I test-shot frames in Manual VS Av mode with identical aperture/shutter speed/ISO, settings and lighting, and I noticed that some frames in Av are sometimes over or underexposed compared to those in Manual.

It can easily be reproduce with spot meter, taking a reading in Av of a slightly darker or lighter tone without change in shutter speed.

I wasn't testing it with tight tolerances, but I took more than enough shots (like two hundred, at every aperture in both Av and M modes for comparison, with redundancy to make sure it was not lighting inconsistencies), and I had pretty consistent results.

Point the spot meter at any subject and take a shot in Av. Then take and lock a reading at a slightly brighter part of the subject, but not so much brighter that the shutter speed would change, reframe exactly like the first and shot. Aperture, shutter speed and ISO are the same in both shots, yet the second will be a bit underexposed. It's even more evident and easy to see that result if the camera was set to 1/2-stop increments.

ALO (and HTP) were both turned off.

I also read on a famous reviewer website, referring to the EOS 6D, that ""In any auto exposure or auto ISO mode, exposure may only be read to halves or thirds, but the camera sets it steplessly anyway."

So, is that correct that the camera fine-tune the exposure at software level in Av mode?

If so, what about image quality? For example, if I shoot a raw in AV, the camera adds +0.17 EV and I further add +0.17 EV in post. Is it the same as shooting in Manual with same aperture/shutter speed/iso, and applying +0.33 EV in post, or worse?

  • \$\begingroup\$ @user49653 What were the specific lighting conditions under which you were shooting? \$\endgroup\$
    – Michael C
    Mar 15, 2016 at 4:53
  • \$\begingroup\$ Test shots I did were done under natural light indoors, at the window, and further tests at night under a regular room light bulb. Both with same reproducible, predictable, controllable results. Regarding the case of "fake" ISOs, aren't they applied directly to the raw data somehow, instead of instructions? If not, then wouldn't they be ignored by third party softwares too? \$\endgroup\$
    – user49679
    Mar 15, 2016 at 7:42
  • \$\begingroup\$ If you have test shots illustrating your question, why not add them here? EXIF data would help too. \$\endgroup\$
    – Caleb
    Mar 15, 2016 at 19:26

1 Answer 1


Most cameras don't set much of anything steplessly. They will meter in very precise graduations but when converting that metering to exposure settings it is going to be rounded to the nearest increment. The finest most cameras adjust exposure parameters is in one-third stop intervals. There are a few cameras that adjust shutter speed or aperture steplessly, some only when shooting in certain modes or with certain parameters selected. But they are far from the norm.

With regard to shutter speed (Tv) and aperture setting (Av) this is the case for your EOS 6D. In the case of ISO, most Canon cameras made in the last decade or so use only whole stop values (e.g. 100, 200, 400, 800, etc.) on the sensor and apply a push/pull or pull/push to images taken with -1/3 stop (e.g. 160, 320, 640, 1250, etc.) and +1/3 stop (e.g. 125, 250, 500, 1000, etc.) ISO settings, respectively. For more on how this works out practically, please see Is it really better to shoot at full-stop ISOs? and Is analog gain really actually power-of-two only?. And then there is this whole thing regarding the precision of target and actual values for Tv, Av, and ISO used by cameras: Is there a sane reason why ¹⁄₁₂₅ is not, instead, exactly half of ¹⁄₆₀?

Most of the inconsistencies you are experiencing are probably a result of minor differences in metering and/or the actual light conditions. If you are outdoors atmospheric conditions can alter the absolute brightness of a scene even on a sunny day. And with "controlled" lighting in a studio setting most flash guns will vary their output slightly from shot-to-shot. You have to spend a lot more to get the stuff that complies with very high tolerances. And it goes without saying that any source of light that flickers (Fluorescent tubes, CFL bulbs, etc.) will meter differently based upon where in the sine wave the light source is when metering is recorded.

When spot metering, how much variation it takes to move the exposure 1/3 stop depends on how close to the center or edge of an ideal brightness for a specific setting the meter was to begin with. Say you are metering a scene and the brightness falls squarely in the middle of f/5.6 for a particular ISO and Tv. The brightness can be increased by almost 1/6 stop before the aperture setting will be rounded up to f/6.3 instead of remaining at f/5.6. On the other hand, if the brightness calls for a value almost exactly between f/5.6 and f/6.3, changing the brightness even a miniscule amount might alter the f-number selected. It is just like rounding cents to the nearest dollar. If you have exactly $5, you need an additional 51¢ to round your balance up to $6. But if you already have $5.49 which would be rounded down to $5, you only need two cents to have $5.51 that would round your balance up to $6.

  • \$\begingroup\$ Michael, I understand that the camera, in any auto mode, meters in fine graduation, and then the measured exposure is rounded to the closest settings mechanically possible. But is there any evidence that the camera isn't bumping up the exposure afterwards, if for example it was rounded down to get back to the fine graduation exposure that was measured in the first place (in the same manner as iso push/pull)? That would makes it stepless like K.Rockwell claimed, doesn't it? \$\endgroup\$
    – user49679
    Mar 15, 2016 at 1:05
  • \$\begingroup\$ @user49653 I moved much of your answer up to your question, and converted the rest to a comment. Please edit as necessary. Michael, please see edits to the question in case they affect your answer \$\endgroup\$
    – MikeW
    Mar 15, 2016 at 2:24
  • \$\begingroup\$ @user49653 To the best of my knowledge there is no "fine adjustment" made to the instructions in the raw .cr2 file to compensate for anything other than actual (vs. stated) ISO sensitivity at any given setting and the offset used for +1/3 and -1/3 stop ISO settings. I've never seen any evidence of such a thing. And whether those instructions that are included in the file are even followed or not depends upon the raw convertor used to demosaic the raw data from the sensor and convert it to a viewable image. \$\endgroup\$
    – Michael C
    Mar 15, 2016 at 4:42
  • \$\begingroup\$ As for Ken Rockwell, he will say whatever he thinks will make him look more knowledgeable than everyone else and/or generate more clicks to his website. Sometimes the only way to be the exclusive purveyor of secret knowledge that no one else has is to make it up. \$\endgroup\$
    – Michael C
    Mar 15, 2016 at 4:46

Your Answer

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