Focus Stacking...First try.

[f8andbethere]

My first attempt at focus stacking. Not as easy as some of the You Tube videos would have you believe. Before I buy software I decided to download the free CombineZP. The results are OK, not sure if the pay for software is any better. I'm sure there is a learning curve and I just started so I'll give this some time.
Anyway, I went to the local grocery store to find a suitable subject and came home with this Lilly. I also just bought my first dedicated Macro lens, a Sigma 150mm f/2.8 APO-Macro DG OS EX HSM and all the other letters behind it that go with it. Distance was about four feet. For light I used a Nikon SB-800 with a ring light attachment on my Nikon D700. A piece of black fabric for the back round.
I tried this several times, using different amounts of images. The first was twelve, it had some strange ghosting around it. Then I tried twenty five images but the flower had drooped while I was shooting, the results were less than impressive. The last attempt was to stop down a little more from f8 to f16 and only use four images. This seemed to work the best.
Let me know what you think. Tips and critiques are surely welcome.

[Glenn NK]

Dean:

It looks fine to me - only one small nit - perhaps allow more space around the flower and leaves as it's a bit cramped in the frame.

Some may not prefer the dark leaves, but I think they let the flower stand out better.

As for software, I used the first version of Combine before acquiring Zerene (the first version was called CombineZM I think), but this later version CombinZP is likely much better.

Good lens - wish I had a 150 mm.

Glenn

[f8andbethere]

Glenn, thanks for the input. I agree about the framing, in the space I had, my back was up against a wall with no room to move. I guess that's the downside to a long Macro.
I'm going to a butterfly sanctuary this week, I think the length of the lens will be an advantage there.

[Nass]

Looks good to me - I can't see any obvious ghosting and the edges are nicely defined. Fake flowers are quite a good thing to practise on because they don't droop . In terms of software, I use Zerene pro but that's for more extreme magnification and this does make a difference, but at this mag and with lighting like this (ie nicely defined, value against black) you're right, software might not make that much difference. Nitty gritty 1 micron big magnification resolution mush, a different case probably

[DanK]

Dean,

It's hard to evaluate this image well because the version you posted is not large, but it looks pretty good to me.

I haven't used Combine for years -- a short time with Zerene (not pro) convinced me, and I have never looked back -- but part of your description doesn't make sense to me. In general, more images are better than fewer.

Without seeing it, it is hard to know, but the ghosting is probably halos from parallax. If you have two surfaces that are adjacent in the plane of the image but far apart in distance from the sensor, the border will be blurred. Different algorithms vary in their vulnerability to this, for reasons I don't understand. In the case of Zerene, the DMap algorithm, which is better for preserving colors and textures, is much more prone to haloing than PMax. Zerene has a retouching tool, so you can often clean up some of this by painting from an image in the stack with a clean edge, and you can also paint from a PMax composite to a DMap composite, and vice versa. However, when the problem is severe, as with some very deep flowers, I have sometimes found it impossible to get rid of the problem entirely.

Dan

[chauncey]

One of the better threads that I've read on the subject is...http://photography-on-the.net/forum/...cking+software
That said, I use Photoshop exclusively and have had quite nice results with it, with the caveat that...
some repair work is sometimes needed/there are numerous results when repeating the stack/fewer images are better if sufficient light is available for higher f/stop numbers.

You posted image was quite nice...would have been better with a larger FOV for the leaves.

[ajohnw]

This tutorial may help using Combine

http://www.wonderfulphotos.com/artic...ocus_stacking/

If the depth of field is very shallow some people move the camera rather than use the focus ring, eg Using the rather expensive Canon dedicated macro lens on insects.

There are plenty examples of what the software can produce here

http://groups.yahoo.com/neo/groups/c.../photos/albums

Combine has been very extensively used for true macro and micro work. I have seen very high quality output from all of the packages but am left with a feeling that the end results depend on technique rather than the software itself. My main interest is using it with microscope objectives - but until I can reliably get good shots with difficult subjects there isn't much point in trying to stack images.

Just checking my records the most used software for micro work is Zerene.

John
-

[ajohnw]

I can add a couple of list entries I have collected. I was asking about 32 and 64 bit linux and it seems as far as run times go there is little difference. These 2 messages cropped up

Hi -- I'm the author of Zerene Stacker and I have a fair amount of experience
with other common software. Let me take a shot at answering some of the
questions posed in this thread.

Dogs Afire asked:
> What things seem to help most when helping the image stacking programs
> create a quality output?

By far the most valuable thing you can do is to shoot clean images to start
with. By "clean" I mean low noise, well exposed, properly color balanced, and
as sharp as your optics can give but not oversharpened.

