How many single shots for focus stacking?

The number of individual shots required for a focus stacking photo is usually determined using a fairly complex calculation procedure. In order to make the procedure easier, I would like to introduce another, considerably simpler method that allows me to do without any abstract calculations.

The number of individual shots required for a focus stacking image like this one from the beetle Polybothris sumptuosa is relatively easy to determine as an approximation

To create a series of single shots for focus stacking, you need to determine two key data: the length of the shooting section with start and end points and the number of individual images. Once you have those values, the step size is determined almost automatically.

The first is usually quite simple: You move your camera on the linear stage so close to the object that it begins to produce sharp contours on the monitor – that, to put it simply, is the starting point. The end point of the shooting path is where the sharpness ends and everything becomes blurry again. You have now determined the total distance, whereby you yourself have to determine how far the rear part of the object – e.g. B. the abdomen of an insect – should also be shown sharply in the photo. The longer this distance, the greater the risk of undesired image disturbances, which are referred to as artifacts – but more on that in a separate article.

Determining the number of individual shots required is usually a much more complex matter. The textbook procedure is to first calculate an approximate value for the depth of field of the lens – using a physical formula (e.g. wavelength of light divided by numerical aperture to the power of two) or an online calculator. The result is a theoretical numerical value for the necessary step length. Then you divide the total distance by the step length and get the theoretical number of individual steps required. But because the software that we want to use afterwards requires a strong overlap of the individual focus zones, a divisor is added by which our step length must be divided. This divisor should be between 3 and 6, depending on the desired image quality. This calculation results in the number of individual shots required as well as the distance that the camera or objective lens must be moved closer to the object per individual step.

That sounds complicated and it really is. High-quality, precise linear stages enable this distance to be read on a scale using the rotary handle for propulsion, and good control devices on a motorized linear table allow the value to be entered. You can do it that way, and it actually leads to good image results – no doubt about it.

The required number of images is entered into the control unit – here Castel micro from Novoflex – as well as the start and end positions

However, this approach is too abstract and too complicated for me. On the one hand, I am a self-confessed math dyslexic, and on the other hand, as a non-fiction author, I am always looking for simplifications to make complex procedures easier to understand. That's why I have developed a method for my work with which I can very easily determine at least a rough estimate of the number of steps required – and that is always absolutely enough for me to work. It should not be overlooked that we no longer work analogue with film material that has to be developed, but with a digital sensor that shows us the image even before it is taken. We can also work fully automated, so that the entire procedure can be repeated at the push of a button. I want to make use of these advantages.

The shooting distance

I proceed as follows: I first determine the starting point of the shooting path – i.e. the camera position for the first shot. To do this, I adjust the focus (the camera position) so that the object on the large computer monitor (which I always work with!) is still completely out of focus. Not a single detail should be seen sharp, but I am close to sharpening the detail closest to the lens. This is exactly my starting point for the first shot. I don't worry about the many blurry structures in the image because the software will eliminate them later. This point is given to the control unit – in my case it is a Castel micro from Novoflex – as a starting point.

In order to find the end point of the shooting path, i.e. the position of the last photo, I move the camera forward motorized. When doing this, the focus zone slowly moves over the entire object until everything is blurred again after the image was in focus. This should be taken absolutely literally, because even the last detail of the object must have completely disappeared into the blur. This is the position of my last individual photo.

The first shot of the distance can be seen on the left and the final shot on the right. In between there are individual images in which the narrow zone of focus moves from front to back. You should be careful at the end, because you often forget the last bit of sharpness in the rear object areas. Here you should extend the total distance far enough back, even if the front parts of the subject are already extremely blurry, as can be seen in the picture on the right. Unnecessary images can later be deleted.

The number of images

If you do this, you will have determined the whole shooting path and developed a feeling for how long it is, and your control unit will now also know. What follows is purely a gut feeling, but it quickly emerges through experience.

Factor number one: The more your lens magnifies, the shallower the depth of field will be – so you will need more individual shots. At 5x it is significantly less than at 10x, and at 20x it is significantly more. How many you need also depends on factor two: the length of the path.

Factor number two: How much depth of subject do you want to fit into the depth of field of your photo? Example: If I have a fly in front of the lens but only want to get a sharp image of its head, this distance measures, for example, two millimeters. This requires far fewer individual photos than if I wanted the entire body of the fly to be in focus from front to back, because that is a distance of perhaps ten or twelve millimeters. The number of individual images per millimeter is the same and depends on the magnification.

Factor number three, which you can ignore with automatic control: the angle of the subject surface to the sensor. The more parallel these two surfaces are, the fewer individual steps you need, e.g. at the front of the fly's head. Theoretically, as the number of degrees of this angle increases, more and shorter individual steps would have to be carried out. You can take this into account when stacking manually, because you then work while looking at the monitor and can vary the step length by eye. However, this is not possible with automatic propulsion, and if you take too few or too large steps, you will later discover blurred lines in the image. These are particularly easy to see on the side surface of the compound eyes, where it is at an almost right angle to the camera sensor.

