UV-reflectance macro z-stacking

[Daniel Geiger]

Below I want to provide some experiments with UV reflectance z-stacking. Feedback welcome!

 

Z-stacking background. In z-stacking, multiple images in successive focal planes are taken and are then computer processed into one image with much greater depth of field than could be achieved with a single frame. This is particularly useful in macrophotography, where stopping down to the lens leads to image degradation due to diffraction.

 

Z-stacking hardware. I use a full spectrum Canon full frame dSLRs 5D mark II, a variety of lenses, mainly Zeiss Makroplanar 100/2, Canon MP-E 65 mm, Nikkor EL 80/5.6 (old, not N). I generally shoot lenses wide open for maximum sharpness, which implies small stacking increments. I provided a table for stacking steps, and a bit more background on stacking of small flowers:

Geiger, D. L. 2013. Imaging small orchid flowers using visible light. Orchid Digest 77: 112–123.

pdf available from my website under http://www.vetigastr...hids/index.html

The Cognisys StackShot motorized and computer controlled focusing rail (or something similar) is critical for consistent quality work.

 

UV z-stacking considerations. Stacking steps do not need to be adjusted because of the shorter wavelength, because depth of field is independent of wavelength (and also independent of focal length). The only variables to consider are f-stop on lens and magnification of image.

 

Nikkor EL 80/5.6 for macro. The lens is used with a step-up ring M39 to M42 on a Praktica M42 bellows ($20 off eBay), then a M42 to EOS adapter, Kenko EOS extension rings, 5D2. This set-up permits magnifications up to about 4:1, which is just about at the diffraction limit of a f/5.6 lens (effective f-stop f/28). I have not tried to reverse the lens yet.

 

UV image acquisition. There is not much different. I use a 365 nm 5W LED flashlight for illumination, and aluminum foil as a reflector for more even lighting. Paper usually fluoresces in UV, and most plastics absorb much UV. A 2" Baader U Venus filter is on the lens with 34.5 mm to 48 mm adapter.

 

Post processing. One problem is that during camera conversion, some of the Bayer filter elements most likely got damaged. This is no big deal for single captures, but if you stack 50 or more, one gets a lot of "creative" color streaks. That leads to a lot of post-processing after stacking. I thought of doing a dust delete data in the 5D2 body, but that does not work, because during conversion, the dust shaker hardware is removed, and without it, the dust delete data cannot be obtained.

 

Below: Blow-up of color dot trails due to Bayer mask damage.

 

File prep in DxO Optics Pro. Run the stack through DxO to adjust exposure and curve. In previous tests, I have found DxO to be quite a bit faster than PS.

Be sure that the source files are all in the same orientation, because the orientation of filter mask in next step needs to match the file orientation.

 

Affinity Photo color dot removal procedure. In the following, I provide a step-by-step guide on how to do efficient post-processing in Affinity Photo; I prefer perpetual licenses, so have abandoned Adobe after using it for 20 or so years; most likely, something similar can also be done in PS.

 

Because these color dots are in fixed positions, they can be screened out with a Gaussian blur mask. Do the following in Affinity Photo:

- Open a dozen images in AP.

- Add a Gaussian Blur live filter on top of the first image.

- Add a mask to the Gaussian Blur live filter. Blur of about 2.5 pixels worked well for me, but that depends a lot on your camera and the type of Bayer damage you have.

- Fill mask layer only with black.

- Paint with a small white brush (4 pixels worked well for me) over any color spots in the mask layer. Do that at 150-300%. You'll spend some quality time at this. Don't bother fix those color spots right at the edge of the frame. The Gaussian Blur from the edge of the image is going to mess it up.

- Once done with first image, copy-paste Gaussian blur filter with Mask into next source image.

- Check for pesky color spots in second image and refine mask.

- Copy-refine for the the rest of the images.

 

Open one of the CR2 source images, develop it. Now start macro recording and paste the final version of the Gaussian Blur live filter with mask on top of the image layer. Stop macro recording and save it as something like "UV prep". The cool thing in AP is, that the live filter with mask is saved in the macro step. So the macro does not paste from computer memory, but the stored filter-mask. That means even if you quit the application, the Blur-Mask is still stored. Super cool.

