DSC_7146.NEF.xmp (11.3 KB)
1 Like
Used RT here, I also don’t suppress all the noise.
Spots removed with GIMP
DSC_7146_RT.jpg.out.pp3 (14.4 KB)
1 Like
Hello Gustavo, I own a Nikon D7000 as well, for years, and it is in no way a noisy camera. I think the noise in your picture is caused by the way how you exposed this scene. Loading your nef in Art shows a very dark picture - way too dark in my opinion. Compensating dark pictures always amplifies noise.
My default exposure compensation setting is -1/3 EV and I don’t use that ETTR technique, just plain old aperture priority. My photos show a bit of noise at standard ISO (100), mainly in the skies, which is easily corrected in any raw converter, just a slight touch of NR will do.
So don’t sell your D7000, it’s really a great camera!
4 Likes
Still have my D7000, intend to use it as a backup with long telephoto for my railroad walkabouts. It does present noise in low light situations, but I’ve found it to be easily correctable. I think its performance is precisely what ETTR is about, exposing to put highlights as high as possible in order to pull more of the scene out of the shadows. I think your +1/3 EV technique is spot-on, especially when one can’t spend quality time doing metering of a scene.
1 Like
I believe you mean +1/3 EV, right?
I think that’s what I’m doing: I have exposure compensation set to +1 EV. Then I spot meter the brightest area of the scene, then reframe and shoot.
But I’ll revise the histograms to see if I’m doing anything wrong. I was used to a Canon 5D and it is just a couple of days I started using the D7k.
Thanks anyway for the advice on this model, and I’m glad to know it is worth to keep it.
Dust on the entrance aperture of the lens should not be visible at all (it is just vignetting and scattering light), dust on the detector is projected sharp. Dust inside is a different issue.
Hermann-Josef
Small aperture shots are very sensitive to lens dust.
Not right! I expose minus 1/3 of a stop by default, in order to avoid blown-out highlights. Using raws I am practically always able to keep those highlights (think white clouds, white clothes) detailed and not washed-out.
So you meter the brightest area using spot. That means that those areas will be well exposed, and you add 1EV to make the scene a bit brighter. That also means that you don’t care about the shadow parts in your scene. In my view this is the very reason why your original photo turns out quite underexposed when opening in Art with a neutral profile!
Using spot-metering is quite a tricky thing, because you must be very well aware what you are metering and how that compares to the rest of the scene.
A thing I forgot to mention in my first reaction was that the Expose-to-the-Right technique implies that you do that while shooting and not in post-processing. So take a shot, look at your histogram and change exposure compensation until the histogram touches nearly the right side.
I nearly always use matrix or center meting with named minus 1/3EV to get good results.
2 Likes
Ouch, I missed that. Metering on the brightest spot does necessitate -EV…
I tried that briefly a while back, as with all of the ETTR pet tricks i tried it got in the way of considering composition, my first priority. The closest I came to success was mid-point metering with +EV, which I think is what @gadolf is thinking about, but even then with, say, +1EV, you can still blow highlights in scenes with large differences in illumination or reflection.
My new camera has a “highlight-weighted matrix” mode, which is oriented to the JPEG exposure so it typically leaves 1 or so stops of headroom, but the camera is also so-called “ISO-invariant”, so the shadows can usually be lifted aggressively without much if any noise penalty. I’m a lazy sort, so that’s working well for me right now… 
Light is noisy. The most perfect camera in the world would still make photos with noise.
ETTR would help here. If the camera had been opened up by one stop, almost no pixels would have become saturated.
Personally, I don’t mind noise when it occurs in the shadows, or on textured surfaces like wood or concrete. But this image has a very dark smooth surfaces as prominent features, looming monsters dwarfing the human-scale but miniature cars and people. The noise on the big dark building spoils the effect.
That building is about 9 stops down from the brightest part of the image. We can’t expect much better than that from a 14-bit camera. So the question when taking the photo would be: do we blow out the sky, or get good noise-free detail in the dark building? If we want both good sky and good dark building, we would need multiple exposures and HDR, or wait for better light.
