11 replies to this topic

[UlfW]

Here are a few more images with simulated Aerochrome:

First an image with yellow tail-lights, (on behalf of Cadmium. ;-) and then an image with more red elements and straight lines to show the effect of the typical lens distortion of a full-frame fisheye lens.


Then an image to test the flare behaviour of the lens with a rear mounted filter with the sun in the image.
The only red elements in the motive are the four flowers at bottom right, here correctly yellow after the AIR-action.
There are a lot of interesting details in the original image as the lens is surprisingly sharp.


The images are my first attempt to use one of my fisheye lenses for IR.
I finally got a suitable filter that was small enough to fit at the rear end of the lens.

Equipment:
Camera: Canon Eos 60D full spectrum modified
Lens: Canon EF 8-15mm at 10mm. (To get full frame fisheye on an APS sensor)
Filter: Rear mounted B+W 040 (OG550), mounted with some greytack-putty over the gelatin-filter holder

Processing:
PhotoNinja: white balance and exposure adjustments
Photoshop: Tiffen12 Action - AIR simuation, NeatImage - noise reduction

[Andrea B.]

Thanks, Ulf. We all do enjoy this Aerochrome look.

Fisheye images are always fun to make. I think that yellow tail light on the car would be also very interesting in a much closer-up composition because of its geometry?

[OlDoinyo]

To me, the salmon-pink tone of the foliage says that there is likely quite a bit of IR in the green and blue channels of these images. Otherwise, an interesting perspective.

[Cadmium]

UlfW, These are very nice, I especially like the top two which include the yellow. Excellent examples of visual red transposed to visual yellow, as it should be.

As far as the exact cutoff point of the longpass filter used for the 'Tiffen12 Action - AIR simulation', the classic filter used for that is the Wratten #12 (AKA: "minus blue" filter), and/or thus the Tiffen #12 filter.
However, I have often used longpass filters that are higher in nm with very pleasing results, in some cases more pleasing than with a #12 filter.

As far as adjustments within the 'Tiffen 12 Action', I sometimes adjust the contrast lower, and also raise the hue (in the last part of the process) to +20 which adjusts the red more to my liking.
It all depends on what you like, but I find those two adjustments can often be useful.

It is good to experiment with a variety of these longpass filters to see how differently they result in the process, and which looks best usually.
Here is an overlay graph of Hoya, Tiffen, and Schott longpass filters in this range.


NOTE: The yellow curves are Tiffen filters (labeled on the graph), the black curves are Schott filters (labeled above the graph), and the others (red, blue, green, pink) are Hoya filter curves (labeled above the graph).

Edited by Cadmium, 06 October 2017 - 00:21.

[Andrea B.]

So I wonder how Tiffen manages to "soften" the shoulder of their cut-in curve? Coatings perhaps?

[Cadmium]

I don't think so, the Tiffen filters I have don't look like they have any coatings. Just their version of absorptive filter glass.
The Schott and Hoya longpass filters look the most sharp, of these examples, but keep in mind that these are just graphs that manufacturers publish, which may not exactly always be true or up to date.
However, Schott provides the most inclusive filter data of any filter glass manufacturer, full range, no drop outs or missing ranges, and 1nm resolution data.
Next best is Hoya, but their data lacks resolution (every 10nm) and has huge range drop outs, such as their U filters which will have missing data for most of the visible range.
Tiffen has only linear graphs, no data, and no way to view their data diabatically.
All of these filters work for this, in one way or the other, but hypothetically if I were picking out a filter to cut off blue, I would probably tend to choose a filter that had a sharper cut off slope.

By the way, all of those graphs are T not Ti (internal transmission), and linear, not diabatic.

Edited by Cadmium, 07 October 2017 - 07:55.

[Andrea B.]

And the difference would be what? Between T and Ti.
(I always forget this for some reason. Will go look up.)

[Andy Perrin]

T includes the partial reflections. Ti does not.

[UlfW]

Andy's answer is correct but quite condensed, possibly a bit difficult to interpret.

I would phrase it as:
T is the combination of the filter material transmission and the losses caused by reflections in the surfaces of the filter.
This is how the real world filter affect the light when you use it.

Ti is the pure filter material transmission.

The surface losses are the result of the difference between the refractive indexes in air and filter material.
The losses are in the same range and roughly around 5% per surface passage.

The "i" in Ti could mean internal.
Ti is always > than T

Do you agree Andy, or have I misunderstood anything?

I hope this clarified things if Andys answer wasn't clear for everybody at the first glance.

[Cadmium]

Here is an example showing the difference between Transmittance (T) and Internal Transmittance (Ti) using the Schott OG515, OG530, OG550, and OG570 filters at 2mm thick.
Transmittance (T) is the real world transmission with no coatings.

Edited by Cadmium, 07 October 2017 - 07:33.

[UlfW]

Andrea,
Thanks, I have always liked working in the wide end of the focal length range.
I like the thee-dimensional feeling you sometimes can get from such lenses.
I bought the FE-zoom four years ago, but soon thereafter my photography inspiration faded away.
Indeed that taillight would be quite interesting as an abstract motive. I'll keep such possibilities in mind in the future.

OlDoinyo,
I adjusted some parameters to taste the way Cadmium mention in his comment just below yours.
Especially the last one needed a lot of work due to the very high initial contrast.
The Tiffen12 action tend to generate a very high contrast too.

Sorry that I didn't have time enough to write all the details as I am not very efficient in writing text. I'm a very slow writer.

[Andy Perrin]

UlfW, that was perfect. I was typing on my iPhone and it's kind of painful to post here on an iPhone! I wanted to minimize the character count...

Specifically, for light entering perpendicular to the filter from the air, if the refractive index of the uncoated glass is n, then the fraction of light reflected off ONE surface (e.g. the front of the filter) is
R1 = (n - 1)^2/(n+1)^2
Typical glass is about n=1.54, so this gives a single-surface reflected fraction ("reflection coefficient") of
R1 = (1.54-1)^2/(1.54+1)^2 = 0.045 = 4.5%
which is pretty close to what UlfW said.

When you have reflections off both the front and back of the filter, the expression is more complicated. Surface roughness and internal scattering from bubbles and such also affects the real world answer. Also this assumes no coatings of any type.

Edited by Andy Perrin, 07 October 2017 - 21:32.