Twelve examples of mFT vs FF with equal exposure and DOF

gardenersassistant wrote:

James Stirling wrote:

Why did you underexpose the A7II shots in every sample underexposing and pushing the exposure introduces noise to an image your A7II shots all have negative EV applied ranging from -0.3 to -1.7 .

As I mentioned in the top post,

"The captures were exposed so as to protect highlights and this meant that some of them looked rather dark “out of the camera”. This is how they looked on being imported into Lightroom with no adjustments.

View: original size

The extreme underexposure for scene 1 was deliberate, and not to do with protecting highlights. I wanted to see how the cameras responded to strong shadow lifting of the type that I sometimes want to do."

Also from the top post,

"In order to choose a starting level for the exposures I chose an aperture for the A7ii and then found a shutter speed and ISO which was practical to use and which just avoided any highlight zebras. I used this for the A7ii Equal exposure shot and worked out the settings for the other two shots. " [One of which was a G80 shot with equal exposure to the A7ii Equal exposure shot, the other one an A7ii shot equivalent to the G80 shot. These two exposures were calculated directly from the A7ii Equal exposure shot settings, Some of the actual settings were 1/3 stop away from the calculated values as I used Av mode some of the time and the settings wandered a little after I had set them to the calculated values.]

So all the shots in each set of 3 had (aside from any random "wandering" of the settings) the same amount of under-exposure. This degree of under-exposure is consistent with the way I work, especially with flowers. Many of my shots are underexposed by those sorts of amounts, and sometimes more. So for my purposes these were realistic exposures.

The A7II is nosier than other FF cameras of that generation

A7II DPreview conclusion https://www.dpreview.com/reviews/sony-alpha-7-ii/12

• Noisy high ISO images compared to full-frame competitors

Example

Equivalence actually describes how to achieve the same results on different systems so no surprises here

Indeed so, which is why I termed one of the two A7ii shots for each set of 3 the "A7ii Equivalent shot". My only surprise was that when viewing equivalent photos the A7ii looked in some cases just a little noisier than the G80 shot, with none the other way round.

First, let me express my appreciation for all the effort behind your tests.  Very well done!

That said, the reason the equivalent A7ii photos were a tad more noisy is due to the fact that, while the two photos are made with the same amount of light (the same amount of light is projected on the sensor for equivalent photos and the two sensors record the same proportion of light falling on them), the A7ii sensor has more electronic noise than the newer G80 sensor, resulting in noticeably more noise for the darker portions of the photo.

The two primary sources of noise are the photon noise (noise from the light itself) and electronic noise (noise from the sensor and supporting hardware).  The lesser the amount of light that makes up the photo, the more the electronic noise matters in comparison to the photon noise.  So, the noise differential for darker equivalent photos will be greater than the noise differential for lighter equivalent photos, in the case of the A7ii vs the G80.

There is much more variation in the electronic noise between sensors than there is between the proportion of light projected on them that gets recorded.  This is why we will see noise equivalence work out so closely for photos taken in "normal" conditions ("properly exposed" photos below, say, ISO 6400) but greater variation as we push the shadows more and/or shoot in light low enough that we'd be using ISO 6400+.

Canon sensors, in particular, have greater electronic noise at lower ISOs than Sony sensors (and everyone seems to be using Sony sensors aside from Canon), and that's why Canon cameras lag behind others with regards to low ISO DR and noise performance at very high ISO settings.  However, in "normal" situations, Canon sensors do as well as the others.

Had you performed the test with a more modern Sony camera, like the A73, you'd have found the differential to be much less (and likely favor the A73 a tad, due to its BSI sensor, which allows it to record about a third stop more of the light that falls on it than non-BSI sensors).  But, still, it would be close, since we're not talking about huge differences here.

Anyway, thanks again for the test -- well done!