Focus Stacking — Part 3

What Works, What Doesn’t

To further understand focus stacking and to compare the good, bad and ugly of Photoshop auto-merge, Helicon Focus and potentially other focus stacking software, I set out to capture sets of images to challenge the technique.  I also spent a lot of time attempting to reduce or eliminate the haloing problem on the rattlesnake orchid image stacks.  I tried dividing the image sets into smaller groups for stacking then combining the stacked images into another stack.  That didn’t improve the image at all.  I also tried reversing the order of the images in the stack, which did not improve the image either.  One approach that salvaged the set of image captures was to change the objective for that image; merge the captures of the blossoms at the front of the plant instead of trying to have sharp details from front to back of the orchid. The image on the left is a section of the nine slice stack and the one on the right is a section of the stack of the five slices at the front of the subject.  All of the background objects (the stem and farthest blossoms) are a softly rendered out-of-focus.

After photographing flowers, mushrooms, spiders and water drops, I now had 30 sets of image captures made to support my experiments with focus stacking. The first thing I did was to process the same sets of image captures in Helicon Focus software and Photoshop.  In many cases one did a better job of merging the slices than the other, and occasionally neither product was able to cope with the elements in the composition.

Helicon Focus software did much better at combining the in-focus background with the in-focus foreground objects of my first image stack, the orchid from Costa Rica.  However, that’s not to say that all the Helicon Focus composite images were error free.  In fact, new types of errors occurred.  In the Costa Rica orchid image, there were no noticeable halos or misalignments, and the composite images looked good with none of the pinched paper or edge overlap effects displayed in the image processed in Photoshop. One thing to keep in mind is that this image has a sharp foreground object, an out-of-focus background, and the images were taken with the camera and subject motionless.  Also, Helicon Focus provides two processing methods and encourages the use of both to choose the best result.  The image below is the result of Method B.

For convenience, I’ll compare a few of the many focus stacks I made from the subjects I photographed.  The comparison of Photoshop and Helicon Focus image stacks is extensive and can be found HERE.

In the photograph of the rattlesnake orchid plant, the halo effects along the stem and the leaves were greatly reduced in Helicon Focus, but other anomalies appeared.  At first glance it appears that Photoshop includes more area in the composite image. However, if the image were cropped to eliminate the mismatch along the edges of the image, it would be very similar to the Helicon Focus images. It is important to try both methods in Helicon Focus; Method A produced a nice soft background and Method B resulted in a mottled background, so method A was clearly the choice for this images set.  There is a pine needle near the base of the large leaf at the back of the plant.  In the Photoshop images it looks normal, but in the Helicon Focus image it appears in pieces.  On review of the separate slices, I discovered I had absentmindedly removed the pine needle half way through the capture of the image slices.  It hadn’t dawned on me to start the image captures from the start.

Photoshop ignored the fact that all the images did not contain the pine needle and masked the layers to include the whole pine needle in the image.  To remove it by changing the masks of the separate layers in Photoshop might be complicated, but Helicon Focus includes a retouching tool that made short work of removing the pine needle from the final image.   Another strange error occurs along the lower edge of the image which is very soft in an odd sort of way.  Image 1 below is a corner of the method A result above, and image 2 is after retouching using the Helicon Focus retouching tool.  It took less than a minute to make the changes.

The mushroom photo sets provided a challenge for both the products.  In the most difficult set, two clusters of mushrooms spaced an inch or two apart are in the composition and part of the rear cluster is close to the background.  This arrangement provided the opportunity to investigate where the halo artifact appears.  The 25 captures were made at f/14.  Unfortunately I did not take a similar set at a larger aperture for comparison.  Method A worked best of the Helicon Focus options. The entire image is shown at the left and three areas are identified for detailed examination.  The red outline indicates the composition of the second mushroom image described.

The cluster of mushrooms at the bottom of the composition (detail A) was a challenge for both Helicon Focus and Photoshop.  The three mushroom caps overlap and the moss in the background is rendered in focus since the focus slices extend to include the other cluster of mushrooms.  Both have a halo around the edge of the mushrooms that meet the background moss.  Helicon Focus (both methods) has less of a halo, but would require some serious editing to use as a large print.

Detail B below shows Photoshop provides a very good composite image of the overlapping mushrooms with nice detail and good separation from the out-of-focus background.  Method A of Helicon Focus shows a halo at the top of the mushroom cap, and the detail is lacking due to the somewhat muddy looking colors of the mushroom cap.

Detail C below is a crop that includes parts of three mushroom caps near the background.  Photoshop again performed well with good merging of the detail in the mushroom caps, no haloing, and good selection of the background information. The Helicon Focus stack has significant problems rendering detail in the mushrooms and the background.  Reviewing the twelve captures in this stack reveals that one or two of the captures had some additional brightness, probably due to a break in the clouds at the time of the captures.

Helicon Focus did not do well with the first mushroom focus stack but performed much better on other mushroom images taken on the same day under similar conditions.  The camera was moved closer to capture only the top cluster of mushrooms.  Although the aperture remained set at f/14, moving closer reduced the depth of field of each capture.   Helicon Focus Method A (below) did the best at assembling the composite images with very few anomalies, and the background was a nice soft rendering.  Photoshop did very well also but had a few areas (red circle) where the layer selected was not the optimum sharpness.  The locations this happened in this image would have been fairly simple to fix by adding the image with the best detail as a layer to the composite, and mask in the proper sections. Helicon Focus Method B had the worst result with strange green bands and a background that was nearly psychedelic.

Another instance where the use of focus stacking has great benefit for me is when I want a foreground subject sharp, and the background softly out-of-focus, as in the example of the spider and web.  I captured six sets of images and Helicon Focus had a difficult time with all of them due to the movement in the spider web, and exacerbated by variations in the lighting.

