Apaflo Loc: Anchorage, Alaska

The sensor temperature will remain at ambient air temperature. How "hot" is that?

big-guy Loc: Peterborough Ontario Canada

Answer stolen from https://photo.stackexchange.com/questions/26002/what-are-causes-for-image-sensor-getting-heated-does-live-view-cause-more-heat



He talks about multiple pictures being taken in quick succession but the same applies to longer exposures as well. A photo taken at 1/60 produces minuscule sensor heat but an exposure of 60 seconds will produce a large increase in sensor heat. In turn producing more noise.

Franku Loc: Wallingford, PA and Parrish, Fl

I have read all of our advices and, in doing so, have learned much. Your explanations and comparisons are far easier to understand that reading a manual.
This afternoon I will again try to put your teachings into practice. I will also attach today's photos to yesterdays and hope that you give another outstanding critique.
Thanks again for your advice on yesterdays photos.

Franku Loc: Wallingford, PA and Parrish, Fl

I have read all of our advices and, in doing so, have learned much. Your explanations and comparisons are far easier to understand that reading a manual.
This afternoon I will again try to put your teachings into practice. I will also attach today's photos to yesterdays and hope that you give another outstanding critique.
Thanks again for your advice on yesterdays photos.

Apaflo Loc: Anchorage, Alaska

If that were true... taking multiple 30 second exposures one right after the other would have a rather dramatic effect in terms of heat and as a result in terms of noise. It doesn't.

In fact, for each camera at any given ambient air temperature it is possible to take and save a "dark frame" exposure that can be subtracted from any long exposure image at the same ambient temperature and the same shutter speed. Hence such a dark frame image taken today at 68F and 30 seconds can be used next week or next year for long exposure noise reduction of images taken with the same camera at 68F and 30 seconds. It's true at 100F, and it's true at -40F. It can be very productive to build a library of dark frames for a given camera!

That is a well known technique used by many because it literally cuts in half the time needed to produce a series of such images using dark field subtraction for noise reduction. (The camera will do it, but it requires two 30 second exposure per image rather than one.)

The critical point for this discussion is that the same stored dark frame image can be used for the 100th out of 100 images taken sequentially. That's right, all the "heat build up" (which is actually very near zero and is not heat from the sensor) from the first 30 seconds of sensor operation is almost exactly the same as the heat build up from 3,000 seconds of sensor operation.

The reason that is true is because the heat production is very small relative to the heat dissipation from the sensor. There simply is no build up, and the sensor stays at the same temperature as the ambient air temperature.

For many uses the saved dark frame image that would be used is best made from averaging multiple dark frame exposures. Some times, for critical work, the multiple dark frames are taken sequentially with other exposures, one every so many multiples.

Do research using Google on dark frame subtraction for long exposure noise reduction to verify all of the above.

Meives Loc: FORT LAUDERDALE

[quote=Franku]I looked at the third picture of man at pool. By the way, after you post you can go back and label each picture, 1,2,3 or by name. I can not address grain, but I would have used different settings. ISO 100 was good with film, but now you can use,400, 800, 1600 with no grain. Also rule of shutter speed is 1/mm lens. You were right on the edge with 28mm and 1/30 shutter speed. Third with the ISO set up you can shut the aperture down from f 4 to f11 and give wider DOF depth of field. David

jimmya Loc: Phoenix

What camera are you shooting with? The cheaper the camera the more noise problems you'll have at high ISO settings. The first shot simply doesn't have enough light so the camera perhaps compensated with high ISO.

Franku Loc: Wallingford, PA and Parrish, Fl

The camera is a Nikon D7100.
I was going to put more pix up today but things happened to prevent the event. We are flying back to PA tomorrow evening but I hope to put up some similar photos with recommended setting before noon.

Franku Loc: Wallingford, PA and Parrish, Fl

[quote=Meives]
Good advice, David. I will follow it!

big-guy Loc: Peterborough Ontario Canada

So, if what you say about heat build up is true and the sensor stays at the current ambient temperature which causes no noise then WHY do you need the dark frame? The dark frame is used to combat the noise added because of sensor heat build up. The longer the sensor is active the more heat it generates and in turn causes noise to appear in some pixels.

If you disagree with this then explain why all electronic devices in use generate heat and do not stay at the ambient temperature.

On a 30 second exposure the sensor generates heat and that induces noise in some pixels. The dark frame method merely exposes which pixels are effected and allows for those pixels to be neutralized. Putting a band aid on a cut does not mean there is no cut.

I am not arguing your eloquent description of the use of dark frames, just your premise and conclusion. The two don't jive and it doesn't make any sense. Maybe this is a, "I meant this but said that" scenario.

Apaflo Loc: Anchorage, Alaska

That is what you say. It isn't true. There is noise, and it is greater with higher temperatures, but the sensor temperature is related to the ambient temperature. Sensor temperature does not increase with the length of exposure time, but the noise does. There is no sensor heat build up. The ambient temperature inside the camera may rise due to the on board CPU, but that is also rather minor.

Read the URL's listed below, and you'll find that hot pixel noise doubles for twice as long an exposure. That is because it is collecting stray electrons for twice as long. To generate twice as many stray electrons a temperature increase of "six to ten degrees C will result in the dark noise in the resulting image doubling".

The point is that if heat build up were causing the noise increases seen in long exposures the camera would literally get too hot to hold!


I discussed how that is not true, and demonstrated proof, in a previous article. The dark frame used for noise reduction can be the same frame used on the first or the one hundredth consecutive 30 second exposure. If what you are saying were true the 100th exposure would be significantly different, and more noisy, than the first. It isn't.


