3

I wonder that if I increase the exposure of a RAW in some software, say, Lightroom, what do I lose in the image.

I know that this is related to how is increase of exposure achieved, here in a few questions below, they gives some explanations. The main stream among them is to say that it's done by increase ISO or analogous way. Can you provide me some reference of this? If it's done by increase ISO, is it same as I increase ISO when capturing the photo? If they won't be the same, what's the difference or which way is better?


I've read through some of related questions, but neither of those perfectly answered my questions. (Here's a list)

and some from other sites.

  • 2
  • If that one doesn't answer your question, please edit to explain what you still need more information on. – Philip Kendall May 1 '16 at 12:57
  • @PhilipKendall: The other question doesn't discuss the ramifications of in-camera vs. post, but I'm biased because I wrote an answer. :-) – Blrfl May 1 '16 at 13:22
  • @PhilipKendall: Well, I think I want something more exact on the difference between how software and camera increase the exposure, which are not given in those questions, but anyway, they are truly closely related. (i.e. I want some more radical). If you think that put this as duplicate will help other, I have no objection:) – YiFei May 1 '16 at 14:20
  • @YiFei I'd suggest adding a bounty to a previous question, asking for more detail. – mattdm May 6 '16 at 12:05
4

...is it same as I increase ISO when capturing the photo? If they won't be the same, what's the difference...

The end result is similar, but how you get there and the side effects are different.

Increasing the ISO setting on the camera results in the addition of gain (amplification) in the path between the sensor and analog-to-digital converter, which is what turns the light levels into numbers. This is done with analog circuitry which amplifies the signals that make up the image to make them brighter. The trade-off is that this process also amplifies any noise in the signal, and that ends up in your image even if you don't make any adjustments in post.

Making exposure adjustments in post requires a bit more explanation:

For the sake of simplicity, say your camera produces grayscale images containing pixels whose light levels range from 0 (black) to 100 (white). A properly-exposed image will have its light levels cover this entire range, giving you 101 distinct shades of gray. (See this tutorial on histograms for further explanation of what this means for your photos.) An underexposed image will cover a smaller range of those values, say 0 to 50, because the amount of light falling on the sensor isn't enough to drive any of the pixels all the way up to 100. Even though the camera can record 100 different shades, this particular example records only 50. Fewer shades means fewer details in lower-contrast areas of the image. The more you underexpose the image, the worse this problem becomes. (Tangentially related is posterization, which is when this is done on purpose.)

Making a positive exposure adjustment with your processing software multiplies each pixel value by some value. In the example above where the brightest pixel value was 50, multiplying by 2 would bring that to 100. This results in an image that appears better-exposed because it covers the full black-to-white range. Because the software can't recreate information the camera didn't record, you're still working with the same 50 shades, and that lack of information becomes easier to spot in the adjusted image.

...or which way is better?

Neither. The ideal for any image is to expose it at the lowest ISO possible (to minimize noise) and long enough to make sure the image's histogram covers the entire light-to-dark range (to maximize detail). Anything else is a compromise that requires weighing what you're shooting with the kind of results you can live with. All other things being equal, you can get more detail by settling for a higher noise level or a lower noise level by settling for less detail.

If I had to pick one over the other, I'd opt for higher ISO, but that's because the body I shoot with does well in the dark and the camera-profiled noise reduction in my processing software does a decent job. Your camera and software may behave differently, and the only way to know which option is right for you is to experiment.

  • 2
    One thing to point out is that ISO Invariant cameras exist in which ISO setting on the camera is exactly the same as increasing exposure in post. They apply gain digitally after ADC, but not gain before. Most cameras apply an analog gain to the signal from the sensor (with possibly digital gain as well), so ISO setting differences is not possible to exactly reproduce in post. – Joe May 4 '16 at 12:44
2

Can you provide me some reference of this? If it's done by increase ISO, is it same as I increase ISO when capturing the photo? If they won't be the same, what's the difference or which way is better?

Some cameras have analog ISO gain, some don't. Some cameras may have analog gain implemented only in certain ISO range.

Different raw processors may have various optimizations that work better if the ISO is set properly in the camera.

The results will depend on camera/raw processor you use and since the implementation details are usually not shared by the manufacturers, you need to do your own testing with the equipment you use (unless somebody already did that and published the results).

That being said, somebody did thorough testing of the camera I am using and found that better results are achieved if one shoots at base ISO and compensates the exposure in post processing. Yet I am still shooting at the proper ISO. This way I can evaluate the shots on the back of my camera and reshoot if I am not happy. This is more important for me than modest improvement in noise...

1

(Disclaimer: I'm Italian, it's fatiguing for me to write in English and moreover to write in technical jargon. Thus, the following explanation take a few shortcuts to be easier to write and to understand)

I'll start from the bottom; increasing ISO in camera cannot obviously be like increasing the exposure of a file, whatever the algorithm used: increasing the ISO means that you are amplyfing the analogical signal (like raising the volume of your stereo, in a certain sense), increasing the RAW exposure means that you are working on the digital data stored in the file. Although in the end you can get a similar result, they are a totally different approach with totally different drawbacks and advantages.

user2719's answer in the third linked question sums it up pretty well: you have a signal, which means Data + Noise. If you use ISO then you are amplifying it, so you end up with Amplified Data and Amplified Noise. Then you digitalize it, which introduces more noises (in a sense), so now you have Amplified Data + Amplified Noise + Additional Noise. Working with RAW and the camera native ISO, instead, (faking that you don't amplify the signal) (we are trying to keep things simple, ok?) you begin with Data + Noise. Digitalize it, and you get Data + Noise + Additional Noise. Apply an exposure bias in a software, and you get Amplified Data + Amplified Noise + Amplified Additional Noise.

All of this is generally speaking. For example, another thing to keep in mind is that usually the electronic circuits in the camera have been thought and optimized (well, we hope at least, for what we pay it) to work together, so when you raise the ISO is not just some kind of third party thingy that casually found itself in the camera body.


So, in short:

  1. Is it same as I increase ISO when capturing the photo? No way: an electronic amplifier before the process of digitalization cannot in any way be the same of taking the color value of a pixel and multiplying it by a given value.

  2. If they won't be the same, what's the difference or which way is better? It's always better to proper expose; after that you can play miracles with a software in case of errors, but otherwise it's like driving with eyes closed because there are airbags in the car.

  3. Can you provide me some reference of this? Well, why not? :-)

Copy paste straight from my chosen RAW developement software. Enjoy:

Utilizing Difference between Exposure Biases of Camera and SILKYPIX®

The exposure bias on cameras slows down shutter speed, opens the aperture and in some cases reduces light with ND filters, thereby changing the amount of light falling on the image sensor. On the other hand, exposure bias when developing corrects the exposure recorded in RAW data and develops the photograph.

For example, consider performing exposure bias of +1.0EV. By performing exposure bias of +1.0EV on a camera, the amount of light recorded on the image sensor is doubled. Performing exposure bias of +1.0EV at the time of development, development processing is performed with the amount of light recorded in RAW data doubled. As a result, both are the same, but they each have their own advantages and disadvantages.

A drawback of increasing exposure bias when developing is that noise is doubled. Compared to exposure bias on the camera, you may result in a photograph with a bit more noise. However, there are huge merits to exposure bias when developing. One is that it gives you the ability to control over exposed highlighted portions. When performing exposure bias on a camera, if the exposure bias is too large, it may not record over exposed highlighted portions. There is no information for portions that go over the image sensors dynamic limit, even in RAW data, and there is nothing to retrieve when developing, even when developing with lowered sensitivity. (*1)

Also, the appropriate amount of exposure bias must be determined and settings on the camera be changed when performing exposure bias on the camera. You can reduce the time and labor when shooting and concentrate more on taking your photograph by performing exposure bias at the time of raw development. Furthermore, exposure bias on the camera often influences the shutter speed, and you must take care concerning camera shake or the subject moving. Increasing sensitivity during raw development normally lessens noise if at around +1.0EV. This is considered an effective photographic method for digital photography.

Also when it is difficult to determine an exact exposure under shooting conditions such as backlight or so on, take a picture with a little underexposed value using the exposure bias. [cut] There is a more advanced technique. When it is so dark that you have to use a slower shutter value or a faster lens, one smart way is to try to use a low shutter speed or aperture value to result in an underexposed image. [cut]

*1 Depending on the type of camera, developing with a lowered sensitivity of around –0.5 to 1.0EV is possible, and you may be able to save skipped portions, but many cameras cannot save whiteouts with lowered sensitivity at development.

Utilizing Difference between ISO Speed Adjustment of Camera and Exposure Bias of SILKYPIX®

The ISO film speed adjustment on the camera usually changes it at an analog level before digitizing the voltage output from an image sensor.

For example, with the ISO film speed doubled, information is doubled and sampled with an amplifier from the image sensor in the camera. Light quantity recorded by an image sensor does not change just because the ISO film speed is adjusted. Doubling the ISO film speed halves the light quantity recorded by the image sensor. There are hardly any differences from performing developments with an increased sensitivity of +1.0.

For example, compare photographing with a camera set at ISO 400 and photographing set at ISO 200 and exposure bias of –1.0EV (one step under). Other photographic conditions such as shutter speed and diaphragm stop are the same. The amount of light recorded by the image sensor for these two photographic methods is the same. The RAW data photographed at ISO 400 is converted and recorded at double the amount of light. When developing the RAW data photographed at ISO 200 into a photograph, sensitivity must be raised to +1.0EV. However, photographs developed with increased sensitivity are almost the same as photographs taken with ISO 400. When comparing these two photographs, the amount of light recorded by the image sensor is the same and both are processed in terms of double the amount of light in the end, and the only difference is whether it was performed in the camera or at the time of development.(*1)

When set and photographed at ISO 400, if highlights are ignored, it is impossible to restore them. However, the possibility of being able to save them by adjusting sensitivity rises if you set and photographed with ISO 200 at an exposure bias of –1.0EV (one step under).

As one application, it becomes possible to confidently take photographs of a dark subject that you have to take by raising the ISO film speed on the camera and working with the shutter speed and diaphragm stop to underexpose on purpose.

*1 The processing of photographs at ISO 400 and doubled in the camera can perform processing to remove noise in the camera at an analog level, which is slightly advantageous for the image quality. However, the difference is not that great, such that the technique of underexposed photography has merits, such as when there is a possibility that highlights may be skipped.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.