Canon's 18mp Sensor

Started Jul 14, 2012 | Discussions thread
KLO82
Contributing MemberPosts: 705
Like?
Re: You're right.
In reply to bobn2, Jul 19, 2012

And here is dxomark:

Color temperature (light spectrum) and intensity of the various lights used to illuminate the test targets are accurately set and controlled. The lighting setup at the DxO Labs testing facility can illuminate targets from about 5 to several thousand Lux with a 5% spatial uniformity over the target area (up to approximately 3m wide and 2m high).

http://www.dxomark.com/index.php/About/In-depth-measurements/DxOMark-testing-protocols/Testing-lab

bobn2 wrote:

KLO82 wrote:

TrojMacReady wrote:

The difference in lighting would not matter as long as they are determining exposure by their same Sekonic (incident) light meter.

I agree if the illumination of the scene would already be equally spread, but seeing as the top is sometimes darker than the bottom, vice versa, etc., that doesn't seem to be the case either. And overall it can result in differing shutterspeeds, thus making it impossible for us to judge sensitivity differences by just looking at the exposure settings.

As long as they are using same light meter to detemine exposure and shoot accordingly, we do not have to consider the difference in shutter speed or f number. For example, say their light meter showed that exposure needed for ISO 100 was f5.6, 1/250 sec, and they shot accordingly to that. After some time, they have completely changed the lighting during testing of another camera. This time, the meter showed that for ISO 100, the settings needed are f5.6, 1/100 sec (due to decrease in light intensity in this case). In both cases, the exposure should be same [same amount of light per unit area should reach the sensor, not considering difference in t stops and stated and actual shutter speed]. But where they mix up is, when they find some camera does not show 18% gray card as mid gray in sRGB output, they disregard the meter reading and increase/ decrease the shutter speed (or f number) to make the gray card as mid gray in jpeg - and note how much they had to compensate and based on that they announce the camera has "overstated" or "understated" ISO. If they would stick to the shutter speed and f number setting according to the meter reading, the comparisons would be fair.

But you are of course correct about the not equally spread lighting will ruin the tests.

I had an exchange with Andy Westlake about their lab procedures, and he said that the reason that they do this is because they will often be running tests in the same room which have their own lighting, so the light on the studio scene will change. Strikes me that it is likely that the distribution of the light on the scene will change also, in which case it becomes meaningless to compare details from the scene, as some are wont to do.
--
Bob

Reply   Reply with quote   Complain
Post (hide subjects)Posted by
Umm.New
See!New
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark post MMy threads
Color scheme? Blue / Yellow