1
\$\begingroup\$

I recently was playing around with my camera (Canon 60D) in a low light situation trying to focus on some subjects across the room. I noticed once getting home and getting the images onto the computer that the image quality was absolutely horrible. I was so very frustrated with such poor quality.

This is atypical, as my lens / camera combination have taken some very sharp pictures in the past (and do most of the time). I could understand if I had an issue with camera shake / long exposures, etc, but that is not the case here.

Below are the specs for the shot:

  • Camera: Canon 60D
  • Lens: Canon 50mm f/1.4 USM with a UV filter
  • Aperture shot at: f/1.4
  • Exposure: 1/160 sec
  • ISO: 640

I shot over 100 images and they almost all look the same, extremely soft, almost like there isn't a focus at all to the image. You'll also notice TONS of chromatic aberrations (In the window, its entirely outlined in purple).

Did my lens not focus? Does low light simply not take as sharp of photos even with decent exposure time/ISO/aperture? Is this caused by a dirty lens? If so, why am I not seeing this in brighter situations?

Any help would be appreciated!

(Large version: https://i.sstatic.net/ydWqf.jpg)

Blurry image

Update

FYI - I used 1 point of focus versus all focal points on the camera (excuse my lack of detail here, I'm not positive what those points are called exactly).

\$\endgroup\$
6
  • 2
    \$\begingroup\$ That lens is known to be soft especially wide open. It is certainly the softest lens that I own. If you need substantially better sharpness I would recommend the Sigma 50mm f/1.4. You simply won't find it in the Canon wide open. \$\endgroup\$
    – dpollitt
    Commented Sep 30, 2013 at 20:23
  • 1
    \$\begingroup\$ If you stop down to f/1.8 the EF 50mm f/1.4 is much sharper than at f/1.4. This is true of most wider aperture lenses. \$\endgroup\$
    – Michael C
    Commented Sep 30, 2013 at 20:28
  • \$\begingroup\$ @dpollitt Nah. At f/1.4 the Sigma is only slightly sharper in the center than the Canon f/1.4 (which is sharper in the mid-frame and edges) and about the same as the Canon f/1.8 II. At common apertures above f/2 there is little difference between any of the three. Click 'Measurements-->Sharpness-->Profiles' to see the chart. dxomark.com/index.php/Lenses/Compare-Camera-Lenses/… \$\endgroup\$
    – Michael C
    Commented Sep 30, 2013 at 22:27
  • \$\begingroup\$ And the-digital-picture.com/Reviews/… \$\endgroup\$
    – Michael C
    Commented Sep 30, 2013 at 22:31
  • \$\begingroup\$ @MichaelClark - I've used both in the field side by side. I found the Sigma to be significantly sharper at f/1.4. That is my opinion. \$\endgroup\$
    – dpollitt
    Commented Oct 1, 2013 at 13:22

3 Answers 3

13
\$\begingroup\$

It seems to me that the image is quite adequately sharp given the lighting conditions. Some things affecting the sharpness of a photo in low light:

  • Almost any lens is going to be somewhat soft at its maximum aperture. As far as I know, the Canon 50mm/1.4 at f/1.4 is a bit on the soft side relative to other fast primes (even compared to the cheapo 50mm/1.8!)

  • The larger the aperture, the tighter the depth of field - most of the things in your photo are simply not in focus due to the razor-thin DoF at f/1.4.

  • The less light, the less contrast and the less accurate the autofocus mechanism is. And the larger the aperture, the easier it is for even a tiny focusing error to throw the focal plane slightly off.

  • Related to the previous two bulletpoints, moving the camera even slightly (eg. when doing "focus-and-recompose") might well be enough to cause noticeable misfocus. Using the non-center autofocus points might help, but they are often less accurate than the center one.

  • The higher the ISO, the softer the image due to noise reduction. This is something that can be tweaked in post-processing if you shoot RAW, but it's always going to be a balancing act between noise and softness.