Really? Some of them get colder!

Again, this was already addressed in a previous article. The heat dissipation from the sensor is greater than the heat generation. Simple as that. The very small amount of heat generated never has a chance to actually raise the temperature of the sensor.

You are also assuming the sensor generates heat anytime it is collecting a charge. That is not true. The tiny amount of heat is does generate happens only when the sensor charge is read. Hence a 30 second exposure means lots of time to let the sensor cool after that last exposure, while a 1/2000 second sequence take at 10 frames per second would generate significantly more heat with very little time to cool between read operations.


That is clearly not true.


I suggested previously that you do some research. What I am saying is not just an opinion. It is a factual accounting of how it works. It is well researched and well documented in many places.

Here is a URL that will at least get you off to a good start.

http://photo.net/learn/dark_noise/

Here is another URL, that might just boggle almost anyone's mind! :-)

http://blog.kasson.com/?p=14229

big-guy Loc: Peterborough Ontario Canada

Thanks for the link and here is just a line from said link:
As the temperature of the sensor goes up, dark noise increases. Different sensors behave differently, but in general, increasing the temperature of a sensor by six to ten degrees C will result in the dark noise in the resulting image doubling.



This link deals specifically with the Nikon D5 and can't be used when dealing with generalities:

This is part of a series of posts about the Nikon D5

There are many sources of noise and if dealing with specific origins, your comments are correct but I am talking about noise in general, particularly thermal, and even your links state what I'm saying along with others. From what I can gather you seem to be discussing shot noise. You can do some further research here if you like: http://www.rmsp.com/blog/2014/04/11/whats-noise-part-2/ where it states,


Maybe we're both right and wrong at the same time when viewed in context. Take the above or leave it, I will continue to research this and many other aspects of photography and as I have discovered, more often than not, things can change over time and it's not a good thing to become locked into any one aspect.

canon Lee

Keep the ISO as low as possible. open the aperture. use a fill flash. Position your subject, if possible, in good natural sun light ( this is tricky and takes time to master. Shoot in Aperture priority and monitor your histogram to make quick exposure adjustments. Less light means grainy images. Artistically speaking a wide aperture creates a more pleasing portrait in that the subject ,not the background, is most important and needs to be pin sharp.

Apaflo Loc: Anchorage, Alaska

That statement is true, but it does not support anything you have said! It does not say that the sensor adds heat or that there is a heat build up over the length of a long exposure. It says at higher ambient temperatures there is more noise.

Hence if there is x amount of noise at 0C, and noise doubles every 10 degrees C, there will be 2x at 10C, 4x at 20C, 8x at 30C, and so on.

The same URL also says that noise doubles as the exposure time doubles. If the noise is significant at 1 second, and doubles at that rate up to 32 seconds... your claim is that the temperature of the sensor went from ambient to at least (at 6C per 2x noise) 36C higher, and maybe as much as 60C higher if the rate is 10C per 2x noise.

The camera would be somewhere between 133F and 176F after a 30 second exposure! It isn't.

Read the URL again, for detail and not just to look for support for you theory. See what is says actually happens, and specifically why the noise doubles with twice the exposure length (even though the temperature does not rise, and no heat is generated).


Two points to take from that URL. One is that it talks about theory, and can help to understand how the camera works. Second, it very clearly states that a D5 does not generate heat during long exposures.


I am not discussing shot noise. Shot noise follows a Poisson distribution and in photography is generally the result of photons arriving at an irregular rate. It has nothing to do with heat.

The links that I've posted do not support what you have said. They also state that the hot pixel noise seen in long exposures is not random. It is a pattern noise, related to stray electrons released by thermal activity into the sensor site. (The release rate in any single sensor site is random, but will vary from one site to another in a non-random manner. Hence some pixels become a "hot pixel" and others do not.)

Your cited URL makes grossly incorrect statements. Like you, the author has "found it on the Internet", and does not actually understand the physics involved.

Guessing that the sensor must heat up when it's being charged is just that: a WAG. Note that the charge claimed to be heating the sensor comes from photons captured... But the way to remove hot pixel noise is to make an exposure where no photons are captured and subtract the noise in that image from the desired image. The hot pixel noise is the same when the sensor collects no photons, so we have to ask why the sensor isn't cooler due to less charge and thus have less noise, which would make dark frame subtraction totally ineffective! But is works rather well, because charging the sensor with photons does not create significant heat.

In fact the only time a sensor is significantly active electrically is when its data is being read out at the end of the exposure. That will create virtually the same noise for long short or long exposures. It's called "read noise". It is a thermal noise and it has a random distribution.


The one single issue between us is the idea that the sensor contributes significant heat, and over the length of a long exposure causes the temperature to rise. It does not.

big-guy Loc: Peterborough Ontario Canada

My apologies for spouting incorrect information. After much searching and getting conflicting data I called Canon tech support. After listening to some gawd awful on hold music for 10 minutes I finally got a definitive answer. Your statement below, including the word "significant" is correct.

From the horses mouth, A long exposure will add heat to the sensor but nothing appreciable and will not damage the sensor or camera and while it may cause a little noise, that noise will be minimal in relation to all the other noise gathered by the long exposure.

Over heating of the camera can be a problem when using live view in a continuous fashion but who in their right mind would do this? Turn it on for extreme focusing and then turn it off is the appropriate method.

With this newly found data I plan on doing some extra long exposures this weekend. Thank you for pushing me to learn yet another new trick.