Motorized bellows

Have questions about the equipment used for macro- or micro- photography? Post those questions in this forum.

Moderators: rjlittlefield, ChrisR, Chris S., Pau

nathanm
Posts: 222
Joined: Thu Jun 02, 2016 8:13 pm
Location: Bellevue, WA
Contact:

Motorized bellows

Post by nathanm »

I have done quite a bit of stacking with both infinity objectives, and conventional macro lenses, mostly by moving the camera + optical train together.

I have read the various posts in the forum on stacks taken "by rail" or "by ring". At low mag (i.e. at less that 1X) by ring is what I normally do.

In a new set up that I am building with a bellows and macro lenses (not infinity objectives!) it seems that I might as well use a stackshot (or other stepper motorized linear stage) to move the camera and the back end of the bellows, while leaving the lens and front of the bellows fixed. That gives the "by ring" approach with the convenience of the stepper motorized macro rail.

It is simple enough - I will take the bellows apart, and fix the front end, while leaving the back end attached to the camera on a rail. It is almost as easy as using the stackshot to move the bellows + camera together.

By moving the camera the lens at the front of the bellows stays put, so the nodal point stays put and the parallax/perspective problems with stacking should be better.

The stepping requirements on the camera end is much more tolerant (bigger step size) than moving the whole set up - once you move higher than 1X magnification, the back focus DOF is much larger than the object side DOF.

I am surprised that I have not found more posts on the forum about setups like this. I read one 2009 post about an Excel macro for making a step size sequence for manually racking a bellows. This is a lot more convenient.

But maybe I have not found the posts!

I use Zerene for controlling the stackshot in other contexts, and one possible problem is that Zerene assumes fixed size steps. The camera side bellows motion steps are in general bigger than the object side DOF steps, but they do vary nonlinearly. In a deep stack the size could vary a lot. One can set the fixed steps to the smallest one, but then you're taking a lot of extra shots. It would be quicker to have variable step size, but that feature isn't in Zerene.

Does anybody have experience with this sort of rig? Any suggestions or comments?
nathanm

barnack-bill
Posts: 36
Joined: Tue Mar 29, 2016 1:55 am

Post by barnack-bill »

This sounds easy to implement on a trial basis moving the bellows manually. Have you done this and how well did it work?

I'm watching this thread with interest as it would be easy enough to fix a stepper motor to the bellows adjustment and set variable steps on an arduino driving the motor.

mjkzz
Posts: 1681
Joined: Wed Jul 01, 2015 3:38 pm
Location: California/Shenzhen
Contact:

Post by mjkzz »

why not move the front of bellow? just curious

nathanm
Posts: 222
Joined: Thu Jun 02, 2016 8:13 pm
Location: Bellevue, WA
Contact:

Post by nathanm »

I am definitely going to try it manually.

The reason to move the camera is to keep the lens to subject to lens distance fixed. This makes the perspective and parallax change less, which makes stacking easier, usually.

There are several discussions of this on the forum; here is one from the Zerene website http://zerenesystems.com/cms/stacker/do ... versusrail
nathanm

elf
Posts: 1416
Joined: Sun Nov 18, 2007 12:10 pm

Post by elf »

Changing the bellows draw works quite well up to 5X and doesn't work well above 10X. If you don't have a telecentric lens, then it is the only reliable way to do macro panoramas.

Here's a link to my desktop setup: http://www.photomacrography.net/forum/v ... php?t=8348

The software is pretty straight forward. Use Lefkowitz's macro DOF formula to calculate the DOF for each individual step and move the camera accordingly.

barnack-bill
Posts: 36
Joined: Tue Mar 29, 2016 1:55 am

Post by barnack-bill »

I'm very taken by this idea and since my very basic rig was set up I had a quick play.

Image

I reversed the bellows completely so the rails came back under the camera and because the weight distribution was all wrong I added a makeshift rear support (bit of nylon rod) though with the fairly light Sony A7 it possibly wasn't necessary. It was quite easy to use and the results were comparable with other methods. On this first try and from my limited experience I couldn't judge whether better or worse but it was noticeably easier to stage and light as there were no moving parts (rails or lenses) near the subject.

