sr. Member
****
Offline Offline

Posts: 7975



WWW
« Reply #2 on: May 28, 2012, 04:40:34 AM »
ReplyReply

Both Helicon and PS allow mask editing which is often time consuming are error prone.

Forest scenes with wind can be a nightmare, but I personally find that less DoF is usually better to reduce visual clutter.

Cheers,
Bernard
Logged

A few images online here!
torger
Sr. Member
****
Online Online

Posts: 1477


« Reply #3 on: May 28, 2012, 05:17:03 AM »
ReplyReply

Both Helicon and PS allow mask editing which is often time consuming are error prone.

Forest scenes with wind can be a nightmare, but I personally find that less DoF is usually better to reduce visual clutter.

Thanks for the reply.

Due to the fairly close foreground in this scene the DOF is only about 20 cm (less than one feet) at f/16 in the closest frame. I did a solution with one-shot f/22 (corresponds to f/16 on 36x24mm) with about 4 degree forward lens tilt, and reduced fall and instead tilting the sensor plane forward as much as I could without having visible keystoning on the background trees that way moving the scheimpflug intersection point as far back as I could. With that I made to me a satisfactory one-shot picture, not 100% tack sharp all over and a bit of diffraction suffering but to my taste better than the stacked image.

I tried many other one-shot solutions too, but the problem I get is with no tilt the DOF just gets too short (the background forest could be out of focus, but I want all laying tree trunks to be in focus and that was not possible), and with tilt it looks kind of strange if things are too much out of focus since the focal plane is not upright. So I find this f/22 with tilt to be the best compromise for this particular scene.

I've attached a scaled down version of this one-shot-solution, you can see some out of focus tree trunks in the top of the frame, also attached a 100% crop showing the somewhat varying sharpness. I'm quite new to these kinds of DOF challenges (never done much forest before) so it is a learning experience. Maybe I'll try to make a print of this, if I'm really lucky perhaps the small variations in sharpness will give a subtle 3D effect rather than just looking blurry. I don't know.
Logged
torger
Sr. Member
****
Online Online

Posts: 1477


« Reply #4 on: May 28, 2012, 05:19:57 AM »
ReplyReply

Oh, another thing - has anyone tried doing focus stacking with tilt? That is tilting the focal plane to different levels and stack together. My guess is that you could get away with fewer images in the stack if that is possible.
Logged
darr
Full Member
***
Offline Offline

Posts: 122



WWW
« Reply #5 on: May 28, 2012, 06:38:38 AM »
ReplyReply

Have you considered f/stop stacking?
http://www.cambridgeincolour.com/tutorials/image-stacking-fstop.htm

I have not tried focus stacking with tilt, but I would guess it introduces a new problem into the equation; lateral movement.
Logged

darlene almeda
what am i up too?
henrikfoto
Sr. Member
****
Offline Offline

Posts: 708


« Reply #6 on: May 28, 2012, 07:39:51 AM »
ReplyReply

Have you considered f/stop stacking?
http://www.cambridgeincolour.com/tutorials/image-stacking-fstop.htm

I have not tried focus stacking with tilt, but I would guess it introduces a new problem into the equation; lateral movement.

That's a smart way to do it. First time I see it. Could maybe combine theis with a few normal stacks.
Fewer movements and less chances of arteffects?
Logged
torger
Sr. Member
****
Online Online

Posts: 1477


« Reply #7 on: May 28, 2012, 08:07:02 AM »
ReplyReply

With stacking using different focus distances there's always focus-breathing, that is the FOV changes slightly. The stackers rescale the images and aligns them before stacking (some of the peak sharpness is lost due to this rescaling, so if you shoot at f/11 the stacked image may look more like f/16 sharp anyway in terms of peak sharpness). With tilt I think there's some other type of FOV changing going on which maybe the current stackers cannot adapt for. Perhaps time for an experiment.

f-stop stacking is an interesting technique, and I have done things in the past like having say f/22 close range, and f/8 at infinity to reduce the number of images in the stack and still have high sharpness in the horizon where details are usually smaller than up close, but I also used moving focus distance. When I do stacking now there's usually so large DOF problem that there is no available aperture that brings all into focus, so the f-stop stacking with fixed focus distance technique would not work.