 

In Open-batch, select the Macro, select your output format (e.g., tif for Zerene), add files, and press OK [Another cool thing in AP, saving/export is not part of macros, but is in the batch dialog. No more infuriating confusions about save functions as in PS].

 

Computer Z-stacking. PS is famously useless for macro z-stacking. I use ZereneStacker. With AP files, make sure that in ZereneStacker the "Use external tif reader" is off. Otherwise, the AP files are read as either B&W or single channel files. The pmax algorithm works quite fine for me.

 

I also tried APs focus merge function, and am quite impressed, but Zerene still has the edge. I have not found a way to touch up the final stack layer by layer, as you can in ZS or HF. If I stack with the color artifacts still in place, AP has trouble with proper alignment, while Zerene nails it. Last but not least, tone mapping is much better in Zerene than in AP. This could be fixed before feeding the files into AP.

 

The specimen here is Barbosella duesenii (Orchidaceae) about 2:1, about 50 frames, about 4s exposure each. Nikkor EL 80/5.6, Praktica bellows, Baader U Venus, Canon 5D2 full spectrum, Cognisys Stackshot, 365 nm 5W LED flashlight, white balance on white plastic.

 

- Stacktrail Zerene. Notice the color streaks, but clean edges of flower and good tonality straight out of stacking.

 

- Stacktrails AP. Notice color streaks, and edge of flower is blurry (alignment issue) and image is overexposed.

 

- AP stack: stack of images after batch processing with Gaussian blur mask in AP. Flower edges are clean, but still a bit over exposed. It seems that AP-alignment is misled by color artifacts.

 

- B&W Zerene: Cleaned tifs stacked in Zerene, with use external tif reader chosen, resulting in B&W image

 

A second example of a bit better shot is Oberonia costeriana (Orchidaceae), flowers are around 1.5-2 mm tall.

 

First image is visible light. Z-stacked on Zeiss Discovery V20mot with Planapo 1.5x, Objective slider, Axiocam HRc, illumination with dual gooseneck fiber optics diffused with ping-pong balls. About 30 frames.

 

Second image is UV reflectance, about 3:1, Nikkor EL 80/5.6, Praktica bellows, Baader U Venus, Canon 5D2 full spectrum, Cognisys Stackshot, 365 nm 5W LED flashlight, aluminum foil reflector, about 40 frames. Processed as detailed above.

[ecliptique]

Hi, Daniel,

 

I use z-stacking since many years with macro equipment or microscope. I belive that it's possible to get better results with this technique.

 

I use also Cognisys StackShot and Zerene or Helicon Focus softwares with Nikon D800 broadband and many different lenses.

 

 

Look here for full resolution version of a Gazania picture (30 frames + EL Nikkor 105mm + Baader U + 3 UV modified SB24 Nikon flashes)

 

Laurent

[Andrea B.]

Daniel, thank you for the nice write-up about your stacking techniques.

The Cognisys is a cool tool. I decided though, that I did not have the patience for it and sold it to Bjørn. I think he does not have the patience for it either. :D :D :D But we love seeing the nice stacking results that others obtain.

 

Daniel, who performed this conversion? If they did scratch the sensor pack, then that is not a good thing! Could it possibly be bits of lint?

 

---

 

Here are some links to stacking techniques which might also be useful to stacking newcomers.

These are free E-Books. If you click the link, the books will download in PDF form.

The Art of Focus Stacking - A Primer: http://spiritgrooves...ocuStacking.pdf

The Art of Focus Stacking - Book Two: http://spiritgrooves...FS_Workbook.pdf

 

 

Laurent, that is a beautiful Salvia.

[ecliptique]

Laurent, that is a beautiful Salvia.

 

Thank you for identification Andrea. I'm just a poor photograph and not a botanist ! Sorry.

