4 ms·

Wet Plate Photography Makes Tattoos Disappear

32 points by DanBC 6y ago

amluto 6y ago
I wonder if the spectral response of modern cameras is such that a suitable linear combination of R, G, and B would reproduce the effect.
- chmod775 6y ago
  If you're compressing an intensity curve on a spectrum to three samples (RGB), you're always going to lose information, regardless of what your original sensor was capable of. Basically someone is throwing balls with floating point numbers at you and you have to sort them into three buckets and a trash bin, then you only report the number of balls in each bucket. For all I know that photographic technique doesn't even capture any part of the spectrum that falls within the RGB area. I don't know much about these photographic techniques and their terminology, so I had trouble looking information up, but it appears that the technique captures blue and ultraviolet.
  - amluto 6y ago
    I have no doubt that wet plate photography does not closely resemble any part of our eye’s response, and Bayer filter designers put a decent amount of effort into getting (a nondegerate linear combination of) our cones’ response. And the potential or UV and IR sensitivity is a good point. It could nonetheless be interesting to try to learn a 3-vector (r, g, b) such that the actual linear-light output from some real camera dotted with the vector is insensitive to tattoos.
- dTal 6y ago
  In color? No. In monochrome? Mayyybe... According to Wikipedia, the wet-collodion process was sensitive only to blue and ultraviolet light. The tattoos reflect mainly blue, whereas skin tone has very little blue; this has the effect of lightening the tattoo and darkening the skin, so it's hard to distinguish them in monochrome. But there's likely no wavelength of "red" you can select where the skin reflects the same amount as the blue tattoo - the tattoo will always look darker. Or to put it another way - there's no stage lighting you could come up with that will make blue tattoos look skin colored, let alone while also making skin look skin colored. The best you could do is to use the blue channel only, or even light the scene only with the deepest blue light the camera can pick up (and convert the photo to monochrome). Generally speaking you can't make things look very different to cameras than the human eye, as they have similar frequency response. Moreover, you can't distort the appearance of objects very much by picking specific narrow frequencies to light them with, or objects would look incredibly strange under narrowband LED lighting (they don't look perfect, but it's not a heavy distortion). Spectral response of objects isn't totally flat, but it's usually not very "spiky" either, so RGB captures a pretty good approximation of the spectral response in most situations. Interestingly, this question has direct application in the field of holography, where the scene must be lit by RGB laser. You only get 3 wavelengths to work with (and you don't have a huge amount of latitude as to which 3) and the entire rest of the spectrum is discarded, and yet full color holograms usually look impressively accurate. [edit] I've played around in GIMP a bit, and you can indeed remove most of the tattoo simply by taking the blue channel. You can further reduce it by shifting the response even more towards blue by subtracting the red and green, albeit with increasing amounts of color banding and other aliasing. I'm pretty sure you could decently approximate this effect entirely in software, with the correct coefficients.
simonblack 6y ago
It took a lot of research to invent panchromatic film. Up until about 1940, most films had problems in being sensitive to the whole range of colors. Most films on sale were 'orthochromatic' which signified that they weren't sensitive to red, and so you could develop them in a darkroom with a red 'safelight'. It also meant that red showed up as black on the final photo. When I first started playing with photography in the mid 1950s, I killed several films while developing them because the books I was learning from assumed you were using 'orthochromatic' films but I was using the newer 'panchromatic' films. I soon learned that I needed to develop my films in total darkness, but I could still use a safelight when printing those black and white negatives, because the printing paper didn't need to be panchromatic seeing that it was only working with shades of grey.

32 points by DanBC 6y ago

I wonder if the spectral response of modern cameras is such that a suitable linear combination of R, G, and B would reproduce the effect.

chmod775 6y ago
If you're compressing an intensity curve on a spectrum to three samples (RGB), you're always going to lose information, regardless of what your original sensor was capable of. Basically someone is throwing balls with floating point numbers at you and you have to sort them into three buckets and a trash bin, then you only report the number of balls in each bucket. For all I know that photographic technique doesn't even capture any part of the spectrum that falls within the RGB area. I don't know much about these photographic techniques and their terminology, so I had trouble looking information up, but it appears that the technique captures blue and ultraviolet.
- amluto 6y ago
  I have no doubt that wet plate photography does not closely resemble any part of our eye’s response, and Bayer filter designers put a decent amount of effort into getting (a nondegerate linear combination of) our cones’ response. And the potential or UV and IR sensitivity is a good point. It could nonetheless be interesting to try to learn a 3-vector (r, g, b) such that the actual linear-light output from some real camera dotted with the vector is insensitive to tattoos.
dTal 6y ago
In color? No. In monochrome? Mayyybe... According to Wikipedia, the wet-collodion process was sensitive only to blue and ultraviolet light. The tattoos reflect mainly blue, whereas skin tone has very little blue; this has the effect of lightening the tattoo and darkening the skin, so it's hard to distinguish them in monochrome. But there's likely no wavelength of "red" you can select where the skin reflects the same amount as the blue tattoo - the tattoo will always look darker. Or to put it another way - there's no stage lighting you could come up with that will make blue tattoos look skin colored, let alone while also making skin look skin colored. The best you could do is to use the blue channel only, or even light the scene only with the deepest blue light the camera can pick up (and convert the photo to monochrome). Generally speaking you can't make things look very different to cameras than the human eye, as they have similar frequency response. Moreover, you can't distort the appearance of objects very much by picking specific narrow frequencies to light them with, or objects would look incredibly strange under narrowband LED lighting (they don't look perfect, but it's not a heavy distortion). Spectral response of objects isn't totally flat, but it's usually not very "spiky" either, so RGB captures a pretty good approximation of the spectral response in most situations. Interestingly, this question has direct application in the field of holography, where the scene must be lit by RGB laser. You only get 3 wavelengths to work with (and you don't have a huge amount of latitude as to which 3) and the entire rest of the spectrum is discarded, and yet full color holograms usually look impressively accurate. [edit] I've played around in GIMP a bit, and you can indeed remove most of the tattoo simply by taking the blue channel. You can further reduce it by shifting the response even more towards blue by subtracting the red and green, albeit with increasing amounts of color banding and other aliasing. I'm pretty sure you could decently approximate this effect entirely in software, with the correct coefficients.

simonblack 6y ago

It took a lot of research to invent panchromatic film. Up until about 1940, most films had problems in being sensitive to the whole range of colors. Most films on sale were 'orthochromatic' which signified that they weren't sensitive to red, and so you could develop them in a darkroom with a red 'safelight'. It also meant that red showed up as black on the final photo. When I first started playing with photography in the mid 1950s, I killed several films while developing them because the books I was learning from assumed you were using 'orthochromatic' films but I was using the newer 'panchromatic' films. I soon learned that I needed to develop my films in total darkness, but I could still use a safelight when printing those black and white negatives, because the printing paper didn't need to be panchromatic seeing that it was only working with shades of grey.