IS IT TOO HOT TO PHOTOGRAPH OUTSIDE TODAY? If your region does not pass the wet-bulb test, it’s also too hot to exist inside without air conditioning. This is actually really a test of whether global climate change has rendered your locale dangerous to live in. Ultimately this doesn’t belong on this site, but I made a thing and I want a place to put it.
Eager to see if this new code works!
This is based off Roland Stull’s 2011 wet bulb approximation equation, which looks like:
Where T is temperature in Celsius, RH is relative humidity in percent, and Tw is the wet bulb temperature in celsius. For more information on this equation, and why it’s important, see: https://open.library.ubc.ca/media/stream/pdf/52383/1.0041967/1
Exposure Values are a somewhat important concept, kind of, if you are trying to maintain consistent exposure while fiddling with your camera in manual mode. Realistically, by shooting in aperture or shutter priority, with your ISO locked down, your camera will maintain the proper exposure value for you anyway.
A photographer might need to prepare for any values between -6 and 16, and the difference of 1 exposure value is called a “stop”. Remember, though “stopping down” generally refers to using a narrower aperture, you can also decrease your shutter speed or ISO by one stop!
The actual math of this does not matter too much to a normal photographer. Or an abnormal photographer. It’s somewhat fascinating to think about this if you are designing experiments or writing software that does crazy camera things though.
Honestly, if you’re doing crazy math-based camera experiments, you probably didn’t need to read this. However, I’m upset that the equations for the exposure triangle (which dictate the relationship between iso/shutter speed/f-number) are not readily available without you solving for these elements yourself, so I figured I’d commit them to one place on the internet. If you need these, look no further.
Note many resources will give versions of these equations using EV_100, or Exposure Value at ISO100. No standards body has codified use of EV_100, and while some websites claim that “Light Value” refers to non-ISO100 EV (and sometimes the opposite, EV_100), this is inconsistent with contradicting definitions around the internet. Forget about light value, for now it’s not a useful term, it’s the math photography equivalent of “nonplussed”. Just use these equations, and if you want ISO100 or EV_100 plug in 100 for the ISO.
If these all feel like they are yielding answers that are just a littttle off- welp, congratulations, you’ve learned the dirty secret of photography. The fnumbers are very, very slightly rounded. Below is mathematically sound, where:
F is the relative aperture (f-number)
t is the exposure time (“shutter speed”) in seconds
S is the ISO arithmetic speed
EV is the exposure value
ISO
S = \frac{100 \times F^2}{t \times 2^{EV}}
Python
iso =(100*(fnumber **2))/(shutter_speed *(2** exposure_value))
While I was waiting for burlesque bunny models to prepare for an upcoming shoot, a Danish traveler wandered into our set. She was goodly enough to pose so I could try out my brand new lens, which I threw on the nearest camera, my modified infrared Nikon. As has been demonstrated in past explorations, this is not the sharpest sensor, as there is no guarantee that a critical focus plane will even exist if your aperture is too open.
To my surprise, the new Daguerrotype Achromat lens, a recreation of the Petzlav design from 1839, worked perfectly with this modification. Apparently simplistic configurations are sharp as a tack in infrared, even while wide open!
As she was captured, straight out of camera:
Madeleine in Near Infrared + Infrared (590nm-~1100nm)
Of course, this is barely even a photograph. Let’s adjust this, and make it black and white:
This is a more presentable version of Madeleine
Great!
But what if we want color?
As we all know, infrared pictures look really weird in traditional “false color”, which is how modern digital photographers often present their infrared work. Obviously, it’s impossible to present an invisible spectrum without modifying it to become visible, so SOME color has to be given to it. Further articles on traditional false coloring will come, but a cursory Googling will provide a litany of tutorials.
As an example, let’s look at the lovely Madeleine in this original infrared photo, after it’s been white balanced to pull the highlights into a blue range. She looks weird. And she’s not a weird looking girl:
White balancing the infrared to something more… “presentable”
We swap the Red and Blue channel, as is the standard jumping off point. Ummm… Great? Now she looks weird AND sickly. We have done no good here.
This might be the traditional end-point of false coloring infrared photographs, but obviously this can’t be used on a human subject.
But perhaps there is another way? I’ve been poking at machine learning for some separate projects, so maybe we can use an artificial intelligence to fill in some gaps for us.
I’ve been interested in the work of Richard Zhang and the work he did with Phillip Isola and Alexei A. Efros on using artificial intelligence to colorize things. It seems a solid use of “Deep Learning” to train an AI on a large set of images, then have it color an image for you. You can read more about here, if you want! http://richzhang.github.io/colorization/
Let’s pop the first black and white image from above into Zhang, Isola, and Efros’ automatic artificial-intelligence colorizer, a demo of which he provides freely at https://demos.algorithmia.com/colorize-photos/
AI, in the form of Algorithmia’s “Colorize-It”. Interesting, if not good. This does give us the distinct impression of… kind of color.
Interesting. Kind of.
Ultimately we see that AI recognizes that there is, in fact, a girl, and the AI’s model (think of this as the memories the AI has retained from past studies) recognizes foliage pretty well.
The only problem is, I’m not interested in foliage turning green, or the model’s eyes turning brown. I don’t much care for the colors bleeding out of the edges, either. I came to make a candy-colored infrared nightmare, and make a candy-colored infrared nightmare I shall!
