You wrote compressed vs. uncompressed. I think you meant lossy vs. lossless.
It depends on the model of camera. Lossy RAW can be quite bad (Sony) or indistinguishable from lossless for photographic purposes (Nikon). Lossy compression algorithms can change between models (Panasonic).
You are 100% right and I agree with you. That's not the point though. We all want decent conversation and reciprocity. I also do think about these discussions like about some kind of transaction. At least in civility, understandable and some very, very basic (hint of) gratitude. I am not helping these undeserving individuals. But that is where their punishment is supposed to stop, in the name of holding meaningful public discussion.
What is not right, is devoting pages of text to not getting that response you require. it is so much unecessary, and brings no value to the forum.
Yes, that is spot on in terms of clipping unimportant highlights.
But in this case a couple of small patches of white clouds were being clipped. I didn't notice it originally but normally I don't like clipping clouds and so 2/3 of a stop extra exposure* is what I would have applied.
As I mentioned earlier, those histograms were just quick and simple demos to prove the benefit of using raw histograms if someone really wanted to get the maximum image quality.
* exposure - amount of light striking the sensor per unit area while the shutter is open
** optimal exposure - the maximum exposure* within dof and motion blur requirements without clipping important highlights.
*** under exposed - more exposure* could have been added with the DOF and blur constraints still being met without clipping important highlights.
Maybe others are not reading the significance into EV0 that you are. If you really want to know where the raws clip at any given ISO setting on your camera, just blow out an image in all color channels and load it into RD. The clipped values will pile up high on the Y axis. You can double-check any raw histogram, too, to see if it is subtracting raw black levels by loading a lens cap photo. If a histogram does not place black at 0, then the values are not linear, and a stop isn't a stop. No matter what you think a tool is doing with the settings that you are using, it is always good to double-check by throwing a clipped frame and a black frame at it.
People who calibrate sensors and lightmeters do care. And that would be much more than 1% of our users. To avoid clipping, it's good to know how the meter is calibrated.
For the purpose of determining just clipping FastRawViewer is much faster.
Is that a dark field histogram? If so, crop to 600x600, turn off black point subtraction, set the x-axis to linear and the bin size to one, and repost. The standard deviation is the rms value of the noise.
Possibly this is down to a common misunderstanding about what 'EV' means. If you think 'EV' means scene luminance, then it seems 'intutively obvious' that 0 EV would mean absolute darkness, and then having it in the middle of the scale would seem a bit strange. Possibly Don's failure to recognise an answer is a case of prior false learning obstructing his understanding.
Looks to me me like the rms read noise is about 3 least significant bits, or maybe a bit more. I don't know your camera, but that kind of read noise is not unusual at base ISO for Canon cameras. Sonys and Fujis are better, in my limited experience with Canons.
Raw data pre-processing, sometimes even vignetting compensation (one of the reasons I look at limited size, often 256 by 256, slightly off-centre samples, about 100 pixels to the left (and/or to the right) of the sensor center, and prefer to use shots taken without a lens or at least with the lens stopped down fully).
All good points. I forgot to mention that it's best to make images of the back side of the body cap, and offsetting the crop is a good way to avoid artifacts from stitched sensors.