I have not verified this on pixel-peeping level. But I have observed that photos taken with flash (built-in one in my case) appear to have very less noise at ISO 800/1600 than the ones taken without flash. I use Canon T3/1100D which gives a lot of noise at ISO 1600 normally.

Why does this happen? Or is it just my misconception?

  • \$\begingroup\$ Are you sure the flash photos were taken at ISO1600? It may be the camera is safety shifting the ISO to prevent overexposure. \$\endgroup\$
    – Matt Grum
    Nov 28, 2012 at 10:17
  • 1
    \$\begingroup\$ Alternately, are your flash photos at 1600 actually brighter? \$\endgroup\$
    – mattdm
    Nov 28, 2012 at 14:48
  • 1
    \$\begingroup\$ Either the ISO is getting silently lowered by the camera when shooting with flash, or the picture brightness is inadvertently increased for the flash off case while viewing, thus exposing more noise. \$\endgroup\$ Nov 28, 2012 at 19:30
  • \$\begingroup\$ I don't have any examples handy but I will try reproducing the scenario again. Also I shoot in full manual, I don't use Auto ISO. So how can camera change the ISO? @mattdm I use more ISO to capture more ambient light (when I am maxed out on aperture side of course). @ AlexanderShcheblikin I don't know any photo viewer which does that. \$\endgroup\$ Nov 30, 2012 at 5:40
  • \$\begingroup\$ Relevant reference reading: photo.stackexchange.com/questions/11910/… \$\endgroup\$ Nov 30, 2012 at 5:50

2 Answers 2


Assuming all other exposure settings with and without flash are equal, then using flash means you are adding light to the scene. Increased light in the scene means increased light down the lens, which means more light at the sensor. That means you have a higher signal to noise ratio at the sensor, which generally means less noise.

Signal to Noise ratio, or S/N, is extremely important to the amount and quality of noise in digital photography. Every pixel in a sensor has a certain range of luminance it supports, from the read noise floor (not quite pure black, but close to it) to pure red, green, or blue (at ISO 100 this is called full well capacity, and generally termed maximum saturation at all ISO settings). A pure red, for example, would be maximum S/N for that pixel. A pixel of an 18% gray card would be roughly 50% S/N. The lower the signal, the more apparent noise (of which photon noise dominates) will be. The higher the signal, the less apparent noise will be.

Any time you can add light to your scene, be it flash, other artificial light, or even sources of natural light, will usually improve the signal to noise ratio. At higher ISO, your maximum saturation will be a fraction of the full well capacity, which is almost the same as underexposing (at ISO 100) one stop for each stop of ISO, then pushing exposure in post by the same amount. If you underexpose at ISO 100 by two stops then boost by two stops in post, that would be similar to using ISO 400 (with the added caveat that you'll boost read noise as well when pushing in post, but let's assume an "ideal" sensor for the sake of simplifying discussion.)

If you underexposed by three stops, then pushed in post, that would be similar to using ISO 800. Before pushing in post, you'll notice that the image is pretty dark. If you do the same thing, underexpose by three stops at ISO 100...but add flash...the unmodified image in post should be a bit brighter than the first shot that was underexposed by three stops. Flash adds light, which increases signal level. Boost both images such that they appear the same post, and the second shot should appear less noisy, even if only by a small margin....it will require LESS of a boost to exposure to correct.


Because they have less noise. Or as @jrista points out, a better signal to noise ratio.

We take photographs with light. The flash simply adds light.

Imagine taking a photo of a black cat on a coal pile at midnight with no additional light. There are in fact, difference in the light coming from the cat, the coal, or the room. But the differences are tiny. So the sensors will emit a random value, based on thermal noise, random cosmic radiation, etc. Its all noise. Now add a flash, and there will be differences between the cat, the coal and the room, you have signal.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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