\$\endgroup\$
5
  • 2
    \$\begingroup\$ If both lenses are set at f/1.8, my EF 50mm f/1.4 is sharper than the 50mm f/1.8 but not by much. For me the advantage of the f/1.4 is the smoother bokeh and the much more usable manual focus ring. And durability. \$\endgroup\$
    – Michael C
    Commented Sep 30, 2013 at 20:30
  • \$\begingroup\$ @MichaelClark Fair point. I've heard from another source that the 50mm/1.8 would be a bit sharper in the center at f/1.8 but don't have first-hand experience. \$\endgroup\$
    – JohannesD
    Commented Sep 30, 2013 at 21:13
  • \$\begingroup\$ At f/1.8 the 1.4 is a little sharper in the center, but the 1.8 is sharper on the edges. The same is true at f/2. From f/2.8 up there is no practical difference in image quality. the-digital-picture.com/Reviews/… \$\endgroup\$
    – Michael C
    Commented Sep 30, 2013 at 22:14
  • \$\begingroup\$ DxO Mark measure the 1.4 as sharp in the center at f/1.4 as the 1.8 II is at f/1.8. Click 'Measurements-->Sharpness-->Profiles' to see the chart. They're both much better by f/2.8. dxomark.com/index.php/Lenses/Compare-Camera-Lenses/… \$\endgroup\$
    – Michael C
    Commented Sep 30, 2013 at 22:18
  • 2
    \$\begingroup\$ even without noise reduction the noise (not just from the gain aka iso but also the photon noise) diffuses the edges and thus the perception of sharpness. \$\endgroup\$ Commented Oct 2, 2013 at 16:17
4
\$\begingroup\$

JohannesD covers all of the basics very well and I won't repeat what he has already covered. I would like to add some additional observations.

  • The image is in focus. The point of focus is the lobster bib and corner of the table. The narrow depth of field created by the f/1.4 aperture means everything even slightly closer or further away from the camera will begin to soften rather quickly.

  • Almost any lens will show some chromatic aberration under the conditions in your photo. The extreme contrast between the bright daylight outside and the dark drop shadows in the window lettering will almost always induce some CA. Assuming you shot RAW files, you should be able to control the CA more in post processing.

  • The chromatic aberration in the reflection on the top of the booth and in the window is likely exacerbated by the UV filter. Step back slowly, take the filter off, and nobody gets hurt! LOL. The front element of the EF 50mm f/1.4 is recessed about 3/4 inch from the front of the lens housing. If you drop that lens the most likely thing to be damaged is the focus collar inside, which is a known issue with this lens. (Most lenses have strengths and weaknesses. This is the most serious vulnerability of this lens' design. If you don't drop it, it should never be an issue for you.) A filter won't do a thing for that. What even a very high quality filter will do is slightly degrade the Image Quality, and a cheap filter will do it more than slightly and add reflections and flare as well.

\$\endgroup\$
3
\$\begingroup\$

Low light results in quantum mottle. Statistically, there are so few photons detected by each pixel that statistical probablilty begins to appear. Imagine a perfectly smooth homogenous 50% reflective gray surface perfectly aligned perpendicular to the central axis of the lense and parallel to the detector grid. Theorectically each pixel should be receiving exactly the same number of photons per unit time of every single color but only 50% of the ambient light in the room is reflected back to the lens.

If the room is bright enough for tens of thousands of photons to register in each pixel over the exposure period, it would be like flipping a coin tens of thousands of times. You might not get exactly half heads and half tails, but even if you are off by several hundred coin flips, it amounts to a variation per pixel of less than a thousanth of one percent.

Darken the room enough, and it may be like flipping a coin only 10 times, and recording the number of heads in each series of coin flips once in each pixel. Even if you are off by only one (6 heads) that amounts to a 10% difference per pixel from the average.

Image processing that attempts to eliminate quantum mottle will average or smooth the results from pixel to pixel resulting in a loss of contrast and resolution.

Further, the efficiency of the detectors to different energy photons will exaggerate these effects and cause shifts in color to those wavelenths that the detectors are most efficient at detecting.

\$\endgroup\$

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.