Prompted by reading Thom Hogan’s review and comments made about the D7000 began to make me wonder about this issue.
It was also spurred on by my experience with the measurement procedure brought on by a response by Robp to a post over in the ‘Printers, Scanners and Color Management’ forum. It was a procedure that allowed you to use your cameras meter to estimate the cd/m2 that the monitor is displaying at. I used my D7000 and then out of curiosity used my D80. Both cameras produced a shutter speed of 1/125 which put my monitor’s brightness at the top of acceptable levels. But the more interesting result was when I took an image of the white screen which should have come out middle gray. On the D80 the histogram peak was just left of center which coincides with Thom’s “Meter’s don’t See 18% Gray”. But the D7000 histogram was just right of center.
Both cameras were set to manufacturer defaults as far as camera JPEG output was concerned. Both cameras chose an exposure of 1/125 @ f/5.6 and ISO 400, but the resulting images had dramatically different result.
Which brings up the question; does the D7000 actually over expose (possible sensor site saturation) or is the tone curve in the D7000 tweaked differently then preceding models which tends to provide a hotter PP making it seem like it is over-exposing (clipping)?
If it is the latter and you are prone to chimping adjusting the exposure accordingly, then you actually may be under-exposing reducing the DNR ability and increasing the noise in shadow areas.
#1. "RE: Tendency to over-expose… or maybe not!" In response to Reply # 0 Tue 21-Dec-10 09:00 PM by rasworth
I assume you used matrix metering, which does not aim for middle gray, rather trys to expose such that the entire luminance range of a specific scene is captured to minimize clipping at either end. I don't know where the camera would try to place the exposure for a constant brightness image using MM, but my prior experience (D300) is that very flat lighted scenes tended to be over-exposed from a visual appeal standpoint, but not clipped. Thom Hogan has reported on some degree of weighting by the portion within the focus rectangle, although I personally haven't seen such with my D7000.
I assume if you repeated your experiment with spot metering (and probably also center weight) the monitor screen image would be closer to middle gray.
#2. "RE: Tendency to over-expose… or maybe not!" In response to Reply # 1 Tue 21-Dec-10 09:43 PM by elec164
>I assume you used matrix metering
Actually no, I used spot metering.
Edited to add:
Out of curiosity prompted by your response I just tried the three metering modes and they all provide the same shutter speed as I suspected they would.
For a metering system to be relevant and useful it needs to be calibrated to a standard. And through reading material on this subject it appears most manufactures use between 12 and 14% reflectance. In fact my Olympus P&S returned the same shutter speed indicating that all three cameras seem to have a similar calibration standard.
In fact when I think about what you wrote, it does not make sense in this case. Yes 3d matrix metering is complex and takes into account not only all the colors of the scene, but seems to place emphasis on the focus distance as well. But in this circumstance the whole viewfinder is filled with the white screen. The instructions also state that it is important to use infinity focus to insure that you do not pickup a possible false reading if the LCD mask or RGB segments are recognized. That means that whether the screen is black, gray or white they will all be rendered the same 12 to 14% reflectance and you would not, for the most part, be able to tell which was which when viewing the capture.
But all that aside, it still would not explain the variation in the histogram. Both cameras chose the same shutter speed for the given aperture and ISO setting in Aperture Priority mode. Depending on how many stops you say the histogram represents. If its 5 stops then there is about a 2/3 stop difference; but if it is 8 stops then there is more like a 1 stop difference in the spike placement.