Instead, I did more reading, and found this paper, a further exploration that involved Zhang, Isola and Efros. https://richzhang.github.io/ideepcolor/ By “steering” an AI, pointing to where certain colors should be, perhaps we can make a more interesting picture?
I followed some simple directions to run this code on a Mac, and within a few minutes of getting the program started, I had colorized a version I liked! No training or reading instructions was necessary. It was largely very intuitive.
Using iDeepColor is fairly intuitive.
To use iDeepColor, simply mouse-over a spot on the image on the left, click, wait a second, and see what the AI recommends. Chances are pretty good many of the points are going to already be a reasonable value, so simply click a point of color you like, and when the palette is recommended, do nothing. Click to a point where you’d like to change the color, then choose a color in the a b color gamut. I believe this to be a* b* channels, from the 1976 CIE L*a*b* space, but we’ll explore this later. You may have to get finicky, adding a few points around an area you’ve just defined, to keep the color from bleeding out.
Honestly, despite the frustration of not having a “zoom” ability, or an “undo” function, this was so intuitive there seems to be little left for me to explain.
After about 20 minutes futzing about, I was left with this:
iDeepColor after about 20 minutes. Now this is exactly what I was looking for.
Is it perfect? No. Is it novel? Absolutely! In fact, it gives me some ideas for colorizing more competently in Photoshop for future endeavors.
I hope this inspires some creative steps with colorization and AI in hyperspectral photography!
After reading my post on how to recreate the look of tintype and wet-plate collodion digitally, you may have been left wanting the code to appropriately demosaic/debayer.
This is a photography blog, not a coding blog, so please bear with me for my first go at making this accessible.
To recap from my earlier posts:
In looking at an image captured through a B25 glass, you will only capture 1 out of every 4 pixels correctly, thanks to the debayer color filter array (CFA) on top of your digital camera’s sensor. That means 3 pixels are “dead” or, at least “incorrect”. Here’s a quick script to throw them out.
As we can see in this close-up of the raw data, this image has a lot of dead pixels. Let’s get rid of ’em.
I know I will get some feedback that this is not a debayer, or demosaic, in any traditional sense. You are technically right, haters. This is simply a downsample to strip out the three dead pixels.
You will quarter the resolution of your image with this process. I’m sorry, but it has to happen. The resulting clarity will play nicely with any upscaling processes you then wish to engage upon, however, and, especially in portraiture, artificial intelligence can work wonders.
I originally wrote this in MATLAB, but instead let me give it to you with freeware.
Software Requirements
You will need the following command-line tools. These are free, but if you aren’t comfortable using a command line interface (CLI) then I have bad news for you- this is unfortunately how you’re going to have to do it.
ImageMagick – This is THE tool for altering an image from the command line. It exists for all platforms, so whatever computer you are running, you can use it.
DCRaw – This is THE tool for opening raw files from a digital camera. RawTherapee is a front-end for this tool. Unfortunately, you still will have to install it, even if you have RawTherapee, to use the command-line interface.
I recommend, if you are on a Mac/Linux, simply using homebrew to grab these- open up a terminal window, and type
homebrew install imagemagick
and then
homebrew install dcraw
On PC, this might be a bit more complex, but these tools exist for all platforms, which is why I chose to code in this fashion.
Assuming these are installed correctly, you should be able to run this single line of code to demosaic down to the blue sensor pixel.
The Code
Open a terminal window, navigate to where your RAW file lives, and type:
(-c) Output in a fashion that can be handed off to ImageMagick (STDOUT for piping)
(-D) Monochromatic, untouched output
(-T) Tiff
(-6) 16 bit
(-g 2.4 12.92) gamma 2.4 with a toe-slope of 12.92, which is, more simply put, the appropriate setting for sRGB.
(-o 1) Set the colorspace to sRGB with a d65 white point. You can tinker with this, but ultimately it shouldn’t make too big of a difference, since we are compressing to a monochromatic pipeline here regardless.
Note that some tutorials will tell you to use a -4. This will output a linear electro-optical transfer function, and unless you really know what you’re doing, you probably don’t want this. I would argue you really don’t want this.
This then pipes ( | )the data to ImageMagick- or rather the utility it provides named “convert”:
(sample 50%) Downsample at 2:1, implied that you are dropping every other pixel in both X and Y directions
(-roll +1+1) Offset your image by the requisite number of pixels to place a blue pixel at the top left of the image. Assuming RGGB, this is +1+1. If you have BGGR, then this will be +0+0. I recommend trying them both. Heck, you can try +0+1 and +1+0 to see what would typically be the two green pixels. The one that looks the brightest is your correct image.
Close-up on the result: This looks dim and, dare I say, stupid. It looks like we’re peering at the data captured by the red pixel of the color filter array, which is the “leakiest” of the three color filters, accepting some blue light.
Close-up on the result: Looks identical to the last one- this means that, as anticipated, we are looking at one of the two G pixels on the color filter array.
Closeup of the result: Bingo! This is the blue pixel!
Not surprisingly, OutputTest_D matches the blue debayer pixel for my Nikon.
Note this is for a traditional bayer filter. You might wish to review the description of color filter arrays, and see if there is a better fit for you and your sensor. I don’t have the foggiest of how this might be accomplished with an x-Trans color array filter with free command-line tools, unfortunately.