This all still sounds complicated, but it's actually quite simple: I know how much the lens I use magnifies, e.g. B. 2:1, 5:1 or 10:1. So I have an intuitive idea of whether a few individual shots per millimeter are enough for me or I need a lot of them. Now I estimate my desired recording distance: Do I have an entire insect body in front of me with a long recording distance? Or am I photographing the wing of a butterfly, which is very two-dimensional and requires a very shallow shooting depth?

Repeat the shooting process

For an insect body with a length of 10 mm and a 5x lens (i.e. magnification 5:1), I would have a gut feeling and start with 150 images after determining the exposure distance. If I then look at the automatically generated and readily stacked image upscaled to 100 percent or more, I can immediately see whether there are any blur stripes that indicate that the individual focus zones have not overlapped sufficiently. Then the recording distance is too long for the number of images or the number of images is too small for the distance. Depending on the width of these blur stripes in the photo, I then have to enter a correspondingly larger number of shots for a repeat. How about 250? Try it out – the system does it at the push of a button, because the recording distance from the start to the end has still been entered. During the next pass, the blur stripes should no longer be visible in the finished photo, or at least should be significantly narrower.

Focus stacking photograph with cropping, 25 single shots

Focus stacking photograph with cropping, 50 single shots

Focus stacking photograph with cropping, 100 single shots

Focus stacking photograph with cropping, 200 single shots

These four images with cropping for detail show the same shooting path with different numbers of individual images (25, 50, 100, 200) with otherwise identical settings. You can see that if the number of images is too small, blurred stripes appear between the individual focus zones (arrows), which become narrower as the number of images increases. These blur stripes become more noticeable the more the stacked photo is enlarged.

Second example: I take a photo of a butterfly wing with a 20x lens. The recording distance is very short due to the two-dimensional nature of the subject, only slightly more than the thickness of a few sheets of paper. But I know that at 20x I need a lot of individual exposures per millimeter of exposure distance. That's why I start with at least 200 shots – for a test image that I can use, for example, to optimize the lighting for the shadow relief. Once everything is set optimally after a few test shots like this, then comes the final series of shots I make to create the high-quality photo. Here I choose at least 300 individual shots.

However, if my butterfly wing is not actually at a precise 90 degree angle to the optical axis, i.e. it is not perfectly parallel to the sensor plane, then this extends the necessary recording distance, sometimes even considerably. In this case, I may need significantly more individual images. That's why it's important to always look for blur lines in the finished image and to leave the control unit settings unchanged in order to possibly repeat the recording with more individual images.

The step length of the feed automatically results from the other values for total distance length and number of images. Focus stacking control devices automatically calculate this step size after entering the start and finish points and the number of steps.

Manual linear stage

In principle, the same applies to manual focus stacking with a linear stage with a fine adjustment screw, except that you do not first determine the entire shooting distance from start to end. You simply move the camera to the starting point and begin your photo series with tiny advance increments, but under visual control on a large computer monitor that is connected directly to the camera and shows you the live view image. Here the feed can only be so slow that the individual sharpness zones massively overlap each other. This can be clearly seen by looking at the monitor, as long as you are not looking at the small camera display but at a connected large computer monitor.

When looking at the monitor, the strip-shaped focus zone is clearly visible, here located behind the eyes. This makes working much easier than with a small camera display. That's why you should, if possible, choose a camera model that allows you to connect an external monitor in addition to the remote shutter release for the motorized linear stage.

At the beginning of such a manual series of photographs, you don't yet know how many individual photographs you will take, but you already have a rough idea based on the image scale (2x, 10x or 20x?) and the three-dimensionality of your object.

For macro shots (up to 1:1), e.g. B. the blossom of a flower, five to ten images are often enough because the field of focus (depth of field) is still quite wide. A macro lens also has an aperture that, when set appropriately, increases your depth of field.

At a microscopic magnification scale of 20:1, however, there are rarely fewer than 300 individual images, sometimes considerably more. It is important that your individual movement steps are not too large. When working manually (which happens very rarely), I achieve the best results when every tiny detail of the object, e.g. a certain bristle on the fly's head, is within the depth of field zone that I can see on the monitor in at least three of the images. It is not enough that the individual depth of field zones lie one behind the other and cover the entire object, because the focus stacking software requires large-area overlays of the individual focus zones.

Of course, you can certainly use the orthodox method to precisely calculate the theoretically necessary number of steps. However, in my experience, the physical-mathematical method does not bring any advantages that would be visible in the image compared to the rough estimate of image number and step size presented here. This would only be useful to avoid too many shooting steps when working manually, as these cost time and effort. But as mentioned, I always use the technical possibilities to automate this process, even on a conventional microscope (which I would like to discuss in a separate article).

And ultimately, the method presented here is also similar to what happens with internal focus stacking or focus bracketing in modern mirrorless system cameras (which I will also discuss in a separate article): you determine the number of individual shots based on gut feeling after you have developed an idea of the length of the entire distance, e.g. the body length of the insect you want to capture.