Photoshop photo stacking, great and ugly - Zerene rescues

[LVF]

The following should be very valuable to those members who are thinking of using Photoshop for photo stacking. Stay with me on this long post, it gets very interesting.

On May 2nd, I posted here, my first attempt at 1:2 close-up photo stacking using Photoshop CS6 (actually, this was my first attempt at photo stacking period). The 1:2 close-up photos used for stacking, were taken with the in-front of the lens 4 feet from the subject being photographed (4 foot working distance). The Photoshop CS6 photo stacking results were somewhat successful but there were slight problems.

In that post I used the Kenko 68mm extension tubes to get the 1:2 close-up photos. Today, I am going to show photo stacking of photos that were taken using the Nikon 17E II teleconverter instead of the extension tubes. The use of the teleconverter also gives 1:2 close-up photographs at a 4 foot working distance.

The photo equipment used consisted of a Nikon D500 camera with the Nikkor 300mm f/4E PF lens with the Nikon17E II teleconverter between the camera and lens. The camera was mounted on a tripod and I used live view to manual focus the lens by rotating the lens focus ring. I used a Nikon MC-36 Cable Release to fire the shutter. I waited a few seconds after touching the focus ring before firing the shutter to reduce vibration.

The subject used for photo stacking was the Lens Align Ruler shown here:



The ruler was laid at a very shallow angle so that it was in full view on the LCD of the camera. The ruler was illuminated with an OTT Lamp which gave a WB around 5000 degrees Kelvin.

Since the vertical height from the front to the back of the ruler was around 2 inches, I had to move the camera setup back to 6 feet (6 foot working distance) to get a 1:3 close-up photo which is approximately 2 inches in height.

Here is the first of 32 photos taken at 6 ft. which shows the bottom of the ruler (I photographed the ruler from the bottom of the ruler to the top):



The following photo may seem irrelevant but its use will become clear soon. Here is the photoshop alignment of the 32 photos in my forming the final stack:



It looks the same as the first photo but it is an alignment of 32 photos, which looks promising.

Here is photoshops blending of the 32 photos:



Wow, the 32 photo stack of the 10.5 inch long ruler looks great. No distortion, no blurs, and has sharp numbers and black squares. I did a great photo stack of 1:3 close-up photos taken at 6 feet working distance. I really have a great setup and Photoshop did a great job at stacking.

Stay tuned, it gets real interesting.

I was really feeling elated. So I decided to move the camera setup up closer to ruler to get a 1:2 close-up stack of the ruler a 4 foot working distance.

Here is the first photo of 20 photos taken at 4 feet:



Looking good. To show that the 20 photos were very good for stacking I am going to show a couple more photos selected from the 20 photos.

Here is a photo showing the "20" number on the ruler:



Here is a photo showing the "4" number on the ruler:



And here is a photo showing the center "0" number on the ruler:



All 20 photos were very sharp and ready for stacking by Photoshop!

Here is the Photoshop alignment of the 20 photos; it should show just the bottom zero just like my successful stack at 6 feet shown above:



What the ___. What happened? I must have made a mistake in stacking the 20 photos in Photoshop. I did it again, same results, did it again very carefully, same results???

Here is the resulting Photoshop blend:



Something is wrong. I got great results with my 6 foot stack. I look at the photos I took at 6 feet and noticed that the right side edge and the bottom edge of the ruler was in the photos. So I am thinking Photoshop needs straight lines like the edges of the ruler to do a good stack. So I rephotographed the ruler with the right edge and bottom edge in the photos.

Here is the first photo of the 27 photos I took with the edges in the photos:



Looking good, edges look good. Here is the center "0" number of the 27 photos:



I checked all 27 photos and they showed sharp numbers, sharp black squares, and good edges. So I proceeded with the Photoshop stacking.

Here is the alignment of the 27 photos:



What in the world is going on with this Photoshop stacking?

Here is the Photoshop blend of the aligned 27 photos:



I give up! Photoshop is really letting me down. I cannot believe what I am seeing. Terrific Photoshop stack of 1:3 close-up photos taken at 6 feet but really bad stacks at 4 feet?

Before I gave up, I remembered Rik Littlefield suggested in my May 2nd post to try Zerene Stacker. So I download the 30 day trial version of the latest beta of Zerene Stacker. I absolutely do not know how to use it, but I forged ahead right after downloading the software to see what it can do.

I fumbled around and finally figured out how to get my first stack out of Zerene Stacker. Here is what Rik calls the ZS DMAP stack of the 27 photos I took at 4 feet:



Wow, it stacked the 27 photos that were 1:2 close-up photos taken at 4 feet. These are the same 27 photos that Photoshop stacked a complete mess.

Obvious, the stacked photo needs some editing to clear up gray areas in the white squares, around the numbers and in the words. But I know nothing about editing in Zerene Stacker so the editing comes later.

I notice that there is another algorithm in Zerene Stacker called PMAX. So I tried it on the 27 photos. Here is the results of the PMAX stack:



I do not have to do any editing of this PMAX stack. It is perfect. I am buying Zerene Stacker. Thank you Rik, you have a great stacking program.

This is a very long post but very informative for those who are thinking of using Photoshop to stack photos. I will be using Zerene Stacker from now on.

Leon

[rjlittlefield]

Interesting... I was not expecting such a dramatic difference between the tools.

Obvious, the stacked photo needs some editing to clear up gray areas in the white squares

Editing is not the right way to tackle this problem. There is a much better method, though it is not obvious.

Half-way through running DMap, the program pauses, offering a slider for the user to adjust. The slider sets a value called "contrast threshold", which is used in conjunction with some other information to figure out which source image to use at each pixel position.

What's happening in your example is that the contrast threshold is set too low, so low in fact that the program decides it should show the edges of OOF blurs in areas that really should be completely featureless black or white. If this sounds crazy, bear in mind that the edges of OOF blurs have more pixel-to-pixel variation than featureless black or white do, and the program works by preserving pixel-to-pixel variation. Unfortunately, the edges of OOF blurs are always some intermediate tone, light and dark grays in your example, so showing those instead of featureless subject produces ugly blobs like we see in your example.

The solution to this problem is to set the contrast threshold higher, so that the program decides what to show based solely on the high contrast regions around the edges of those black/white rectangles.

There is a sort of "preview" of what the slider is doing, which masks out regions of the image that will be ignored for purposes of decision-making.

Historically those masked-out areas have only been displayed as black, leading to the description "black in preview". That's the term that appears in the "How To Use DMap" tutorial at http://zerenesystems.com/cms/stacker/do ... rialsindex, and in all previous discussions of this issue.

However, with recent betas of Zerene Stacker, there is now the option to adjust the mask appearance. With the mask appearance adjusted to look similar to Photoshop's "rubylith", an appropriate setting of the slider for your test chart would look something like this:



In other words, look for any undesired blotches and adjust the slider to mask them out.

The equivalent "black in preview" representation looks like this:



These produce the same result from DMap, but I'm working toward using the rubylith mask as a default in the user interface because I think it will be simpler to explain and use.

--Rik

[LVF]

Thank you for the explanation on the workings of Zerene Stacker. I will study the tutorials and learn how to use Zerene Stacker.

Really appreciate your taking the time to explain this program. And thank you for correcting my miss spelling the word Zerene.

Leon