From that single image, I would be tempted to aggressively remove noise from the building, and replace the sky with a looming storm.
1 Like
Here is where I think I’m making a mistake: if I measure light in the most bright spot of the scene, then that spot will be middle grey and everything else will be underexposed. So per the histogram (before applying any modules), I understand that I have room to compensate +2EV (say, 1.7EV to stay safe):
EDIT: The arrow in the middle of the scene points to where I think I spot metered it, but maybe it was the cloud above, not 100% sure.
EDIT2: Actually I think I should compensate +3EV, since that histogram is from an image which was shot with +1EV
Now thinking about it, I think here’s the most difficult part of ETTR. If the dynamic range is very big, even the +2EV compensation might still leave the shadows very dark.
Agreed, although usually I don’t look at the histogram, once I have already determined the average compensation value that works most of the time. Then, it’s as easy as aiming the brightest spot, lock exposure, recomposing and shooting. I kind of get used to it (although I agree I loose some seconds on this)
This way you just compose, meter and shoot, right?
@snibgo I liked your interpretation of the scene! I work at that corner and always got some kind of feeling that matches, now I see, your description. And a good reminder that sometimes blowing out highlights might be an option. In other words, to always be aware of trade offs we must make in some scenes. This is something I have to practice a lot, no doubt, and it means training the look, more than anything else. Thanks for that.
You took the shot at ISO 640, then raised the exposure by 2.5 EV (-1EV applied by darktable to counter the in-camera +1EV, then you raised it by +3.5EV), ending up at ‘effective ISO’ of 640 * 2^2.5 = 3620 for a daylight shot. Maybe a wider (faster) aperture (instead of f/11) would have given enough DoF, given the wide angle lens.
You applied contrast EQ and sharpen, which both raise noise levels if you’re not careful.
1 Like
This is probably why there’s no traction among the camera manufacturers to support real ETTR - it’s nigh well impossible to mechanically know the tone curve needed for all the possible DRs, so they can’t deploy relevant in-camera processing. I have highlight-weighted metering in the new camera and I use it, and know that my data has unused headroom. But I can easily tone-curve it to perception, with little to no noise accentuation. Lazy way to do it, but it works well for walk-about grab-shooting…
1 Like
EDIT: I think the pixls website isn’t offering a large version of that image, so the glorious gloom can’t be seen in gorgeous detail. So I’ve uploaded the large JPG to http://snibgo.com/imforums/7146_comp.jpg .
12 Likes
Thanks for keeping the creative spirit alive @snibgo!
I miss the adventurous takes and by extension @chroma_ghost RIP.
1 Like
I didn’t read 100% of this thread, so my apologies in case I took the too early exit.
My 2ct
If you are that new to a cam, you shouldn’t think that fast to ditch it again. But learn the new beast and where it differs from the former beast
AFAIK Canon is stronger in preserving highlights, where Nikon is stronger in preserving shadows.
That being said, ETTR implies “as much as possible without ruining wanted highlights”
I shoot several bodies, one is D7100. My way of shooting often does not allow me to take quality time. Hence I have -0.7EV as std. and matrix measurement. If I’m afraid of blown highlights, I further decrease exp. comp and take a second shot. That usually suffice.
Besides denoise wavelet auto as @Jade_NL explains is also my way of handling and it is pretty sufficient on the even W-A-Y more noisy OMD Em5 mk2
yep, +3EV is usually about right if you want to take this sort of spot-metering approach.
1 Like
And finally my take. Thanks for that nice picture
DSC_7146.NEF.xmp (76,9 KB)
(sorry 3rd revision, I should really finish first)
Hope you like it
2 Likes
Yeah, I over reacted, definitely 
. And I completely agree on that. Maybe the drama comes from the fact that it was me who probably bricked the 5D, against all odds 
Still digesting it.
It took me around a year to feel at ease while shooting with the 5D (and it was the ettr/spot meter way 99% of the time) so I know I’ll have a learning curve in front of me.
And thanks for sharing your usual shooting mode.
I do!