Starry Sky

By the negative effect of the 'starry sky', photographers usually mean a collection of hot and broken pixels, which resembles a real sky with billions of stars. The photo below shows a 1: 1 crop for a very long shutter speedon which such an effect is strongly manifested.

Starry sky created by Canon EOS Digital 350D camera at ISO 1600 and shutter speed of about 1 hour.

Canon EOS Digital Camera Starry Sky 350D at ISO 1600 and shutter speed about 1 hour.

I want to warn you right away that this article only deals with hot pixels. Unlike 'dead' or 'dead' pixels, their number and intensity of luminescence increase with increasing ISO and / or with increasing excerpts. At low ISO and / or relatively short shutter speeds (up to 30 seconds), hot pixels may not manifest themselves in any way and are a kind of time bomb. Dead / dead pixels are always present in the pictures, with any settings.

Hot pixels make a huge contribution to the overall digital image noise. The difference between hot pixels and normal digital noise is that hot pixels usually appear only at slow shutter speeds and even at ISO 50 or ISO 100. Usually hot pixels are lonely, but with very slow shutter speeds, their number increases, as shown in the examples from this articles.

Personally, I rarely shoot at very long exposures, and therefore I do not encounter the problem of hot pixels. But here are photographers who use very long excerpts, for example, in astrophotography, more than once faced with the fact that even at the lowest ISO values ​​appear hot pixels.

Below are examples of shots taken at very slow exposures in 'BULB' mode (manual excerpts) During exposure the camera mount was closed with a conventional bayonet cover. Unfortunately, the Canon EOS Digital 350D и Canon EOS 5D not quite correctly record the duration excerpts in BULB mode, but I can assure you that Exposition spent about one hour. So very long excerpts were used only to maximally demonstrate the strong effect of the starry sky, in real life they are rarely used.

Link to the archive with the originals. Pictures were taken in RAW + JPEG mode with different settings for noise reduction functions.

The examples of photographs show that the matrixes of cameras suffer not only from the effect of the 'starry sky', but also have serious 'flare' at the corners and sides of the matrix. Due to the fact that matrices Nikon D80, Canon EOS 5D and Canon EOS Digital 350D have the usual Bayer structure, the resulting hot pixels correspond to one of the shades of red, green or blue, since usually one of the subpixels a certain shade.

Despite the rather terrible picture of the "starry sky" of my cameras, I do not have any serious problem for real everyday shooting. I am pretty easy on hot pixels. My experience tells me that the number of hot pixels can increase over time and should not be seriously feared. When shooting at fast shutter speeds, the occurrence of hot pixels is extremely rare.

Hot pixels can be treated fairly easily by enabling 'long exposure noise reduction'. This function removes hot pixels with one trick - after the end exposure the main shot, the camera exposes the additional shot with the shutter closed, equal in time to exposure main frame. In the second, additional frame, which is not written to the memory card, the camera's processor finds hot pixels and 'subtracts' them from the main picture. The function works very well and even with such an apocalyptic sky, as shown in my examples, it copes with a bang, including removing side flares from the matrices. But she has one serious problem - the shooting time is doubled. And if the photographer used a shutter speed of two hours, then in the end he would have to wait another two hours for the camera to take an additional corrective shot.

Some cameras have modifications of this function for which the camera simply memorizes the 'star map' once and subtracts it from all subsequent images. This hot pixel map can be updated via the camera menu. Functions based on the creation of such 'maps' are usually called a general term - 'remaping' (from the English 'Remapping'). If you shoot in RAW format and process the filmed material on a computer with modern powerful tools for working with RAW files, then the remap function is turned on automatically. Modern processing programs are quite good at detecting hot pixels and removing them.

Two parts of the same picture with and without remapping. Auto Remapping with Adobe Camera Raw

Two parts of the same photo, with and without remapping. Automatic remapping with Adobe Camera Raw. This snapshot can be found in the archive with originals under the name 'IMG_1265.CR2'

Also, hot pixels are partially removed by the 'High ISO Noise Reduction' function. This function can detect a hot pixel as a normal part of digital noise and programmatically remove it.