Most of the deep stacks that you see at photomacrography.net are shot as 
JPEGs.
This is because 1) JPEGs are more efficient to shoot and store, and 2) the
conditions that allow deep stacks to be shot at all also allow the lighting 
and
camera settings to be tuned so that the resulting JPEGs are high quality --
"clean" in the sense used above. The bad reputation of JPEGs is primarily due
to abuse -- too much compression, incorrect exposure or color balance 
requiring
big corrections, excessive sharpening due to camera settings, and so on. To be
sure, the very highest quality results come from shooting raw and converting 
to
16-bit TIFF for processing. But when JPEGs are clean to start with, the
difference is quite small. If you're not sure about your own equipment and
workflow, try it both ways.

By the way, there are no stacking programs that work directly with raw files,
contrary to what you may think from looking at the user interfaces. Some of
them just do the conversion to TIFF behind the scenes so you're not aware of 
it.

> It seems that I might want to modify the imput images
> to help the software.

A few sorts of modifications can be helpful.

1. Color correction may work better before than after stacking.
2. Sharpening to emphasize the detail that you care about may help the 
software
to distinguish detail from noise.
3. Preprocessing to remove dust spots will prevent them from turning into 
trails
that may be difficult to touch out.
4. Noise reduction, again if it's properly tuned to preserve the detail you 
care
about while reducing variation that you don't.

Dogs Afire suggested several ways "to indicate to the program not to use
information from certain areas". I have never seen this done effectively, and
I'm having trouble imagining a situation where it would be the method of 
choice.
Certainly most users end up telling the programs TO use information from 
certain
areas of certain frames. That's what retouching is all about. But those
decisions are best made after the program has already done most of the work.
Proactively editing the source frames in the hope that the edits will 
cooperate
with the stacking algorithm is something that I've never seen done.

> Are there specific parameters for any of the programs
> that you find particularly useful?

In Zerene Stacker, the most popular stacking method is PMax. It has no
parameters at all, and it's also the best method around for finding and
preserving even low contrast or fuzzy detail. However, PMax also has the
downsides that it tends to increase noise, increase contrast, and may alter
colors. DMap is just the opposite -- it faithfully preserves contrast, colors,
and noise, but may miss low contrast or fuzzy detail and requires tuning to 
give
the best result.

The default parameters for DMap are chosen to work well with images that are
sharp when viewed at 100% = actual pixels. Images that do not look sharp at
100% generally require larger settings for the two radius parameters. A good
strategy for setting those radii is to find a frame that contains the level of
detail you care about, reduce the display Scale until that frame appears sharp
on screen, then make the radii larger than default by the same ratio. For
example if you find that the image looks sharp when displayed at 50%, then 
make
the radii 2X bigger than default. If it has to be reduced to 25% to look 
sharp,
then make them 4X bigger. This same strategy works well in other programs that
use radii also.

> Is anyone familiar with the complex wavelet method that appears
> to be available for ImageJ?

That's one of the methods I studied before writing Zerene Stacker. The math is
different in its details, but the general approach and results are similar to
Zerene's PMax.

> One effect I notice in the stacks is a muddiness of the image
> and I would like to minimize that effect.

I'm not sure what the term "muddiness" means here. There is an effect we've
come to call "stacking mush" that results when the stacking method fails to
recognize low contrast detail and ends up using unfocused frames instead. See
http://www.photomacrography.net/forum/viewtopic.php?t=9335 for an extreme
example.

If stacking mush is what's being described, then the best fix I know is to use
Zerene's PMax method or CombineZP's Pyramoid Maximum Contrast. (The complex
wavelet method would probably give good results too, but I haven't 
specifically
tested it.) Most of the other methods are vulnerable to stacking mush,
including Zerene's DMap, both methods A and B in Helicon Focus, and the depth
map methods like Do Stack in CombineZP. (The fact that Dogs Afire mentions the
interactive threshold setting tells me that he's using DMap, which is 
vulnerable
to mush. PMax has no threshold setting.)

If "muddiness" means something else, then it would help if I could see an
example.

Alex Cummins asked:
> So my question is should you just move the focus as slightly as possible
> even if you don't see any change and then plow thru the photos.

The penalties for having more frames than you need are pretty mild, like
increased processing time and (for some methods) somewhat increased noise. The
penalties for having too few frames can be pretty severe, like focus banding
that essentially ruins the stack. When in doubt, err on the side of too many.

All of the rest of Alex's posting looks dead on target to me.

I agree with Paul Martin's observations and suggestions as well.

The depth of a focus step depends strongly on magnification. The interactive
calculator at http://www.microscopyu.com/tutorials/java/depthoffield/ is 
pretty
good. For critical work, it's a good idea to step at say 70-80% of the value
shown there.

Regarding the time required to shoot stacks, be aware that automation 
continues
to advance. See for example the data and method shown at
http://www.photomacrography.net/forum/viewtopic.php?t=11519 , using the
StackShot controller and a separate stepper motor to turn the fine focus knob 
of
a microscope. There's an example at
http://www.photomacrography.net/forum/viewtopic.php?t=11464 of shooting with a
40X objective using the StackShot rail by itself (no microscope involved).

