Hair of the dog

[boomblurt]

Inspired by Beatsy's recent "Knot [in] the hair of my chinny chin chin", I tried to do a hair image. It took quite a few attempts to get anything worth sharing - getting decent light at 50x is, to me at least, a black magic that still bamboozles, despite all the excellent information here.

The model was my dog, a Blue Heeler named Jack, who has both dark and light coloured hair. The white hair had a mostly colourless cuticle (Edit: cortex is actually the clear bit here and cuticle is the scaly outer) so I'm thinking that it's the same effect that gives his polar bear cousins their colour.

Mitutoyo 50x, Raynox 150, Canon 6D, flash.

[Saul]

Very interesting - looks like I can see inner and outer layers ?

[rjlittlefield]

I agree it looks like there are some sort of layers. Perhaps not inner/outer, but at least front/back.

Any chance we can see this in stereo?

--Rik

[boomblurt]

Google informs me that we are seeing three layers there - the cuticle is the outer, scaly layer, while the innermost part is called the medulla. The clear middle portion is the cortex and it normally contains the pigment.

Rik, I don't think I'd be able to stereo this - the retouching was too fiddly.

[rjlittlefield]

You might try it without any retouching and see how it looks. I often present a retouched single image, and then a stereo pair with no retouching. Of course the pair will have artifacts, but if the added insight from the stereo is enough, I just live with them.

--Rik

[boomblurt]

The stereo was done at +/- 8 but doesn't really help to show the internal structure, to my eyes at least.

To the naked eye the hair is pure white, but under the microscope the clear cortex becomes more apparent as magnification is increased.

[rjlittlefield]

Thank you for the stereo. I have pulled your pair into StereoPhoto Maker and examined it with magnification, in parallel view using my best viewer (https://www.berezin.com/3d/pocket_3dvu.htm).

The stereo does make clear that the white portion is internal, not back surface, which is a nice bit of insight.

Other than that, I have to agree that in this case the stereo does not add much. I think this is a tribute to your illumination, which makes the general 3D structure clear even in a single view. Nicely done!

--Rik

[rjlittlefield]

By the way, I'm having some trouble believing that the stereo was done with +-8, unless you've done some significant post-processing.

With +-8, some subject features should move a lot to the right, others to the left (in each view).

But in the pair that you've presented, it seems like all features have moved in the same direction although by different amounts. Does this make sense, and can you explain?

--Rik

[boomblurt]

I just redid the stereos to confirm that it is +/- 8 ... initially I had done it using smaller values and was surprised to see not much effect. Retouching was cloning out a couple of dust trails, plus levels and a little sharpening.

I've aligned (by positioning) the two images in Photoshop and there is a rotation there, but it's not massive. (Edit: So the right image's right/rear side moves to the left, while the left/front of the right image moves to the right.) Could slabbing be the cause? The stack was 20 slabs, with about 200 images in total.

Edit: Also I don't think that the hair and knot are the same distance from the lens. I suspect (but can't recall for sure, as I did a few versions) that the knot is furthest away, with the view from above resembling a ^ with the knot at the apex.

Here's a gif of the stereo images :

[boomblurt]

Aaahhh ... (/inspiration after a couple of Vodkas) ... Zerene produced a stereo with the knot off centre (to the left in the left pic and vice versa) - perhaps due to the +/- 8? Most subjects I've done wouldn't be possible at +/-8, but here, with no hairs etc to give depth cues that's what I ended up using. Then I cropped in stereophotomaker and this aspect remained. So if you compare the knot, as cropped, it is all moving in one direction.

When the images are repositioned so the knot is in basically the same place, as in the gif, the rotation around the knot is evident. Could this explain it?

[ChrisR]

These look to be flattened, once air-filled tubes, like polar bear guard hairs?

found a ref:
http://azadocents.org/wordpress/?p=211

[rjlittlefield]

Could slabbing be the cause? The stack was 20 slabs, with about 200 images in total.

Bingo!

So, I'll bet that what you did was to add the slabs to the end of the regular source files, set stereo for +-8, then Stack Selected of just the slabs, the last 20 of the total 200.

When ZS processes Stack Selected with stereo turned on, the shifts are interpreted as applying to all the input files as a single stack. That's what allows Stack Selected outputs to serve as retouching inputs for a full stack.

But in the scenario I've outlined, it means that the shifts for the 10% of images that actually got stacked corresponded to a rotational component that was only 10% as large as you expected, plus a constant offset that accounts for the other 90% of the nominal shift. Effectively you got a stereo rotation corresponding to only +-0.8 instead of +-8.0.

A better way to do the stereo would be to either
(1) open a separate project containing just the slab outputs and Stack All those, or
(2) Stack Selected the original 180 images instead of the 20 slabs.

In either case, appropriate shifts then would be in a range of maybe +-2 to +-3, giving a stereo separation that is roughly 2.2 to 3.3 times larger than you have now. Remember that you can use arbitrary shifts, like 2.5, if that looks better than either 2 or 3.

--Rik

[boomblurt]

Interesting link, Chris. It is a guard hair, and I only mentioned polar bears as I seemed to recall from somewhere that their "white" hairs are actually clear.

And thanks Rik, that makes sense. I usually use that sort of shift so it was confusing, and I'll know in future.

[Beatsy]

Nice one boomblurt - your efforts paid off handsomely. I find lighting tricky at 50x too (although my hair knot had the mitty on a 135mm tube lens, so only 33.75x on the sensor - but still tricky nonetheless).

Interesting to compare the differences. I see this hair also has a flattened cross section like my beard hair - was it fairly coarse?

Perhaps we're at the start of a series here...

[boomblurt]

Thanks, Beatsy

The hair was almost as thick as a beard hair, but the cuticle scales were larger. Untied, the shape didn't seem to be too far off cylindrical (I had another look after Chris' post) - maybe the tension in the knot increases any flattening present?

And I'd wager that Chris is right and the hair is hollow, just from the way it deforms; seems lots of animals have hollow hair including some dogs.