This is at about 3x (full frame, fly wingspan about 8mm)

Image

It will be interesting to see the results from the OP who no doubt has more experience and a more sophisticated rig.

rjlittlefield
Site Admin
Posts: 23564
Joined: Tue Aug 01, 2006 8:34 am
Location: Richland, Washington State, USA
Contact:

Re: Motorized bellows

Post by rjlittlefield »

nathanm wrote:Does anybody have experience with this sort of rig?
To the best of my knowledge, elf is the only person who has routinely used automated focus stepping by rear bellows extension.
The camera side bellows motion steps are in general bigger than the object side DOF steps, but they do vary nonlinearly
True, but my personal feeling is that the non-linearity is less of an issue than it might seem at first.

With a fixed aperture setting on the lens, the ideal rear extension step size (by wave optics calculation) varies as the square of the effective aperture.

Of course the exposure changes by the same amount.

As a result, if the ideal step size varies "a lot", then so does the exposure.

The change in exposure can be compensated by levels adjustment in processing, but the most convenient automated methods such as brightness adjustment in Zerene Stacker and Helicon Focus are only approximations, at least when normal color profiles are used.

As a result, my usual recommendation is to avoid large changes in extension unless there's a really compelling reason to use them.

One example of a compelling reason is provided by Charles Kreb's OlyMinCan, shooting from 1:1 to infinity using a 28 mm lens on a full frame camera. But in that case the effect of extension on step size and uneven exposure was probably reduced by the use of a retrofocus lens whose exit pupil is farther forward than you would expect from just its focal length.
Any suggestions or comments?
It's always a matter of experimentation to see how the tradeoffs go.

But if I wanted a truly perfect result then I would think about controlling the illumination intensity also, increasing its brightness in synchronization with the change in effective aperture.

When using asymmetric lenses, all the calculations including step size should incorporate the pupil ratio also.
barnack-bill wrote:On this first try and from my limited experience I couldn't judge whether better or worse but it was noticeably easier to stage and light as there were no moving parts (rails or lenses) near the subject.
It seems to me that this improvement is due to reversing the bellows, not to focusing by rear extension. Am I missing something?

--Rik

Edit: change "this" to "the non-linearity", to clarify that it's the non-linearity that is less of an issue than it might seem at first.
Last edited by rjlittlefield on Fri Jun 24, 2016 5:44 pm, edited 1 time in total.

nathanm
Posts: 222
Joined: Thu Jun 02, 2016 8:13 pm
Location: Bellevue, WA
Contact:

Post by nathanm »

Any reason why elf (and now maybe me) are the only ones? It seems to make a lot of sense, and not be that hard. Saving time messing with stacking artifacts is a lot easier than just making this set up.

---------
I have thought about varying the illumination, but that is not easy.

I either use studio strobes, or older Canon speedlights (580ex) on manual, which typically need to be at very low settings (1/64 power to 1/32 power) which is their minimum power. The studio strobes are typically already on the minimum setting.

In principle one could use the Canon (or Nikon, or other) speedlights with a smart controller where I could tell the controller via a programmable interface to dial down the power, but I'm not aware of an easy off the shelf way to do this. Also, the flash would have to be diffused or set farther away to the point that it was not at the minimum power setting.

One could use continuous light, assuming that one had a light with easily programmable light levels. I like flash as extra insurance against vibration.

If anybody knows of digitally programmable flash, or digitally programmable continuous light (LED ?) I am eager to hear about it.

Or, one could use continuous light and change the shutter speed, but that might mean changing it more than would be good for vibration reasons.

The most easily programmable setting is the camera ISO - that could be varied pretty easily. To get a 2-stop variation, you could go from ISO 50 to ISO 200. On most modern DSLR you could go to ISO 400 with no noticeable loss of quality. ISO can be changed programmatically pretty easily.
=============
But, I don't understand the point about large step size leading to large exposure variation.

In a very deep stack (physically deep, so large distance from foreground to background) then the exposure at the closest level will be different than the farthest.

So in the example of a stack like Charlie's set up (or like the new Laowa 14mm, which I have but haven't used seriously yet) where you go from 1:1 to infinity, the effective f-stop will go up by 2 stops. The focal distance doubles and the lens has the same aperture.