I hope this is helpful. Sorry I couldn't figure out how to use fewer words.

--Rik

> Are there stacker programs designed for transmitted light microscopy?
> I ask from a position of complete ignorance. Jim

Picolay fits that description.

Interestingly, as I read the Picolay manual, most of its stacking methods do
incorporate the "depth map" assumption that was bothering Don earlier in this
thread.

In a prior post, I wrote that "methods with that assumption were not designed
for transmitted light microscopy."

But really I should not have ended that sentence with a period. It would have
been more accurate as "...were not designed for transmitted light microscopy 
of
subjects with overlapping structures at different depths."

I think it's important not to infer too much from the general theory of 
various
methods. Many times what matters most is the details of how they're
implemented. Even if a method is based on the depth map assumption, it can
still work very nicely with overlapping multilayered subjects in many cases.

It's also important to remember that commonly used stacking programs all 
provide
methods that are different from depth mapping. In Zerene Stacker, there's 
PMax;
in Helicon Focus, Method A; in CombineZP, either of the Weighted Average's and
Pyramoid Maximum Contrast.

While I'm here, there's one more issue I'd like to touch on: handling of
out-of-focus regions.

It's fairly common in microscopy and extreme macro photography to encounter
stacks where some features never do come into sharp focus, although they are
definitely better focused in some images than others. The colonial algae at
http://www.photomacrography.net/forum/viewtopic.php?t=11753 are a good 
example.
In general, stacks like these are handled better by one of the pyramid methods
such as PMax in ZS or Pyramoid Maximum Contrast in CZP. Methods based on
fixed-size neighborhoods are often unable to identify the best of several
out-of-focus versions, so it's common to end up with a result that is blotchy 
or
more blurred than it needs to be in the never-focused regions.

--Rik

John
-

[DanK]

John,

Thanks for posting the material from Rik, the designer of Zerene. I have never had to adjust the parameters for DMap, but it's good to know of this method for doing so. BTW, in my experience, this exemplifies one advantage of Zerene: good support.

Re this:

If the depth of field is very shallow some people move the camera rather than use the focus ring, eg Using the rather expensive Canon dedicated macro lens on insects.

Unless you are using very high levels of magnification, it doesn't matter--either method works fine. Check out my site, which has macros up to a bit over 2:1 but not higher. I have never used a rail, but at very high magnifications, moving the camera with a rail is probably more precise. In the field, people move the camera because they find it easier than changing focus, but it won't matter for the quality of images. I do both, depending on the circumstances.

[f8andbethere]

Dean,

I haven't used Combine for years -- a short time with Zerene (not pro) convinced me, and I have never looked back -- but part of your description doesn't make sense to me. In general, more images are better than fewer.

Dan

After going back an looking at the images from that string of shots, it seems that in the first and the last frame of that string the flower seemed to be bigger in relationship to the other images in the same set. Not sure how that happens, I started my focus on the front side of the flower and moved back with each new focal point. I'm going to assume thats where the ghosting came from.
Thanks for all the input guys, lots of good reading and learning ahead of me. I think photography is the most fun you can have with your pants on.

[f8andbethere]

As far as the image I posted not being very large, I brought it in via PhotoBucket. If I upload from my computer, will it come in at full size? My computer skills and knowledge are very limited, please forgive any novice 'puter questions.

Thanks again for all the help.

[Nass]

After going back an looking at the images from that string of shots, it seems that in the first and the last frame of that string the flower seemed to be bigger in relationship to the other images in the same set. Not sure how that happens, I started my focus on the front side of the flower and moved back with each new focal point. I'm going to assume thats where the ghosting came from.
Thanks for all the input guys, lots of good reading and learning ahead of me. I think photography is the most fun you can have with your pants on.

Give yourself a little googling session on the subject of telecentric lenses. Telecentricity avoids what I think you're describing

[ajohnw]

As far as the image I posted not being very large, I brought it in via PhotoBucket. If I upload from my computer, will it come in at full size? My computer skills and knowledge are very limited, please forgive any novice 'puter questions.

Thanks again for all the help.

If you use the "click here to upload photo's from......................" just click choose, navigate to it, select it and then upload now. Something will appear in tags, right click, copy and then paste the link into the post. Leave resize as default. It's probably a huge image so may be best to 1/2 size it 1st. Use the preview post to see that the upload etc has been successful. Also you may have to switch the post to advanced to get the "click to ..............." button to work,

On size of images I think that the stacking software has to take care of that and that telecentric come into play if the images are also used to create a panorama.

John
-

[f8andbethere]

Thanks for the info guys, I'm off to Google telecentric.

[HaseebM]

Very nice first attempt, looks clean.

[f8andbethere]

Very nice first attempt, looks clean.

Thanks Haseeb.