 

Here is another 16 frames stack fluorescent picture. Rose de Noël en français ! :)

 

 

And a fluorescent tulip (26 frames)

 

[Daniel Geiger]

Andrea: It is a Bayer pattern issue, not lint. I can see individual pixels either all red, blue, or green, no halo, once I magnify the file sufficiently on screen. Just read David Pruchi's book, and notice that Bayer pattern is under a separate protective layer, other then the hot mirror. Accordingly, I am no longer sure that this is a problem with conversion, or whether those already existing errors are just more prominent in a narrow band UV reflectance image.

 

Laurent: I whole-heartedly agree, there is room for improvement. However, there is a major difference between imaging a 2 mm flower and a 2 cm flower, and stacking one flower vs. a multifloral inflorescence.

 

Next step is converting a couple of flashes. Should be fun.

[Andrea B.]

Hmmmmm.....I'm thinking that it must then simply be bad (hot?) pixels on the sensor. Does the 5D2 have a routine for "mapping out" bad pixels? There are some apps which automatically take care of that during conversion. I think Iridient Developer is one such.

 

Was it an older D52 which was converted? There are little burn-outs that happen over time. In the past we have discussed whether exposure to UV light might damage a sensor in some way over time. No conclusions reached however.

 

Quite interesting how we see the path traced out by the stacking -- although quite an annoyance to you having to deal with it.

 

Good luck with the flash conversion. Be very careful with any lurking capacitors!!

 

Added later:

I was wondering if you have yet experimented with fewer stacks at, say, f/11 versus more stacks at f/5.6??

Just curious as to which might prove more useful for such tiny flowers.

 

Also, it might be really useful to rent a CO/60 for such work? About the sharpest in UV that I've ever seen.

[Daniel Geiger]

Hi Andrea, It is an older body, January 2009 first image in my database.

 

There is a "Dust delete" function, but after conversion, in which the dust shaker hardware is removed, dust delete function does not work anymore. Accumulation of bad pixels is certainly a possibility. It's not just one or two, but rather 50-100. Not something I want to clean up manually for every shot. So I want something a bit more automated. Last night, I thought whether a digital color-noise filter could do the same.

 

Re smaller f-stop, you quickly run into diffraction issues at >>1:1. I haven't tested the EL 80, but on the MP-E 65 f/2.8 is markedly sharper than f/4; apparently on that lens, there is quite a bit of copy variation. Consider that f/5.6 at 4:1 effective f-stop is 28, so pretty close to f/32 where diffraction blur shows up at 8x10 prints. On the 5DsR with ~4 µm pixels, maximum information content is around f/11, so at f/2.8 maximum magnification is 4:1. Above that you only get empty magnification. On 5D2 with ~6 µm pixels limit is at about 6-7:1. That is all with visible light. There UV will give a bit of an advantage by a factor of around 550/380 nm ~ 1.4 (= 1 f-stop). Again, no hard rules, just some ball bark indications.

 

Re number of images in stacks, that I have experimented with and the 70% of DOF is quite spot on. Zeiss programs 0.2 into their microscope stack acquisition module in Zen Blue, but that is WAY too narrow. Ran test with 0.1, 0.2, 0.3 .... 1, 1.4 and 2x DOF. I see no difference up to 1xDOF, so 70% gives a bit of safety margin without going overboard. DOF is wavelength IN-dependent, so no changes in UV.

 

Will play a bit more with the EL 80, and if I like the overall idea, I will buy a proper UV lens. My project is on-going, so renting makes no sense. I first want to keep parameter space simple, and get consistent results. Then add elements, if warranted.

 

Have you put the CO60 on a bellows?

[Andrea B.]

Re number of images in stacks, that I have experimented with and the 70% of DOF is quite spot on.

 

Good to know!

 

Have you put the CO60 on a bellows?

 

I have not had the CO60 on a bellows. I have had it on a P11 extension only. But then I'm mostly making botanical documentation "in the field", so stacking and true macro is impossible (well, almost) when the little breezes are blowing and tossing the flowers around. :D

 

Although whenever I see these interesting focus stacks like you and Laurent are making, I think that I really must try this again sometime. I do encounter many common wildflowers not on park properties which could perhaps be occasionally picked and taken home for a studio stacking attempt. And there's always the garden flowers, too.

 

P11 -- this was the extension for the UV-Nikkor 105/4.5.