Up in the mountains field shooting, writeups follow when I’m back in a week or so.
Sunil A writes:
I am a diglloyd subscriber, I immensely appreciate and admire your work. I am not a pro, but I am photography enthusiast and like the technical aspect of DSLRs, I find your blog and DAP very educational.
Since you have been using D810 for couple weeks (I am a regular blog reader), I thought to inquire about an issue I see. I have D810, Nikon 70-200mm VRII, Sigma 35mm 1.4 A and Nikon 85mm 1.8g. I was testing D810 with 70-200mm, I consistently find images darker shot at maximum aperture. In Aperture priority mode, going from F 2.8 to F 4, the shutter speed is doubled but the images are brighter. Compared to F4, F2.8 image RGB is more towards left.
Maximum aperture on a lens may not be real maximum (T Stop). So is the camera looking at marketed aperture vs T stop. And later at F4.. is the T stop and F stop same.
So do you always do exposure compensation to get the same exposure between apertures. I find your images to be exposed the same across apertures.
I found the same issue on Sigma 35mm 1.4 as well, although not to the same degree as Nikon 70-200mm. I appreciate your advice, as I am not sure if it is just my camera. I also posted a thread with images on dpreview as well.
DIGLLOYD: First, I’m assuming that “shutter speed is doubled at ƒ/4” means twice as long, so that the exposure value (EV) is equivalent, e.g., ƒ/2.8 @ 1/500 vs 1/250 @ ƒ/4.
At full aperture, several effects are at play. I regularly compensate for this by giving 1/3 stop to 1/2 stop or so more exposure at full aperture for some lenses (particularly wide angles), though I usually reverse that boost when doing a series. I do so to give better exposure to the peripheral areas. The amount needed ranges from almost no compensation to nearly 2/3 of a stop with certain wide angle designs. How much depends on the lens design and the particular sensor. Across an aperture series, it is common to see fluctuations of 1/10 or 1/20 stop from other factors (e.g. diaphragm and shutter speed accuracy); I normally correct this kind of normal variation for presentation.
Vignetting (field illumination) is a multi-factorial behavior; primary losses quickly improve even one stop down. But this does not account for center brightness. See also Ray Angle, Vignetting, Color Shading on a Digital Sensor.
Sometimes there is plain old “cheating”; a nominal f/2.8 might really be f/2.9 or f/3 but vendors prefer to state a nominal value. Sometimes this derives from rounding the focal length figure, e.g., 50mm might be 48mm or 52mm, or 200mm might be 180mm (!). Up to 10% is considered “acceptable” (by whom?). But it might also be rounding off the true diameter of the lens diaphragm. Since f-stop = focal / entrance pupil diameter (a ratio), judicious rounding can go a long way when it sounds better for sales. Basically, conventional numbers sound better; no vendor wants to advertise an ƒ/2.9 lens.
Then there is transmission (T-stop) which can reduce whatever brightness is actually there (see What are F-stop and T-stop? in Making Sharp Images). T-stop is typically only 1/10 stop less than f-stop Zeiss ZF.2 lenses, but few other vendors say.
Then there is the loss due to digital sensor technology and peripheral ray angle, which can be up to half a stop for some f/1.2 lens designs and remains very significant at ƒ/1.4.
Many vendors silently adjust for their own brand lenses and/or chipped lenses from others. For example, Canon silently gains-up the f/1.2 lenses to compensate! This can be seen by inserting paper between the lens contacts and camera body; the camera not recognizing the lens won’t make the silent boost (compare with/without). Many point and shoots and other cameras (e.g., Fujifilm X) just build in correction for their own lenses (and also vignetting correction on top of that), whereas in reality the true brightness might be significantly less than advertised. Hence it's rather silly to see “lens tests” for vignetting with some cameras where the vignetting has been taken out by the camera already. Such tests are system evaluations, not optical.