I will describe one interesting observation of the noise reduction function at high ISO sensitivity using the example of a camera Nikon D80... In the menu, this function is called 'High ISO NW' and four operating modes: 'Norm Normal', 'Low Low', 'High High', 'Off'. And if you click on the help button, you can see an interesting explanation of how this function works:

Process pictures taken at sensitivities above ISO 400 equivalent to reduce noise (ripple). The capacity of the memory buffer decreases as noise is removed from pictures. Note that when 'Off' is selected for High ISO NN, noise removal is still performed at sensitivities above ISO 800, although the amount of noise removed is reduced.

In short, at ISO 800 and above, the camera always uses noise reduction. As a result, you won't be able to get a 'naked' JPEG file at a high ISO value. The most interesting thing is that due to forced noise suppression, sometimes there is a situation when the noise at ISO 640 is higher than at ISO 800. In the gallery from this article there are two pictures with Nikon D80 at ISO 640 and at ISO HI-1 (equivalent to ISO 3200), forced noise reduction made the picture at ISO 3200 cleaner.

Many years ago, I saw some cameras tested for noise. In those tests, after passing the average ISO values ​​on which the camera could work, the graphs showed a jump with a decrease in noise, and then again its further growth. Then no one could solve the riddle of such matrix behavior. But the solution most likely lies precisely in the fact that when shooting in JPEG after a certain ISO value, automatic forced noise reduction is turned on, which makes the picture cleaner.

By the way, forced noise reduction does not only apply to Nikon D80, most Nikon cameras always produce noise reduction at maximum ISO values. Each camera has its own separate threshold, from which forced noise reduction is turned on. As for cameras of other brands, I have no exact information.

Also, many cameras have an ISO threshold to which noise reduction will not be carried out, even if the noise reduction function at high ISO is turned on manually. For example, here's what you can find in the camera menu Nikon D700 about the 'Under. noise for high. ISO ':

Reduces noise at ISO 1600 or higher (also reduces the number of pictures that can be taken in a single burst). Even when 'Off' is selected, minimum noise reduction will be performed at ISO sensitivities HI 0.3 or higher.

It turns out that Nikon D700 noise processing starts at ISO 1600, while Nikon D80 at ISO 400.

And one more observation about the Canon EOS 5D - in the pictures it is noticeable that hot pixels in the 'starry sky' accumulate in special constellations and even the 'Keno Milky Way :)' is visible in the center of the picture.

Milky Way?

Milky Way? This photo is enhanced with contrast. The rest of the photos are in the article without any additional processing. The original of this snapshot is archived under the name 'IMG_5374.CR2'

Hot pixels do appear even at the lowest ISO values, you just need to make the shutter speed longer (see the file 'IMG_5372.JPG' from the archive with the sources).

Noise reduction is turned off in the archive file has the name IMG_5377.CR2 / JPEG

Noise reduction is off. The file in the archive is named 'IMG_5377.CR2 / JPEG'

In real shooting conditions, all these hot pixels will simply be superimposed on the resulting image :(.

Feel free to experiment with your digital cameras. To do this, simply cover the camera bayonet, or the front lens cap, find 'Bulb' or 'B' or 'Time' mode and shoot at really long exposures and different noise reduction settings. and at different ISO values. The 'Bulb' mode can most often be selected in mode 'M' after setting the slowest shutter speed (30 or 60 seconds).

I propose in the comments to discuss the following questions, to which I do not have a clear answer:

  1. Does the number of hot pixels increase with the age of the camera, i.e. Is there an effect of 'matrix burnout' from its constant work?
  2. What determines the appearance of new hot pixels?
  3. What matrices - CCD, CMOS, FAVEON, LBCAST, etc. more resistant to hot pixels? How secure is film against this digital ailment?
  4. What causes the 'flare' at the corners and edges of the image, which is described in the article?
  5. What kind of vertical stripes appear in pictures with Canon EOS 5D?
  6. Do some Canon, Sony, Pentax cameras force noise reduction at high ISO if the noise reduction function itself was turned off?
  7. How to clearly distinguish digital noise and hot pixels at very long exposures? Or is digital noise at long exposures composed entirely of hot pixels of varying degrees of 'hotness'?

↓↓↓ Like it :) ↓↓↓ Thank you for your attention. Arkady Shapoval.

Add a comment:



Comments: 119, on the topic: Starry sky

  • Denis

    A little story from my practice may be useful ... I wanted to take an interesting long exposure photo on my Nikon D80, but not stars, but a very ordinary subject. With the help of a cable, I set the shooting time to hour. However, when I returned to the place of shooting earlier, I found that the picture had already been taken, i.e. having not fully worked out the time I needed, I cannot say for sure what influenced this, but with a probability of 97% we can say that the remaining battery charge is low, I also thought that this could be a function of protecting the matrix from overheating. Naturally, the picture was underexposed, instead of 1 hour - 25 minutes (1505,4 seconds) and this is what came of it Most likely, the shooting was carried out without JVI (it seems to have regretted it), since there was a flare in the upper corners of the frame. Later, I still decided to see what can be obtained from this image, and, increasing the exposure in the Republic of Lithuania by 10 stops and adjusting the BB, I got the following . So this question can be useful not only when shooting astrophotography. Thanks to Arkady for the article and readers for comments.

    • Denis

      The room had a temperature of about 10 degrees C + - 5 degrees.

  • anonym

    So this amplifier heats up - from this and the characteristic flare, Canon's amplifier turns off, unlike Nikon, this explains that astrophotographers are either canon or special cameras.

  • Dmitriy

    1. "How much" -> how much
    2. "What vertical stripes appear on the Canon EOS 5D shots?" - banding
    3. "Do some Canon, Sony, Pentax cameras force noise reduction at high ISO values ​​when the noise reduction function itself has been turned off?" - produce. even in RAW. from a certain threshold. complete information is not yet available for all manufacturers. here, for example, by Sony

    • Arkady Shapoval


  • Alexey

    I own several years Nikon D600.
    Recently noticed a red hot pixel near the center.
    How can I remove it forever?
    Is this due to the fact that lately I shot a lot of videos, because the matrix works constantly and heats up?

    • Jury

      carry the camera to a service or search the network for service software and clean it yourself

    • Valery A.

      I'm sorry, I was a little late. I also had a burning red pixel near the center at d600, and at short shutter speeds and at ISO 100, it can be clearly seen. I began to look for solutions on the Internet (I thought, in the form of a thread of the program), I came across a simple advice to the Nikon Club - to forcefully clear the matrix (my cleaning function was disabled). I cleaned it, lo and behold - it was lost, it was so retouched that it could not be found. I looked through the manual - there is nothing like that in the section about cleaning, why?

      • Nicholas

        and it is possible to clarify whether the internal function of cleaning the matrix is ​​meant? or straight clean with a mop? also found a hot pixel in the camera yesterday. filmed a comet at night.

  • Sergei

    Hello Arkady!
    I have a Nikon D700! I decided to check it for broken, hot pixels. Found 2 hot, but it's not scary! Found another! When approaching and enlarging this black photograph, it is black, but not uniformly black, it is all in squares and rectangles of different shades, some kind of light black, gray, light gray, dark gray, randomly arranged, and it gives the impression as if the entire matrix is ​​chopped and pokotsan, in general it is not clear! And with every black photograph, these squares and rectangles change their locations! Although compared with other black photographs of other cameras, for some reason everything else is uniformly black on others, no matter how you bring it closer! Arkady, what could it be? Something with the matrix, or should it be? Explain please! Thank you in advance!!! I tried to insert this photo, but not inserted!

  • Alexander

    I looked. Everyday, it makes no sense to shoot for more than 30 seconds. At ISO100 / 30sec - if you do not look pixel by pixel, it is not noticeable. At ISO6400 / 30sec - the starry sky within a large metropolis.
    How I solved the problem - I just don't bother, which is what I wish everyone)

  • Sergei

    About “crazy alpha particles”.
    The range of alpha particles in air is a few centimeters.
    To influence alpha radiation on the matrix, an alpha radiation source should be placed in front of the matrix and all glass in front of the matrix should be removed.

  • Edward

    He conducted several of his own experiments on this topic. I got unexpected (for myself) conclusions. Ready to share.
    First, the first unexpected conclusion: the viewer program is of fundamental importance for viewing a picture with hot pixels. I will show in the pictures from the computer screen.
    The experiment carried out as follows. Pictures were taken with the lens cap closed.
    The first shot at high light sensitivity (800) with a shutter speed of 15 s. This is how it looks in XnView 2.39 2015. Like a continuous noise. Same picture
    in the program ViewNX 2 2.8.1 2013 looks quite decent. 400% increase. Defective pixels are visible.
    So, in order not to be intimidated by the "starry sky", you need to choose a viewing program.

    About the second experiment and the second conclusion - a little later.

    • Valentine

      VieNX looks at the NEF file well because it is a native Nikon program that understands the algorithm of its raw file. The XnView program is a third-party software viewer, which is not sharpened for raw in principle, but can view them. Nikon's proprietary raw viewing algorithms have not been passed on to XnView and other viewers, so they basically reproduce raw raw poorly. The exception is the jpeg embedded in the raw file for previewing - they play it normally, like all other jpegs.

      • Edward

        It is obvious. Another thing is more interesting: it is possible that Nikon's program masks the "starry sky", and XnView shows it as it is.

        • Valentine

          Nikon's program, when interpreting a raw RAW file, uses its algorithms and many parameters that form the final image on the screen. XnView interprets the file as it can (in practice, very primitive), so it often displays with multi-colored pixels even those files in which there is no 'starry sky' effect. Nikon's program (camera firmware or converter core) must remove all kinds of extraneous defects that reduce image quality. If you like to apply the word masking to this interpretation of the file, let it be masking. Note that such a negative effect as noise is not always removed by default, because eliminating it can both improve image quality (in terms of image clarity) and worsen it (in terms of reducing detail). The manufacturer already partially transfers control of the parameters of such effects to the user.

  • Edward

    Second experiment. Dependence of defective pixels on whether the viewfinder is open or closed.

    Shooting with the lens cap closed indoors under home lighting. Pictures were taken on two devices: D40X and D610. In all cases, ISO 800, aperture fully open, shutter speed 15 sec. From the point of view of the effect on the pictures of an open or closed viewfinder, I note that the experiment took place indoors under home lighting. Defective pixels in the images were counted using the 1.0 program DeadPixelTest 2001.

    The full comparison table is here.

    The results of the experiment. On one device (new), when the JVI was open, three “hot” pixels were found, with the closed one, four. On the second device (old), 99 “hot” pixels were found with open JVI and 111 with closed pixels. On repeated images on the same device, the coordinates of the hot pixels remain the same, which indicates the repeatability of the experiment and the randomness of the results. I can’t explain yet why there are more “hot” pixels when the JVI is closed, than when it is open.

    Output. Preliminary conclusion: the number and brightness of hot pixels are not strongly affected, the JVI is closed or open. For clarity, it is worthwhile to conduct an experiment in bright sunlight. Wait for spring.

    • Valentine

      Flashlight in the JVI. There was already such an experiment. Only a direct hit of sufficiently bright light affects a little. In other cases, you can neglect. In general, you are right, there is no significant influence. But when shooting in bright light in LiveView mode (when the eye is not pressed to the JVI), you can play it safe and cover the viewfinder with available tools.

  • Edward

    Here is the link

  • Edward

    With what, in fact, the experiments began.
    I noticed in my pictures a lot of gray round spots. Hundreds Thought: dust on the matrix. I took the device for cleaning. Cleaned. But they said that it was not dust, but broken or hot pixels. Began to experiment. Found out that defective pixels are not like these round spots. The spots are much larger in size, they do not have sharp edges with a halo and a round shape.
    In the attached picture, these differences are clearly visible.

    What are these specks? Who can guess?

    Big picture

    • Valentine

      As an option for the influence of the lens. Lens dust, inhomogeneous glass structure, etc. On the diaphragm 22, this can occur. On real shots, this can interfere very, very rarely, only if you do not have any purely specific tasks when shooting.

      • Edward

        I tried to shoot with this camera with a different lens. The defects are the same. Apparently it's not the lens. Only the matrix remains. But it was being cleaned ... Riddle.

        • Edward

          ... "I tried it on this device" ...

          • anonym

            "Cleaned" and "Cleaned up" are very different concepts.

        • Michael

          Classic frame with dust. Only her more than a lot. I've never seen so much. I have not seen other options for the appearance of such a picture

        • Michael

          Although met, the oil on the matrix gives a similar picture

  • Andrei

    Damn, I decided to check the camera down for broken pixels (I’ve been using it for two years) set ISO100 and 8 shutter speed. Found a bunch of hot pixels. Got upset. I decided to check for less shutter speed. I put 500 and almost did not find them, what does this mean? can I wasted in vain?

    • KalekseyG

      "Hot pixels are usually lonely, but with very long exposures, their number increases, as shown in the examples in this article."
      If you don’t shoot stars and a night city, then don’t worry.

  • Paul

    How to make a remap at home on Nikon D3xxx?
    Bold, hot pixels at ISO 1600 and shutter speeds of 1/100.

  • Dmitriy

    What to do with a lot of dead pixels on the matrix? More than 15.

    • Michael

      Remap. SC is waiting for you

  • Dmitriy

    What is the price in ye?

    • Dmitriy

      50 dollars remap. I think it’s worth it to make an old device for $ 250.

  • Dmitriy

    If white and colored dots appear in the photo on iso 640 and higher, and with iso 100 at 1/200 on black I can’t find them, then maybe it’s not broken pixels? Clear, confused.

    • Michael

      These are the so-called “hot pixels”. The article is just about them

  • Dmitriy

    Thanks. And I thought that whites are beaten.

  • Maria

    I first noticed it on Canon. And with low mileage. 5d Mark 2. Before that, I had no idea what this could be with the Nikons ...

    • Roman

      Well, if 5D2 was taken away by filming, because they were shot by everyone who is not lazy, but Nikon is not, then how can one be compared?

      • Maria

        Her, well, Nikon also experienced, which D800). In Canon, by the way, the mileage is not so very. I just chose this principle. The service confirmed 20 k.
        I can’t say that the 800-ka in any more advantageous position. unless on a computer 800 displays 7k points, and five to 5k. Maybe this also somehow affects.

        • Jury

          a lot of video was filmed on 5d Mark 2, therefore, the mileage is small (the shutter is not worn out), but the matrix is ​​all in hot pixels. On D800, the pixel is smaller - hot ones are less noticeable, plus they can be easily removed.

          • Maria

            Clear. HERE, too, such an idea overtook. But everything is logical. It remains to google how to programmatically remove. In general, without zooming it is not particularly visible.

      • Maria

        By the way, I apologize to everyone for the misinformation.
        I looked closely. The pixels are “knocked out” there. Where, out of ignorance of the converter, I “twisted” this software contour sharpness. Without her, problems are generally not visible. And each pixel is filled with some information according to the corresponding, not starry tone.
        Although the seller from the overbought confirmed that the camera was often used for video.
        So your truth turned out to be)

        In general, it is clear that the miser pays twice, of course. If you want a BU - expect surprises.
        After all, as I understand it, for Cannon and programin like ShowExif there is no way to check the shutter mileage.

        Here with this photo everything should be in order
        And here - whitish spots
        I don’t know what hosting to throw, so that it can be seen and you can give advice on photos.

        Thanks for helping everyone)

        • Maria

          And in fact, the hot pixels were confused with the single killed post-processing. After carefully reading the article. Full Fail) Sorry.

        • Iskander

          To check the mileage of Canon there is a program Canon EOS Digital INFO.

          • Yurko

            Tilkyot on exclusive OS is not a canaє .. (

            The one that is very principled - that try? .. To cope with your tasks, then th ok. If you can't cope - it’s tricky))

  • Larion

    I got rid of hot pixels on the Nikon d7500 by cleaning the matrix twice in a row. The method has been described here.

  • Yurko

    oh єs! When video is shown, especially if you catch fire .. - mind - you need to minify the camerat (drive to steer! Not to beat "zirok", not to play)).

    and for a photo, it is effective - rav ryatuє. Tai, jpeґ in a large number of correctly projected scenes є in solidarity, to tsyo zorepadu.

    ps I, to the tsії statty, do not know the same, tsі "zirki" in the photographs. Zrobiv 30sec test camera. І .. oh-yo-y - my eyes didn't see it)))

Add a comment

christening photographer price Photography for lovers

Copyright © Blog author - Photographer in Kiev Arkady Shapoval. 2009-2021

English-version of this article