![]() ![]() The maximum turned out to be 2133, instead of the physically possible 3840, which is already, by itself, a sign of total underexposure by log 2(3840/2133)= 0.8 EV (see math above). However, +0.3 EV has already been manually added to the automatic exposure - this is visible in EXIF. Such a small total amount of saturated pixels means that the image is in fact underexposed - with normal exposure, the specular highlights on metal and glass surfaces should indeed appear as specular highlights, not monotone featureless areas, which only lower the apparent contrast of the image and do not bring any artistic merit. For illustration purposes we also added three arrows pointing to those blown out highlights. Turning the overexposure warning on and off one can see that there are also a few saturated pixels on the metal watch bracelet. Switching on the OvExp warning checkbox puts a red/purple overlay on saturated pixels. ![]() These pixels are located on a few tiny specular highlights caused by curvatures on metallic surfaces such as the button on the jacket pocket and metal picture frame. In the overexposure and underexposure section (OvExp/UnExp Stats), it's visible that only 367+290 pixels in the green channels are blown out (saturated). So, if we know the EV value, we can find by how much the linear count has changed by raising 2 to the power equal to the amount of change in EV units. ![]() So, analogously, changing the exposure by 1/3 EV would be the same thing as changing the linear count by the cube root of 2, that is 1.26 times (1000 to 1260 for +1/3 EV), while changing the exposure by 1/2 EV is the same as changing the linear count by the square root of 2, 1.414 times (1000 to 1414 for +1/2 EV). A change of 2 stops means changing the linear count 4 times, 1000 to 250 is -2 EV 1000 to 4000 is +2EV. Therefore, a 1 stop (+1 EV) change in exposure is equal to the 2x change (2^1) in linear raw count, like going from the count of 1000 to the count of 2000 if it is +1 EV or going from the count of 1000 to the count of 500 if it is -1 EV. Change in EV units is equal to log 2(the ratio of the second count to first count). We're going to use a few simple mathematic formulae, which reflect the relationship between the amount of the change in EV units (photographic stops) and the change in linear counts (linear RAW values). See ETTR topics and A CALL TO ACTION: Build Intelligent ETTR Into Digital Cameras and Why Don’t Digital Cameras Offer a Perfect ETTR exposure?.Here goes a little bit, (really, just a bit) of mathematics. A very serious failure in design thinking, and by all vendors. And auto exposure often leaves two stops of dynamic range unused. It is not uncommon for me to see a blown out exposure on a camera histogram that could have accepted another full stop of exposure, yet with zero loss of highlights. * Camera histograms are an abject design failure (Leica M Monochrom excepted) they force through white balance and a truncated color space, undocumented as to what is actually done, and are thus highly pessimistic on exposure, if not flat-out misleading. But it has many other useful features, including a detailed EXIF info view that I prefer to just about anything else.įor me, the true raw histogram alone is immensely useful I use it to check the actual exposure behavior of new cameras and to fine-tune my sense of what the camera histogram means*, and as well as to check for the amount of blow-out highlights far more clearly than the “blinkies” in some programs. For anyone doing ETTR, RawDigger provides a clear-cut answer as to whether the file is perfectly exposed, overexposed, or underexposed. RawDigger is not a workflow tool so much as a feedback loop on your own technical success with exposure. I recommend it for all shooters looking to get the best from their camera. It comes in three editions, all of which are priced modestly. I’ve been using the RawDigger image analysis tool for some months now. SEND FEEDBACK Related: Expose To The Right, exposure, histogram, LibRaw, RawDigger ![]()
0 Comments
Leave a Reply. |