So there will be 2 stops of exposure difference between the closest frame and the farthest frame. That is independent of the number of frames in the stack in between. You will always have 2 stops total variation from front to back.

If you do that in (made up example) 20 variable size steps, or 100 constant size steps you still have 2 stops of difference from back to front.

Or, are you saying that Zerene and Helicon have a limit on frame-to-frame exposure difference? But don't have a limit (or less of a limit) on total exposure difference?

Finally, I am not talking about doing 1:1 to infinity stacks - at least with this set up. It is really aimed at 1X to say 5X, and no individual stack would be super deep.
nathanm

Charles Krebs
Posts: 5865
Joined: Tue Aug 01, 2006 8:02 pm
Location: Issaquah, WA USA
Contact:

Post by Charles Krebs »

Any reason why elf (and now maybe me) are the only ones?
Since here we are (sort of by "definition") working at 1:1 and higher there is less incentive to do this. At higher magnifications "parallax error" stacking artifacts are typically not a significant issue. Elf was heavily involved in stitching together large numbers of separate images so it made much more sense for his project. Depending on the imaging parameters, if you use rear bellows to step focus you can wind up with a pretty significant variance in the magnification and thus field-of-view from the front of the stack to the rear.

Nevertheless, the idea of keeping the lens aperture in exactly the same location is enticing. It seems the lighting variation would be perhaps a more vexing problem to solve than the bellow rear standard automation. As mentioned, nothing comes readily to mind using electronic flash. Continuous light seems more "doable" When I dabbled with the "OlyMinCan" I finally settled on just using auto-exposure with a three frame exposure bracket at each step. But auto-exposure (or auto-anything for that matter) is generally not a good idea when acquiring a stack of images.

fotoopa
Posts: 194
Joined: Sat Dec 14, 2013 3:14 am
Location: Belgium
Contact:

Post by fotoopa »

Charles Krebs wrote: It seems the lighting variation would be perhaps a more vexing problem to solve than the bellow rear standard automation. As mentioned, nothing comes readily to mind using electronic flash. Continuous light seems more "doable" When I dabbled with the "OlyMinCan" I finally settled on just using auto-exposure with a three frame exposure bracket at each step. But auto-exposure (or auto-anything for that matter) is generally not a good idea when acquiring a stack of images.
With LED flash, it must be able to work more easily. My LEDs are now set between 0 and 40000 usec. The most common used values are between 1000 and 8000 usec. The flash time pulses now remain fixed for the entire step cycle. But nothing would stop me to vary the flash times as functions of the step range. The exposure effect can be measured once in advance with a gray card. So you know immediately the effect of start and end position and possibly also in intermediate positions. For those times Roughly to calculate it may not be so difficult. These calculations can be done in real-time or in advance by a simple list. Just some extra program code would be required in the FPGA.

Currently I do not have an arrangement to carry out the tests. But I'm thinking to test the Raynox-150 / Raynox-250 with bellows. Tests could be included in that combination setup. This would be a great challenge for fotoopa!

barnack-bill
Posts: 36
Joined: Tue Mar 29, 2016 1:55 am

Re: Motorized bellows

Post by barnack-bill »

rjlittlefield wrote:
barnack-bill wrote:On this first try and from my limited experience I couldn't judge whether better or worse but it was noticeably easier to stage and light as there were no moving parts (rails or lenses) near the subject.
It seems to me that this improvement is due to reversing the bellows, not to focusing by rear extension. Am I missing something?

--Rik
Not having to allow for a moving lens and bellows standard made lighting easier with the two light sources I use. Each is 2 x 48 LED arrays in a diffuser and were designed for close up rather than high magnification macro work. With the lens fixed I can position the diffusers with more freedom, even touching the lens body. Possibly a benefit of value only in my set up.
Last edited by barnack-bill on Sat Jun 18, 2016 12:25 pm, edited 1 time in total.

fotoopa
Posts: 194
Joined: Sat Dec 14, 2013 3:14 am
Location: Belgium
Contact:

Post by fotoopa »

At my current setup is the flash LED module is on a separate mechanically track. The exposure across the object remains at the same distance. I have 6800 lumen power, hence the short exposure times who works as a flash. I use 6 channels to control the light direction If necessary.

