This picture was taken in very low light after sunset with unsuitable equipment, aka my Nikon D5300 APS-C DSLR and the 18-105mm kit lens. A note on the hardware: I am quite attached to this camera and as it is serving me well since 11 years, even with a cracked display, I don’t see the point in replacing it even though more capable and affordable alternatives exist. I would rather buy a used 35mm lens which allows me to shoot at low light.
This shot is severely underexposed and noisy and it has some dead pixels. Let me see what you can do with modern processing software to squeeze the most out of some old hardware! I know AI denoising software exists and is quite capable, something the we did not dream of in 2013 when this camera hit the market. Feel free to incorporate the noise as a kind of grain, or to remove it if you can, and to fiddle with white balance and exposure. There is also a significant red shift in high ISO with this camera, sometimes it almost feels like the red tint of expired color film. Feel free to do with it as you please. I can also imagine very dark versions with just some leaves appearing, monochrome grainy film look, what ever
Also, this is my first play raw. I have more exciting images in stock than this one of a few leaves in the dark, but I deemed it good enough for a test ride, first because I like the faint afterglow of the leaves after sunset, and second because it represents my shooting tactic: Take the photo at a feasible handheld shutter speed (1/200 in this case), push ISO to the limit (ISO 2000 is the upper ceiling for reasonable image quality on this camera) and hope for the best in post processing.
I am not sure what the image is supposed to show — is it purely for testing noise, or is there a compositional intent?
In any case, given that you are shooting at a nominal 2000 ISO and it needs about a +6EV boost, the actual ISO is about 128000. I just did profiled denoising, then did it again on the red channel, as the image does not have a lot of red yet it was very noisy to begin with. This is just a 2-second adjustment, but it goes quite far.
Thank you for your contribution. Well, maybe I did not choose the right image to play raw, if there is no evident compositional intent, I mainly wanted to have the contrast of the dangling leaves against the dark lower third. The exposure you applied to the image makes it look too bright in my eyes, as it was quite dark at the time the image was taken and the natural lighting is not visible anymore. So with all due diligence, I prefer the underexposed version, I hope you won’t be offended by this.
What I found interesting, how did you calculate the actual ISO and why is it informative to know that it is about 128000? And my second question, the slider for exposure has a natural end at +4 EV, how did you get +6? I know this is possible with scrolling but is there an easier way?
It is informative because one can estimate how far each sensor can be pushed using simple heuristics.
I usually think like this: a current full frame sensor can be pushed to ISO 32000 without any problems using denoising current denoising algorithms. APS-C makes it 16000, the fact that it is a bit older makes it 8000.
So, you would need to exposure time and aperture to get more light. It is unclear why this shot needed a 138mm equiv focal length, which, with this lens, gives you f/5.6. If you want to capture shots like this, I would go for a bright prime, eg the 50mm f/1.8 which you can get for around 50 EUR for F-mount. Then you would have about 10 times the light, so if you expose for 1/50s you could shoot the same scene at effective 1000 ISO.
Thanks, this is quite informative! So basically, with ISO 2000, I could go about +2EV which makes it 2000\cdot 2^2=8000. And with ISO 1600, it would be more like +2.3EV. How did you come up with these numbers?
I am just looking into getting a used 35mm f/1.8 lens which would be like ~50mm ? on a full frame sensor.
If I understand correctly, the shot was taken at ISO 2000 and you boosted the exposure in post by +6 EV. Wouldn’t that make this an effective ISO of 12,800 rather than an actual? I’m not trying to be pedantic, I just want to understand what your saying.
One way to assess needed exposure is through “Exposure Value” based on lighting. According to the Wiki your scene lighting value was about 9 Ev. But your camera settings (aperture, shutter, ISO) are for 15 Ev, so you did indeed under-expose by about 6 EV.
I do a lot of bracketed exposures and with my Canon cameras find that the under exposed images are very noisy. I know that Nikon suffers this issue less than Canon, but if the leaves were not blowing in the wind the image stabilisation if available would have permitted a longer exposure which probably would have less noise. Certainly would have with my Canon. Another trick I try in low light is to avoid zooming and try to shot wide angle. This gives less camera shake so slow shutter speeds can be used and many lens then have a larger aperture to let more light in. I suspect your lens is f3.5 at 18mm and 5.6 at 105mm. Good luck. The noise in this image didn’t worry me too much. DSC_3200.NEF.xmp (12.0 KB)
Thanks for sharing! It’s always hard to bring up colour and detail without blowing up the noise in these cases. I tried to keep it blue-ish since it’s a dusk shot. (DT 5.2)
Just as an exercise, I ran the raw through DxO PureRaw 5 first and applied a similar pass. I think the colours are more natural, here, but there’s a loss of detail, too.
