DXOMark just tested the D3X

Started Jan 15, 2009 | Discussions thread
leping Regular Member • Posts: 429
DxO DR is at 1:1 RMS noise | DxO Insight: Higher-Resolution Sensors = Better Results

Read the DxO DR definition (RMS noise level = luminance level) before comparing to measures using something quite different:

"3. Dynamic range

Dynamic range is defined as the ratio between the highest and lowest gray luminance a sensor can capture. However, the lowest gray luminance makes sense only if it is not drowned by noise, thus this lower boundary is defined as the gray luminance for which the SNR is larger than 1. The dynamic range is a ratio of gray luminance; it has no defined unit per se, but it can be expressed in Ev, or f-stops. "


How could more pixels does not bring down DR?

DR is determined by the sensor's full well photom capacity and thermal and read noise level.

It is obvious according to the science of physics, but many would just not accept. If the micro-lens(es) are "gapless", and you devide the one pixel area into four. End results? When you binning the 4 pixels together, or properly downsample to 50% size, you get back exactly the same SNR, since the level of noise is proportional to Sqrt(N), where N is the number of photons hit the single low resolution sensor pixel area. You lose SNR by a factor of 2, or 6dB, by checking the high resolution pixels at 100%, since only 1/4 of the photon is hitting, but after bining everything comes back exactly the same (just devided into four parts and individually read and properly combined), since the total number of photoms, and hense the level of fluation are the same.

However, by going 4 pixels instead of one, you GAIN the obth ways:

1. The option to use the highre resolution, in this case doubling;

2. The option of downsizing to 50% to half the noise. If the sensors are not Bayer type, that's it, and you do not lose or gain anything. Nevertheless, if the sensors are Bayer, which introduces another 3dB true resolution loss, you actually gain here again, because of the much less demosiacing artifacts, better color definition and smoother tonor transition after downsizing, all the higher resolution brings in in the first place for Bayer sensors, even when your lenses can not resolve to the single sensor size level, with or without CA, out of focus, and out of DOF.

These has been, as I tried to mention, but rudely insulted, adequately and confincely tested and confirmed in Lloyd Chamber's DAP reviews, with these free side-to-side comparison 100% unsampled/downsampled samples:


In the D3x vs D3 case, the area difference is not 4 but close to 2, so that the noise level differs by 3dB in the ideal (linear) world.

Go back to the DxO site, and checkout the following conclusions under the Insights tab:

"Contrary to conventional wisdom, higher resolution actually compensates for noise:

This Insight uses currently-available DSLRs to demonstrate the technique for objectively comparing noise for cameras with different levels of resolution. Such comparisons conclusively show better results overall for high-resolution sensors, despite the increase in noise. ..."

Thanks for reading.

-- hide signature --

4x5 film / 6x7 film / Canon 5DII & 5D-IR / Nikon D700

Post (hide subjects) Posted by
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow