Problems with rendering details

A forum to ask questions, post setups, and generally discuss anything having to do with photomacrography and photomicroscopy.

Moderators: Pau, rjlittlefield, ChrisR, Chris S.

AlxndrBrg
Posts: 74
Joined: Tue Jan 14, 2014 12:17 pm

Problems with rendering details

Post by AlxndrBrg »

I have some annoying troubles to get details and shapes to render as "readable" in pictures of bee genitalia (it's for work, don't think I sit around photographing bee genitals for fun, no no, I only do that with cuckoo wasps :oops: ). I've tried different lighting, and experimented with different stacking techniques in Zerene, but the improvements are minimal. It doesn't help that the cuticle is amber coloured and semitransparent of course.
The viewing angle has to be like it is, otherwise I would skew it so the important details did not overlap with the structures in the background.

One solution could be to go higher mag, but I don't have a 20x Mitty, although I really should make them order one for me.

The obvious, but less satisfactory solution is to have a go at it in Photoshop, and simply paint in some definition, but that takes time and I have to do it for more than one species (props to anyone who can recognise the species/genus!). It would be good to find a way to solve the problem "pre-post", I'm sure more problems like these will pop up.

Or do you consider the image to be perfectly readable as it is? Sometimes I doubt how well the shapes come across, I can't really tell how someone without experience of these things view the image - I already know how they look in 3D since work with them under a stereo microscope to clean and mount them.

Specs:
  • - Canon EOS 6D, shooting at 1/20 sec, iso100, fullsize .jpg
    - 10x Mitutoyo M Plan Apo 10x/0.28
    - ITL200 reversed as tube lens, at the "correct" distance from objective and sensor, giving ~10x mag
    - White printer paper in a cylinder around the Mitty as a diffuser, I should probably try without a whole cylinder, as it can sometimes diffuse the light too much, lowering contrast
    - lit with 2 IKEA Jansjö LED lamps
Here is a quick and dirty edit of how the end result should look like, with better background removal etc etc
Image

and here's how the shape should be "read", with the red part laying infront of the blue part
Image

to give a sense of how cropped the image is compared to a full sensor 3:2 shot (5472x3648 pixels)
Image

the two following are unedited 1024x1024 pixel "100% crops" (that for some reason in my Chrome browser is shown at 133% size?) of the output TIF from Zerene, showing the most troublesome area (line). There's lack of contrast, hairs that are rendered as transparent, lots of artifacts around the hairs and so on. And these pictures are as good as I could make it using editing of Dmap + Pmax, using sub stacks etc. With Dmap alone the area is a mess with blurry patches due to the hairs, and in just Pmax the details of the tip blends in with the details of the underlying structure. I especially find hairs that are at an oblique angle, towards or away from the viewer, to be almost impossible to come out well after the stacking. I have an iris (not in use when I took the pictures) behind the Mitty, perhaps stopping it down a tad could help?
Image


Image

Any help or suggestions are most welcome!
Last edited by AlxndrBrg on Tue May 12, 2020 11:37 am, edited 2 times in total.

AlxndrBrg
Posts: 74
Joined: Tue Jan 14, 2014 12:17 pm

Post by AlxndrBrg »

and here's a dorsal view, showing how well separated the two structures really are, which does not at all come across in the lateral view:
Image

rjlittlefield
Site Admin
Posts: 21121
Joined: Tue Aug 01, 2006 8:34 am
Location: Richland, Washington State, USA
Contact:

Post by rjlittlefield »

I would have no clue about the 3D structure, just from looking at one of the full-stack images.

When I took a course in Scientific Illustration, many years ago in college, something like this was presented as an example of a subject that might be better drawn than photographed.

The problem was that the foreground and background structures had very similar color, texture, etc, so that when photographed together they did not separate for the viewer.

Of course photography has advanced a lot, but I think you will be in trouble as long as you stick to the idea of presenting a single photographic image and hoping that the viewer will make sense of it.

So I have a few suggestions of things that might help.

1. Do a Stack Selected of just the foreground structure, and present that as a companion to the full stacked result.

2. Present as stereo pair, straight out of PMax with no retouching.

3. If the presentation medium will allow animations, then present as a rocking sequence.

--Rik

rjlittlefield
Site Admin
Posts: 21121
Joined: Tue Aug 01, 2006 8:34 am
Location: Richland, Washington State, USA
Contact:

Re: Problems with rendering details

Post by rjlittlefield »

AlxndrBrg wrote:unedited 1024x1024 pixel "100% crops" (that for some reason in my Chrome browser is shown at 133% size?)
For me they display at 175% (1792 pixels). The reason is my Windows display scaling, which I have set because of using small 4K monitors.

--Rik

iconoclastica
Posts: 325
Joined: Sat Jun 25, 2016 12:34 pm
Location: Wageningen, Gelderland

Post by iconoclastica »

One thing you might try is combining different lightings. For example, the marginal transparent hairs are hard to combine with the white background. In such cases (and immobile subjects), I shoot a second stack with the foreground lighting switched off. This results in a strong silhouette wich is perfect for separating foreground and background with a mask.

This of course will not help with the hairs within the outline of the apparatus. But perhaps a hard skimming light would give you the desired details. In general, I am not a fan of super soft light (BIG lamps plus transparent gobo). Sometimes even an unshielded flash is still too big (soft). But then, I do not normally have shiny ceratinous subjects, so that might not work for you...

BTW, There's more detail in your photo than you show here.
--- felix filicis ---

Peter M. Macdonald
Posts: 180
Joined: Tue Jan 20, 2009 2:59 pm
Location: Berwickshire, Scotland

Post by Peter M. Macdonald »

Looking at some areas of your pictures, the lighting is still a little on the harsh side. I appreciate that this may be a symptom of your problem, rather than a cause, but it may be worth exploring.