One interesting aspect of focus stacking using a view camera with linear focus rail focusing as I do here is that DOF/focal length/focus distance is all a zero-sum game, which means that for f/16 the DOF is always ~0.62mm on the focus rail regardless of focal length or focus distance. This makes stacking very easy, just move the lens 0.5mm between each image (to get some DOF overlap too).
« Last Edit: May 28, 2012, 08:09:26 AM by torger » Logged
Graham Welland
Sr. Member
****
Offline Offline

Posts: 610


« Reply #8 on: May 30, 2012, 01:02:30 AM »
ReplyReply

I've never tried the focus stack using just f stops - could be an interesting technique although the effects of exposure time differences and lens performance wide open could be interesting to work through.

I'm an Alpa shooter so focus stack is definitely in my vocabulary for wides at least where I can't tilt. For the forest scene you show I can't believe that a tilt will ever give your the total depth of field that you are looking for. I would stack using the same f/stop such as f/8 or f/11 and then combine using helicon. I use the masking features quite a lot these days to fix up artifacts and movements. It's certainly not perfect and if the wind is b,owing you need to understand that you're never going to get everything pin sharp.

Helicon is, for me, the very best solution. When I shoot with my Alpa STC I will either shoot a minimal set of images focused on the key scene elements or alternatively a uniform set of stacked images using various focus points from foreground to infinity. Depending upon the scene one or other will work best.
Logged

Graham
torger
Sr. Member
****
Online Online

Posts: 1477


« Reply #9 on: May 31, 2012, 01:13:47 AM »
ReplyReply

Helicon is, for me, the very best solution. When I shoot with my Alpa STC I will either shoot a minimal set of images focused on the key scene elements or alternatively a uniform set of stacked images using various focus points from foreground to infinity. Depending upon the scene one or other will work best.

Interesting! When you shoot a minimal set on key scene elements do you allow that the resulting image has a non-contiguous DOF, that is that the image goes into focus, then out of focus and then into focus again further away, or is that a big no-no?
Logged
BartvanderWolf
Sr. Member
****
Online Online

Posts: 3642


« Reply #10 on: May 31, 2012, 02:08:03 AM »
ReplyReply

Interesting! When you shoot a minimal set on key scene elements do you allow that the resulting image has a non-contiguous DOF, that is that the image goes into focus, then out of focus and then into focus again further away, or is that a big no-no?

Hi,

I would try to avoid such issues by shooting DOF zones that connect. When the DOF-zones overlap, then you'll run the risk of double features on moving objects/branches, when there is a gap between DOF-zones it can look unnatural. See this earlier thread with several suggestions on how to get connecting DOF-zones.

Cheers,
Bart
Logged
torger
Sr. Member
****
Online Online

Posts: 1477


« Reply #11 on: May 31, 2012, 02:49:29 AM »
ReplyReply

I would try to avoid such issues by shooting DOF zones that connect.

I did in my experiment, getting connecting DOF zones on a bellows camera is very easy (constant relative focus rail distance for a specific f-number regardless of focal length or focus distance, at f/16 turn 0.5mm for each shot to get suitable overlap), the question was rather an artistic one - would you accept disjointed DOF zones in an image?

Getting joined DOF zones is easy but may require a huge number of shots, so the reason of using fewer shots is to speed up shooting which can be useful/necessary from various reasons.

My overall suspicion after doing some focus stacking for subjects that cannot be solved with tilt (i e typically tight forest scenes or tight indoor scenes) is that it is very hard to do it without getting some artifacts here and there and I'm trying to get a feel how those that use focus stacking more regularly what expectations one should have, and what's generally considered ok and not by the "stacking community".

I personally have difficulties accepting artifacts in composite images, when I've done stitching for example I go through the seams and adjust manually to be 100% sure that it is impossible to see any seams. If that cannot be achieved I consider the stitch failed. It seems to me that with focus stacking one cannot have the same high expectations. But I shall try helicon focus and see if it is better than zerenestacker that I have used.
« Last Edit: May 31, 2012, 02:51:06 AM by torger » Logged
Graham Welland
Sr. Member
****
Offline Offline

Posts: 610


« Reply #12 on: May 31, 2012, 03:24:51 AM »
ReplyReply

Interesting! When you shoot a minimal set on key scene elements do you allow that the resulting image has a non-contiguous DOF, that is that the image goes into focus, then out of focus and then into focus again further away, or is that a big no-no?

It depends on the scene. If there is a foreground element such as maybe trees and then a far scene such as a distant hills and nothing much of interest in between then it works. If you have detail between the foreground scene and the background then this isn't the best approach because yes you'll have alternating regions of focus and that will look horrible. It is entirely scene depend as to whether to take this approach vs a systematic set of focus steps.
Logged

Graham
Pages: [1]   Top of Page
Print
Jump to:  

Ad
Ad
Ad