Photoshop aligned and merged the four captures in this stack, and the only anomaly in the image is circled above.  It is a section of out-of-focus web that shows larger than the web in focus and somehow gets chosen as being sharpest.  This was easily corrected by erasing the section of the out-of-focus web from two focus slices then merging.  The misalignment in Helicon Focus was partially solved by aligning the images in the stack with Photoshop, saving each layer as a new image file, then combining them in Helicon Focus.  This process worked well with some of the rattlesnake orchid images, and while Helicon Focus assembled the spider correctly (image not shown), it still had trouble with the spider web in some locations.

At first I was mystified by the haloing effect, but since Photoshop and Helicon Focus were exhibiting the same effect in the same places on the image, it was likely not related to the software processing.  In fact it is an optical issue.  To prove this to myself I set up an experiment with a subject that could be arranged so that one part of the subject overlapped a more distant part.  I chose a branch of berries from a tree growing in my yard.  To avoid some of the problems associated with subject movement or changes in brightness, I set up two analog lights indoors, placed the branch of berries in a clamp, and set the camera on a tripod.  I also used the Helicon Remote software to control the collection of the focus slices.  The closest and farthest points of focus are set in the software, which calculates how many focus steps can be made between the two settings.  The user has a choice to reduce the number of captures, but I collected the maximum; I could always eliminate slices from the stack.  Capturing the images in this way eliminated some of the sensitivities of the Helicon Focus software which are that the slices should be roughly equal in steps, should be the same brightness, and should be aligned perfectly.

When a subject is out-of-focus it appears as an ill-defined projection of the subject.  When the focus point is on a background subject, in this case the stem, the berry is out-of-focus and extends beyond the in-focus berry.  When the focus is on the berry, the stem is out-of focus.  The area adjacent to the edge of the berry near the stem is impossible to capture in sharp focus, or at least without the out-of-focus foreground image overlapping.  The images below were captured using the Helicon Remote acquisition control software with the camera set to an aperture of f8.

Neither Helicon Focus nor Photoshop, or any other focus stacking software, have a chance of correctly capturing and merging images of this type due to the optical phenomenon of the out-of-focus foreground subject making it impossible to get in-focus background subjects adjacent to the foreground subject. The examples below are crops (blue outline) of the full frame image above.  The Helicon Focus and Photoshop composite images all fail to render the berry complete, although the Photoshop rendition is clearly closest to complete. Both methods in Helicon Focus appear to have difficulty in selecting the berry parts early in the process select slices that are more out-of-focus than optimum.  They also exhibit significant halo around the berry and below the stem. The fourth image is a copy of the Photoshop image layered with two of the slices that include portions of the berry near the edge and the adjacent one toward the front.  The image requires additional editing on and near the stick, likely layer masks, to completely correct the red color halo on the stick.  Similar editing can be done on the Helicon Focus images using the included Retouching tool although it would most likely require some work using the masking capabilities in Photoshop.

There may also be a benefit to shooting at smaller apertures, and more exploration in that direction is needed. The images below are Helicon Focus processed, one set captured at f/2.8 and the other at f/8.  The image comprising slices taken at f/8 are clearly better as expected; the out-of-focus blur should be smaller at a smaller aperture.  For this composition editing the f/8 image would be manageable as the only difficult areas are between the berry and the stick.  For a more complex image, such as the rattlesnake orchid, editing may be very extensive or even impossible.

Conclusions

Photoshop and Helicon Focus both have their strengths and weaknesses based on my observations. Shooting for focus stacking requires planning, just as photographing panoramas and multiple images for High Dynamic Range processing require planning.  The images must be captured with the processing method in mind.  For focus stacking, the best results will be had when the camera is on a tripod, the subject does not move, and the lighting is constant.  This pretty much defines an indoor setting.  However, much of nature is outdoors, where the light can be variable and the subject may move even though the camera is rock solid.  While Helicon Focus generally does well at putting the image stacks together, added benefits are the Helicon Remote control of the image capture process and a sophisticated retouching tool. Photoshop provides the ability to align images in the stack that might be offset from others prior to stacking and can handle brightness variations between images in the stack, and I belive I will have some uses for this technique.

Focus Stacking in General 
PROs:  Useful tool to extend range of sharpness in an image.
—— — Useful to isolate subject and allow background to be rendered as soft focus.
CONs:  May not be effective in cases where in-focus foreground objects overlap in-focus background objects.

Photoshop
PROs: Auto-Align corrects misalignments of captures prior to merging.
 —- — Auto-Merge handles variations in color and brightness in captures.
CONs: Manual acquisition of image slices can leave gaps.
—— — Retouching final image by correcting mask layers can be tedious.

Helicon Focus  
PROs: Some control over merging process.
— —-  Two processes provided with adjustments.
—- —- Sophisticated retouching tool.
—– — Helicon Remote control of camera ensures sufficient number of slices captured.
CONs: Does not tolerate misaligned images in the stack.
—- —- Does not tolerate variation of brightness between images in the stack.

General Considerations
———-Use tripod or other stable mount for camera and lens
———-Control subject movement (shoot indoors, block air movement, subject choice)
———-Stable illumination of composition (subject and background)
———-Equal focus steps (use Helicon remote or other computer control of camera)
———-Smallest aperture appropriate for image

I haven’t yet decided if I will purchase Helicon Focus, but I will certainly continue to experiment with the free trial, which after 30 days limits the size of images produced and includes some marketing information on the stacked image.  I’ll try additional images of flowers as I’d like to use the technique in my outdoor macro work, and will come up with methods to control illumination and subject movement outdoors.

Focus Stacking – Image Comparison