Try to diffuse the lighting even more than has already been done. You can always put a little more contrast into the final output with a simple "S" curve in Photoshop.

Also, the transparent and near transparent hairs have edges which look very white, i.e. they are rendering very bright. Try to experiment with slight under exposure. It would be interesting to see a set of pictures which have been underexposed by one half stop and one whole stop. Again, the same curve in Photoshop can be used to put a little more punch into the rendered output.

Good luck.

Lou Jost
Posts: 4638
Joined: Fri Sep 04, 2015 7:03 am
Location: Ecuador
Contact:

Post by Lou Jost »

I have a similar problem for some micro-orchid photos. Some of these orchids have fake insect genitalia to fool insects into mating with the flower for pollination. These are taxonomically important for identifying orchid species, just like the real genitalia are for the insects. But in orchid specimens preserved in alcohol, the structures are translucent and colorless, so our false genitalia are even harder to photograph than the real genitalia.

My solution is to make a rocking synthetic stereo GIF in Zerene and convince the journal to publish this in an online appendix. These are easy to read in 3-D.

AlxndrBrg
Posts: 74
Joined: Tue Jan 14, 2014 12:17 pm

Post by AlxndrBrg »

Thanks for everyone's input!

It gave me the idea that I can maybe try and shoot the same image with different lighting and exposure, without moving the subject, and then make a composite in PS to visually "distance" the different parts. Will be labour intensive though.
Looking at some areas of your pictures, the lighting is still a little on the harsh side. I appreciate that this may be a symptom of your problem, rather than a cause, but it may be worth exploring.

Try to diffuse the lighting even more than has already been done.
Thanks, that didn't really occur to me, it's worth trying actually.
When I took a course in Scientific Illustration, many years ago in college, something like this was presented as an example of a subject that might be better drawn than photographed.
You might be right, thought about it too, might just be that I have consider myself defeated and break out the ol' Wacom tablet. At least I will have good photographic material to line-trace over heh!

The images will be published as a single .PSD from an integrated system, so sadly I cannot use any of the digital techniques like rocking gif or stereograms. And in the end probably as a tree carcass version as well.

AlxndrBrg
Posts: 74
Joined: Tue Jan 14, 2014 12:17 pm

Post by AlxndrBrg »

I would have no clue about the 3D structure, just from looking at one of the full-stack images.
Hmm, but would you have an idea of the general shape of the structures from just the lateral picture? In other words, would you have understood the shapes in the very first image as being how they are shown in the 2nd (red/blue) image? The overall 3D structure is not so important, as long as the separation and shape of the blue/red parts is readable.

rjlittlefield
Site Admin
Posts: 21121
Joined: Tue Aug 01, 2006 8:34 am
Location: Richland, Washington State, USA
Contact:

Post by rjlittlefield »

AlxndrBrg wrote:
I would have no clue about the 3D structure, just from looking at one of the full-stack images.
Hmm, but would you have an idea of the general shape of the structures from just the lateral picture? In other words, would you have understood the shapes in the very first image as being how they are shown in the 2nd (red/blue) image?
Not with much confidence. Maybe it would be OK, if my task were to match a physical specimen against a set of images like you show. Depends on how much variation between species. But I would be more comfortable with something like a line drawing or a tinted photo.

--Rik

Lou Jost
Posts: 4638
Joined: Fri Sep 04, 2015 7:03 am
Location: Ecuador
Contact:

Post by Lou Jost »

Another idea would be to use the depth map generated by Zerene DMap. and color it using false colors thayt are a function of elevation. For example a blue (far) to red (near) gradient like those used in GIS programs to make elevation maps. Those are instantly interpretable at a glance.

I've been thinking about using a depth map to create false shadows as well, using GIS programs. Suppose you have a flatly lit image made using epi lighting. There are no shadows. But GIS programs let you do"hill shading" on top of the actual colors. You choose the location of the light source and the software places shadows on thw depth map. I have been thinking about using this tecnhique for some images of leaf surface structure I have been working on.

There is a lot that could be done by marrying stacking software and geographical information software. I believe it would be possible to make stunning photos this way. Further, they could be viewed directly in the GIS application as three-dimensional objects.

Probably there are 3-d rendering programs that already do this.

Troels
Posts: 516
Joined: Mon Feb 15, 2016 11:06 am
Location: Denmark, Engesvang
Contact:

Post by Troels »

If it should be fast and easy I would make two stacks: One where only the nearest part was sharp, another with the distant part sharp. I don't think it is possible to express depth separation in one picture when everything is sharp.
Troels Holm, biologist (retired), environmentalist, amateur photographer.
Visit my Flickr albums

iconoclastica
Posts: 325
Joined: Sat Jun 25, 2016 12:34 pm
Location: Wageningen, Gelderland

Post by iconoclastica »

--- felix filicis ---

MarkSturtevant
Posts: 949
Joined: Sat Nov 21, 2015 6:52 pm
Location: Michigan, U.S.A.
Contact:

Post by MarkSturtevant »

I am agreeing more with rjlittlefield, especially on the idea of using photographs to support making drawings, but then where the drawings, with stippling and so on, are the real goal. Drawings are an excellent teacher of anatomy.
I don't know what is done these days in publications on this sort of thing, but maybe a photograph can also be supplied as a separate panel or made available thru a url link. So the photograph does not have to be perfect.
Mark Sturtevant
Dept. of Still Waters

Lou Jost
Posts: 4638
Joined: Fri Sep 04, 2015 7:03 am
Location: Ecuador
Contact:

Post by Lou Jost »

In my field, botany, drawings are actually preferred over photos in scientific publications, for the reasons we've been discussing here.

Post Reply Previous topicNext topic