G6 vs newest sensors: Real difference in high ISO noize

Started Mar 5, 2014 | Questions thread
texinwien Veteran Member • Posts: 3,326
Re: G6 vs newest sensors: Real difference in high ISO noize

Øyvin Eikeland wrote:

Ulric wrote:

BTW, when you look at the "Measured ISO" that DxOmark use, don't assume that it is more real or more correct than the manufacturer stated ISO. It is just defined differently. Look at their measurements within their own coordinate system and you'll be fine.

Hi, I have problems interpreting the DXO graphs. Some help would be appreciated. DXO is apparently calculating the "real" iso sensitivity of the sensor by exposing the sensor to a known intensity and then calculate the sensitivity from the raw files. Is this correct? The points in the graph are then plotted against measured ISO and not against what you set on the camera. The right-most data-point for the OM-D E-M5 is taken with the camera iso set to 25600, right? The measured ISO for this setting is only 11848. Therefore the data point is plotted at 11848 along the X-axis. What I do not get is this: How can I find the dynamic range (or SNR) you will get if I set the camera to an ISO of 25600? If the dynamic range in the raw-files are indeed 6.61EV, isn´t it unfair to plot the point just below 12800?

You need to be careful when looking at the graph and judging the precision distances. Don't forget that the x-axis scales exponentially and not linearly. So, while there are only ISO 100 steps between ISO 100 and ISO200, there are 6,400 ISO steps between ISO 6,400 and ISO 12,800.

Each doubling in ISO is 60 pixels wide on the graph. Between ISO 100 and ISO 200, each pixel will be (roughly, on average) a little less than 2 ISO steps. Between ISO 6,400 and ISO 12,800, each pixel will be (roughly, on average) around 106 ISO steps.

I say 'roughly, on average' because the last pixel between ISO 100 and ISO 200 is more than 2 ISO steps, whereas the first pixel between ISO 100 and ISO 200 is less than 2 ISO steps (again, because the x-axis is scaled exponentially). The same goes for the jump between ISO 6,400 and ISO 12,800 - the last pixels on the right account for more ISO steps than the first pixels on the left.

The formula for determining the exact ISO value at any pixel on the x-axis is 50 * 2^(n/60), where 'n' is the number of pixels from x=50 ISO (the far left of the graph). The center of the furthest right E-M5 dot is 473 pixels to the right of x=50 ISO. Plugging that into our formula, we have:

50 * 2^(473/60) = 11,806 ISO

If anything, the graph is unfair to the E-M5 by 42 ISO steps


Øyvin Eikeland

 texinwien's gear list:texinwien's gear list
Panasonic Lumix DMC-GM5 Olympus E-M5 II Olympus 12-40mm F2.8 OnePlus One Canon EOS 300D +20 more
Post (hide subjects) Posted by
(unknown member)
(unknown member)
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow