I think that many people set up the highlight warnings and then take some test shot increasing the EV in small increments and then come up with a fairly good guide of how far they can push the DNR of their camera…maybe this changes for different lenses not sure but it seems like a robust and quick way to be able to work quickly without too much fiddling…
Thank you…
My photos were vacation photos - often on the go with few sec. time to adjust…
Didn’t have time to properly assess.
I think this is what happens indeed. I will try to turn OFF D+ then manual adjust to slightly under expose -1/3 to -2/3 to deal with the highlights.
1 Like
I’d hope that once you get all the wacky jpeg-first stuff disabled that you’ll find that you generally have +1/2-1stop of headroom from the jpeg. That’s how its been for the other systems I’ve shot on.
2 Likes
One item not discussed is the metering modes on the camera. Those will have a big effect on the exposure compensation.
I recommend switching your camera to M (manual mode). Keep the iso at 400 and learn how to expose on your camera. Keep it there for a month or more. Correlate to darktable and then eventually switch to A (aperture). If you do this, you won’t be looking at camera histograms.
My answer to this problem is to set the camera to auto exposure bracketing of 1EV. With continuous drive the camera quickly gets the three shots and this covers me in most situations. Also reduces the chance of a person having their eyes closed because of blinking. Camera sounds more professional as well so the BS factor makes you look like a better photographer.
1 Like
I’ve done this quite a bit. Have 9k photos to sort 
My information regarding Fuji cameras, differs from yours: as I understand it the ‘L’ histogram is directly taken from the display/JPG data but the R.G.B. histograms are from the RAW data.
I follow the advice from AP and allow the camera to establish the exposure ‘pivot-point’ (mid gray). My Fuji underexposes this value by a full (protective) stop, which I reverse automatically in processing. The only other exposure correction that I make is through the camera’s compensation control which again is automatically reversed as a part of the exposure preset.
In general I find that only very rarely do I need to mess further with exposure during the development process.
I believe that this way my exposure/development pivots around the 18.45% and filmic takes care of the tone mapping as needed. Using ETTR as a basis for exposure appears to me counter-productive with this ‘modern’ system.
Closer to the raw data given what gets disabled but still not a true raw histogram I don’t think…Fuji Natural Live View | Exposing to the Right | Fuji Frame
I am not aware of any system (sans Magic Lantern) that provides raw histogram data. I welcome specific information to the contrary, as I’d sell all other gear immediately and buy into that system.
I made a small test using my Canon EOS 850D. That model provides the modes D+ and D2+ to avoid clipping.
Indeed the image is exposed at ISO 1 EV or so lower and subsequently the dark parts are processed to look like the set ISO. So far so good … It looks like this processing is baked into the raw. If I turn off almost all moduls and look at the histogram of the same scene shot at different modes. There is a difference between the “normal” more and the D-modes histograms.
Would it be too much to ask if you can include the following?
- the histograms from DT in linear form (in order to match the camera)
- the histograms from the LCD of the camera (in order to match to the one from DT)
What is the black/white point for each picture? And could you upload raw samples?
Given the shape of the histogram, I’d agree.
If the RBG histograms are a true direct readout of the sensor, you’d expect the values to be bunched to the left.
Otoh, if there is a log transform done on the sensor data, is it still a raw histogram? Though even in that case, I’d expect a difference in shape between the L and the RGB histograms (no S-curve on the raw histogram).
Also, I’d like to see an authoritive reference documenting those “raw histograms” (the mentions I found were just “reflect the red, green and blue pixels of the sensor”, which is not the same as “show the sensor values”)
Sure, please find
-
the raw files
20220825_100133-IMG_0804.CR3 (25.8 MB)
20220825_100140-IMG_0805.CR3 (22.8 MB)
20220825_100145-IMG_0806.CR3 (22.8 MB) -
dt histograms (linear)
-
camera (histograms)
The raw white points are (dt |exif specular | exif normal):
*.804 |14338 | 13035 | 11448
*.805 (D+) | 11606 | 11435 | 11606
*.806 (D2+) | 11606 | 11435 | 11606
Looks like DT reads the WP correctly.
For me the message is to be careful with using the modes because something is happening very early in the pipeline, most probably before saving the raw file to the memory card.
1 Like
Thanks, I see one thing wrong. With my 6D, HTP 200(D+) is ISO 100, not ISO 200. HTP ISO 200 (D+) should be compared with ISO 100.
I don’t know about (D2+) and I will not be able to check your raw files before mid September.
That’s quite a big difference 14338 vs. 11448 ?
Thank you! Very interesting results and relation to the histograms. This modes (D+ and D2+) appear to be indeed quite aggressive. But the camera LCD is far from painting the whole picture.
Appreciate the comprehensive test and files!
I know this is not darktable, but rawproc will display a histogram corresponding to the image at the state of the tool selected for display. I’ve found this very helpful for following the state of the image from tool to tool in the chain. Also, it makes it very evident which data is the source for the histogram construction. And, any step in the tool chain can be selected for display, whether it looks good or not. But, by selecting the input image at the top of the chain, that allows for looking at the histogram of the raw data right out of the file, with no processing.
I bring this up because it’s very important to know the exact place in the tool chain from which the histogram is constructed, in any software…
3 Likes