See: http://www.photomacrography.net/forum/v ... ht=fotoopa

mjkzz
Posts: 1681
Joined: Wed Jul 01, 2015 3:38 pm
Location: California/Shenzhen
Contact:

Post by mjkzz »

thanks, I knew I am missing something when I asked that question (why not move the front of bellow).

It is not difficult to change light intensity "continuously" of a speedlite like Canon 580EX without any modification to it but with a custom trigger circuit. I have even modified a YN560 (first gen) for this purpose -- changing its output "continuously". But that modification is kind of messy, it involves opening YN560 up and solder some wires.

The reason I quote "continuously" is because most speed lights are not very consistent from charge to charge, but if we allow enough time for it to charge, we can achieve better consistency

mjkzz
Posts: 1681
Joined: Wed Jul 01, 2015 3:38 pm
Location: California/Shenzhen
Contact:

Post by mjkzz »

oops, submitted too fast.

I think it is the synchronization between motor and light intensity change is the problem -- what are the relationship between the two? Linearly changing light intensity or some other formula?

rjlittlefield
Site Admin
Posts: 23564
Joined: Tue Aug 01, 2006 8:34 am
Location: Richland, Washington State, USA
Contact:

Post by rjlittlefield »

nathanm wrote:But, I don't understand the point about large step size leading to large exposure variation.
I don't see where anybody said that large step size leads to large exposure variation.

What I said -- or at least what I intended to say -- was that large variation in ideal step size goes along with large variation in exposure.
If you do that in (made up example) 20 variable size steps, or 100 constant size steps you still have 2 stops of difference from back to front.
I agree that in your example of infinity-to-1:1 the variation will be two stops of exposure (a ratio of 1:4) no matter how big the step size happens to be. The ideal step sizes will also vary by a range of 1:4. Both exposure and ideal step size vary in proportion to effective aperture squared.

However, the numbers I get are that if you need 100 constant size steps to cover the 2 stops range, then you need 50 ideal-sized steps. So the penalty for using constant size steps, in this extreme case, is to increase the number of frames by 2X, not 5X as your numbers suggest.

With less extreme ratios, the penalty for using fixed size steps is smaller. With a 1-stop difference, I calculate around 1.35X.
I am not talking about doing 1:1 to infinity stacks - at least with this set up. It is really aimed at 1X to say 5X, and no individual stack would be super deep.
Perhaps it will help to work some numbers.

Suppose your subject is as deep as it is wide, and you're shooting with full frame, at 1X at the front of the subject, and you're using a 100 mm lens. Then when focused on the front of the subject the lens will be 200 mm away, with the sensor 200 mm behind that. To focus on the back of the subject, the front focus distance increases to 200+36=236 mm while the rear bellows extension drops to about 173 mm (=1/(1/100-1/236)). That's a rear extension ratio of only 173:200, about 1:1.16. The corresponding penalty for fixed versus ideal steps will be about 15% more frames, and the effect on exposure is less than half a stop.

Using a shorter lens makes both penalties worse. With a 50 mm lens, I get that you're looking at rear extension from 100 mm to 79.07 mm, about 1:1.26, with about 25% more frames for fixed versus ideal, and a little over half a stop.

In both cases, if the ultimate concern is for image quality, then it seems to me that variation in exposure would be the first thing to worry about. The increase in frame count by using fixed step sizes is not prohibitive, and the extra frames will have little or no effect on image quality.
Or, are you saying that Zerene and Helicon have a limit on frame-to-frame exposure difference? But don't have a limit (or less of a limit) on total exposure difference?
In Zerene I don't recall that there's any crisp limit for exposure correction. The quality of the correction just goes down as the variation goes up. Probably it's better if you use a linear profile (gamma=1), but I've never experimented to confirm that. I have no idea what Helicon does.

If you wanted to get a perfect correction for exposure, another place to do that would be in raw-to-RGB conversion. I haven't checked, but I would not be surprised if dcraw has a parameter that would let you do that.

--Rik

Post Reply Previous topicNext topic