Previous page Next page

Low light high ISO (compared)

One long-running concern about cameras' ever-increasing pixel counts is the impact on high ISO noise. Fitting more, smaller pixels onto a given size sensor almost inevitably ends up with higher pixel-level noise when images are viewed are viewed at 100% on-screen. Sony's older 16MP sensor, as used in the NEX-5N, has earned itself an enviable reputation for its high ISO output - here we're seeing how the NEX-7 stacks up in comparison. The question to be answered here is how any increase in pixel-level noise affects the image as a whole; i.e. does the higher pixel count have a negative impact if you compare prints of the same size, rather than view on-screen at 100%.

In the examples below we're comparing the NEX-7 against the SLT-A77 and the NEX-5N; in the case of the latter, we're showing 100% crops from both its native 16MP output, and upsampled to 24MP to match the other cameras (using Photoshop Bicubic Smoother). The samples are shot at ISO 6400 under low colour-temperature (2600K) artificial light, designed to be representative of typical indoor lighting. This accentuates the appearance of noise due to the low level of blue light in the spectrum of the light source. This means that to achieve accurate white balance the blue channel has to be amplified strongly, and the green channel to a lesser extent - thereby increasing the visible noise.

All samples were shot using the same lens (the Sony DT 50mm F1.8 SAM) at the same set aperture (F8). The exposure was chosen to match the output brightness between cameras - under these conditions the SLT-A77 required a shutter speed 1/3 stop faster. Assuming consistent aperture setting between cameras, this would imply that the A77 received less light (which in turn would result in slightly higher noise).

These examples were shot in RAW then developed in Adobe Camera RAW with noise reduction turned to zero. This is intended to show the underlying noise levels, and therefore isn't necessarily representative of how you'd process the files in practice.

Sony NEX-7, ISO 6400, 1/40 sec F8 Sony SLT-A77, ISO 6400, 1/50 sec F8
Sony NEX-5N, ISO 6400, 1/40 sec F8 Sony NEX-5N, resampled to 24MP

You can see immediately from these images that while the two NEX models look rather similar, the SLT-A77's shadow regions are enveloped in blue haze. This due to the lower amount of light the A77's sensor receives, both from the shorter shutter speed and the fixed mirror diverting a proportion of the light to the AF sensor. Crucially, though, the NEX-7 shows that this isn't an inherent property of the sensor itself; just how it behaves in the SLT implementation.

Sony NEX-7 Sony SLT-A77
Sony NEX-5N Sony NEX-5N, resampled to 24MP

These 100% crops show the A77 to be clearly noisier than either of the other two cameras, but between the NEX-7 and NEX-5N there's really not a lot to choose. The NEX-5N looks less noisy when viewed at 100%, but upsample it to match the NEX-7 and the visual difference all but disappears. The same story continues in the highlight and shadow regions of the frame, shown below - the 5N tends to look better at 100%, but little different when the images are compared at the same size.

Sony NEX-7 Sony SLT-A77
Sony NEX-5N Sony NEX-5N, resampled to 24MP
Sony NEX-7 Sony SLT-A77
Sony NEX-5N Sony NEX-5N, resampled to 24MP

The inevitable conclusion from this is that, if you compare images as a whole rather than at the pixel level, the NEX-7's low light, high ISO performance is essentially a match for the NEX-5N's. So overall the increased pixel count delivers potentially more detail at low ISO, without any obvious negative impact on high ISO image quality.

Previous page Next page
483
I own it
200
I want it
112
I had it
Discuss in the forums

Comments