We are bioluminescent

[dabateman]

I only just recently read this interesting study. We do seem to emit photons from our heads. Just at 1000x less intense than our eyes can see.

 

I wonder if Andy moves to a pitch black room and sets his full spectrum sony A7s, if he could could get a portrait with the longest ISO and 30 seconds exposure?

 

 

https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0006256

[Andy Perrin]

David- from the parameters they give in the paper, no way would I be able to get anything. They took 20 minute exposures! With a cooled camera at -120°C! Even my TriWave only gets down to -80°C (I'm not sure how the sensitivity compares to the A7S, partly because the exposure is not set in the same way as for an ordinary camera on the TriWave so it's hard to match settings.).

[dabateman]

The two peaks were at 630–670 nm and 520–580 nm, respectively.

The exposure time was 20 minutes. But I think they were at base gain for the sensor. Allowing cooling to control the noise and optimize for highest quantum efficiency. They reported 75%.

 

They stated that dark current is 0.65 electron/pixel/h and readout noise in the slow scanning mode is less than 4.5 electron rms. Whats crazy though is they used a 7 element lens with NA 0.5. That might be the tricky part.

 

The Sony A7s has around 60% quantum efficiency. And if you boost the ISO to a tolerable noise level in an air conditioned room, you might be able to see something over the noise. But might need a high NA lens.

Whats the maximum exposure duration of the bulb mode? Can you push minutes?

A test shot at high ISO might work.

[Andy Perrin]

You boost the noise when you raise the ISO, and the signal is probably the same size as the noise. I really highly doubt that I can get anything distinguishable from the noise. Also, when I leave the A7S on for a long time then the amp glow becomes quite discernable in parts of the image (although center is usually okay). I can expose up to 30 sec for sure, and longer in bulb mode but that will trigger the auto noise suppression.

 

There's also the problem of finding any place that's dark enough. My apartment leaks light for sure. Just shutting the shades would leave me more than enough light for the A7S to see by under those settings without any bioluminescence at all.

[Stefano]

You can take multiple 30 seconds exposures and then stack/average them. I did a similar thing with my solar panel experiment, do you remember? If your noise is random (not all noise is random) you can average it out. Still, the intensity is very low. We should emit about 3 photons/minute as blackbody radiation (in visible light), if one of then enters your camera you will see our thermal emission.

 

The 3 photons/minute thing is a rough calculation from this Wikipedia article: https://www.google.com/url?sa=t&source=web&rct=j&url=https://en.m.wikipedia.org/wiki/Black-body_radiation%23:~:text%3DAn%2520object%2520that%2520absorbs%2520all,is%2520called%2520black%252Dbody%2520radiation.&ved=2ahUKEwieztG4pK7qAhXGlIsKHR2aAk8QFjAOegQIDBAi&usg=AOvVaw2QCi7uSI3HG3g-Ow5PgYsa&cshid=1593685558151

[Andy Perrin]

You can average random noise as you said, but the obstacle will probably be the non-random noise (especially the line noise, and amp glow, which dominates when you have practically no signal). I really don't think my good-but-consumer camera is going to magically become a (very inefficient) thermal camera no matter how much I average. You need something like one of those cooled astro cameras.

[Bill De Jager]

You can take multiple 30 seconds exposures and then stack/average them. I did a similar thing with my solar panel experiment, do you remember? If your noise is random (not all noise is random) you can average it out. Still, the intensity is very low. We should emit about 3 photons/minute as blackbody radiation (in visible light), if one of then enters your camera you will see our thermal emission.

 

The 3 photons/minute thing is a rough calculation from this Wikipedia article: https://www.google.c...d=1593685558151

 

Those (averaged) three photons per minute will be going in all directions, not just towards the camera lens. What percentage of the surface area of the sphere surrounding a person is subtended by the entrance pupil of a camera lens? Certainly far less than one part per one thousand.

 

One photon won't necessarily be read by the sensor. If it is read, then how do you know it's from the subject as opposed to a stray black-body photon from some other object? Even if you knew, one photon would mean that at most one pixel responds.

 

Averaging out the noise doesn't mean that all pixels have exactly the same brightness from noise. It just means that the percent variation between pixels is reduced substantially, making the noise less severe. The signal-to-noise ratio is greatly improved but there will still be some noise. If you're starting from a position of high noise and very close to no signal then that's a very tall hill to climb to try to get a discernible signal.

 

To sum up, one pixel or twenty, each showing the very dim signal created by one photon per pixel, would still be drowned out by averaged noise. You wouldn't see an image of a person.

[Stefano]

Yes, the probability that one photon from a person has the right trajectory to end up in the lens and strike a pixel on the sensor is very low, close to zero. The blackbody radiation idea was just a random idea I had, since to see the bioluminescence from people you have to have a very sensitive system. Maybe one little dot on the image can actually be thermal radiation, but of course it would be indistinguishable from noise.

 

If you could take a huge number of exposures, then something could emerge. But this is practically impossible.

[Andy Perrin]

If you could take a huge number of exposures, then something could emerge. But this is practically impossible.

Are we still talking about my camera? Because the answer there is a definite "no" even if you had thousands of exposures. You can't average away the line noise, and the line noise will be larger than the signal.