3 ms·

The author's rushed approximation of "-infinity dB" is disingenuous, and discounts the efforts of people who handle low-level signals. For example, his claim m

by nayuki 9y ago

The author's rushed approximation of "-infinity dB" is disingenuous, and discounts the efforts of people who handle low-level signals. For example, his claim makes it impossible to compare the ambient light level of an eclipse versus the light level from a full moon or new moon. For example, not knowing the noise floor will make it impossible to tell whether a 16-bit audio recording, or 20-bit, or 24-bit is the best choice. What is the "larger source of error" in using a decibel scale for brightness? Maybe the fact that human eyes are not perfectly logarithmic? But such a scale is very appropriate for calculating camera exposures.

dahart 9y ago
Aww, come on, don't double-down, just let it be. If you want some benefit of the doubt, feel free to give some benefit of the doubt. We can nit pick each other to death, but that isn't going to be productive or fun at all. You can't take something out of context and claim that it's insulting people that he wasn't talking about. He wasn't discounting anyone. Yours is a bigger and more intentional error than his statement that totality is 0% energy. You're picking a fight instead of acknowledging that totality really is approximately 0% solar power for the purpose of figuring out how much heat you're going to feel on your arm during the eclipse, compared to before the eclipse. He admitted the noise floor is not really zero. He wasn't contradicting himself, he was clarifying for the sake of the nit pickers. You're using that against him instead of accepting that both of his statements can be true at the same time. He was never claiming absolute zero with infinite precision, that is completely clear from the article. He also wasn't asking what bit depth to use for audio recording. That would be weird, wouldn't it? Because he's talking about the sun's heat and not the sound it makes. He wasn't talking about photographing it either, or even measuring the heat precisely. He was showing why it gets cold before it gets apparently dark. You can do that in even less than 16 bits. Yes, human eyes are not perfectly logarithmic. Interesting that you don't care for the sake of camera exposures in the middle of a conversation about the sun and moon, where you just cited bit depths and people who handle low-level signals. The logarithmic approximation isn't particularly good, which the link in the article discusses in detail. Log is good enough for this article, but not for vision scientists. The logarithmic brightness approximation really truly breaks down when you look at the sun or the night sky, human vision does not give an approximately logarithmic response to either the sun or the night sky, not even close. Camera film and camera sensors also break down and begin to clamp at either end of a moderate log range that is narrower than human vision. Photographing the corona well is really hard due to the extreme dynamic range and the way that camera response stops being logarithmic at the ends.
- ars 9y ago
  The person you are replying to is not wrong though. The author of the article is. Even during totality it's not as dark as night. It's around the illumination of just after sunset, and he could have definitely plotted that - it's not that hard to lookup how much there is then, he doesn't have to measure it himself.
  - dahart 9y ago
    Ugh. Be very precise and tell me, what does it matter? What about the story would change if he'd used super extremely precise numbers? How would the conclusion change, exactly? Does the correct number for solar power during totality explain why it gets cold before it gets dark? The article was not wrong, it was an approximation, which he stated clearly. Did you perhaps not understand this sentence: "Anyway, there is never really "zero" radiant energy to detect. In reality there's always some kind of background noise, maybe at -90 dB or so."? If he had used the correct number, you would see that it is approximately 0% power. Again, and I'll state it as fact now rather than rhetorical question, the article was accurate to more than 3 significant figures. BTW, the discrepancy between night time levels and the darkness of totality is due in part to atmospheric scattering during the eclipse, so you're introducing a red herring. > he could have definitely plotted that Given that the heat plot goes from 0% to 100%, and that the brightness plot goes from -70dB to 0dB, will you estimate for me how many pixels in both plots would be different if he used really precise numbers for solar power during totality? My current estimate is 0 pixels would change across both plots. Do you think the number is higher? If so, why?

by nayuki 9y ago

dahart 9y ago

Aww, come on, don't double-down, just let it be. If you want some benefit of the doubt, feel free to give some benefit of the doubt. We can nit pick each other to death, but that isn't going to be productive or fun at all. You can't take something out of context and claim that it's insulting people that he wasn't talking about. He wasn't discounting anyone. Yours is a bigger and more intentional error than his statement that totality is 0% energy. You're picking a fight instead of acknowledging that totality really is approximately 0% solar power for the purpose of figuring out how much heat you're going to feel on your arm during the eclipse, compared to before the eclipse. He admitted the noise floor is not really zero. He wasn't contradicting himself, he was clarifying for the sake of the nit pickers. You're using that against him instead of accepting that both of his statements can be true at the same time. He was never claiming absolute zero with infinite precision, that is completely clear from the article. He also wasn't asking what bit depth to use for audio recording. That would be weird, wouldn't it? Because he's talking about the sun's heat and not the sound it makes. He wasn't talking about photographing it either, or even measuring the heat precisely. He was showing why it gets cold before it gets apparently dark. You can do that in even less than 16 bits. Yes, human eyes are not perfectly logarithmic. Interesting that you don't care for the sake of camera exposures in the middle of a conversation about the sun and moon, where you just cited bit depths and people who handle low-level signals. The logarithmic approximation isn't particularly good, which the link in the article discusses in detail. Log is good enough for this article, but not for vision scientists. The logarithmic brightness approximation really truly breaks down when you look at the sun or the night sky, human vision does not give an approximately logarithmic response to either the sun or the night sky, not even close. Camera film and camera sensors also break down and begin to clamp at either end of a moderate log range that is narrower than human vision. Photographing the corona well is really hard due to the extreme dynamic range and the way that camera response stops being logarithmic at the ends.

ars 9y ago
The person you are replying to is not wrong though. The author of the article is. Even during totality it's not as dark as night. It's around the illumination of just after sunset, and he could have definitely plotted that - it's not that hard to lookup how much there is then, he doesn't have to measure it himself.
- dahart 9y ago
  Ugh. Be very precise and tell me, what does it matter? What about the story would change if he'd used super extremely precise numbers? How would the conclusion change, exactly? Does the correct number for solar power during totality explain why it gets cold before it gets dark? The article was not wrong, it was an approximation, which he stated clearly. Did you perhaps not understand this sentence: "Anyway, there is never really "zero" radiant energy to detect. In reality there's always some kind of background noise, maybe at -90 dB or so."? If he had used the correct number, you would see that it is approximately 0% power. Again, and I'll state it as fact now rather than rhetorical question, the article was accurate to more than 3 significant figures. BTW, the discrepancy between night time levels and the darkness of totality is due in part to atmospheric scattering during the eclipse, so you're introducing a red herring. > he could have definitely plotted that Given that the heat plot goes from 0% to 100%, and that the brightness plot goes from -70dB to 0dB, will you estimate for me how many pixels in both plots would be different if he used really precise numbers for solar power during totality? My current estimate is 0 pixels would change across both plots. Do you think the number is higher? If so, why?