Frame Averaging for Ultra Low Noise with the Nikon D850 Monochrome (but works for any camera, color or monochrome): Moots MootoX YBB 29er
Background: The Nikon D850 monochrome is a Nikon D850 with its color filter array (CFA) removed by maxmax.com. The NEF files are converted to monochrome DNG via LibRaw Monochrome2DNG and “Method B”, then processed using Adobe Camera Raw. Doing so avoids any demosaicing and thus retains full spatial resolution.
After working from 4 AM to 7 PM, I was getting bored* with lens quality control problems, and so I turned my attention to the intriguing results in Wheel and Tire. So I shot another scene and this time I verified that the Nikon D850 in-camera averaged exposure is the same as merging in Photoshop—pretty much identical.
Frame averaging takes remarkably little work compared to things like focus stacking, the main chore being taking the images, which could have been (and could be automated with a firmware update!) fully automated at high speed, but Nikon screwed up its multiple exposure mode implementation in the D850. Nikon are you paying attention? But at least Nikon offers a useful form of multiple exposure—Sony lacks multiple exposure support in the needed form AFAIK (can’t find anything in-camera or in the manual).
Indeed, Apple is at the forefront of frame averaging (one of many variants of computational photography) with Night Mode in the latest iPhone, with most all “real” camera vendors sitting with their thumbs up their backsides. They’re all gonna fail from their own lack of imagination, and the deserve it.
Includes images up to full camera resolution for single shot, camera 10-frame average, and Photoshop 10-frame average, plus enlarged crops with commentary.
The results, I suspect, are probably superior to an ultra-exotic large format monochrome camera costing $100K. Just an educated guess—a guess only, but it’s a visual treat.
My trip to the mountains has been delayed a week now, pre-flighting a crop of crappy quality control lenses—I’m getting antsy what with the first touches of snow hitting the Eastern Sierra—I want to shoot the Sony A7R IV, but remarkably, the Nikon D850m intrigues me much more as offering a breakthrough capability of unique quality potential. The change of seasons waits for no man or woman or variant**.
* While I still have some lapses and lingering side effects, my brain is working almost as well as pre-concussion again—finally, 15 months later!
** But it’s also that my perspective has changed—I want to take a big bite out of the apple of life every damn day if my mind and body will cooperate, because I don’t know if I have one day or one year or what the Simulation is gonna grant me. Good health and fitness are things I take as gifts that can disappear any time, even if I have no retirement savings, because money cannot buy the former.
Below, the Moots MootoX YBB 29er is my favorite mountain bike. With its “soft tail” no-maintenance rear suspension, it adds comfort way over a hardtail, particularly when carrying a pack of camera gear up mountains. And its titanium frame is just about indestructible with years of hard usage—and it can be bead-blasted to look brand-new all over again, if desired.
Ryan M writes:
I was excited to learn in your article that the D850 supported an AVG multi-exposure mode. Somehow I thought it only had ADD which isn't very useful to me.
I played around with it a bit this morning and found two things worth mentioning one good and one bad. On the good side, in the self timer release mode it will automatically exposure the number of images selected in multi-exposure settings. Turns out that's also true for bracketing which I also didn't know. So you can have a slow but fully automated exposure-delayed multi-image taken for excellent results without mirror slap. I don't think it's using EFCS in this mode though, so you would still need to use Qc and hold down a trigger to eliminate shutter shock.
On the bad side, my first use of multi-exposure had very poor results because the camera was in 12 bit mode and showed significant posterization in the shadows. I guess the way they are averaging the images is accumulating rounding error in the lower bits. In 14 bit mode the results are still slightly different than you'd get by averaging in Photoshop, but just barely. If I end up using multi-exposure mode I'll probably keep the individual images to average in Photoshop but the in-camera result makes for an excellent preview.
DIGLLOYD: Ryan is correct, self timer and bracketing can get the job done, and I tried them, but be sure there is a shutter delay to avoid vibration. The D850 can and does use EFC shutter if set to (M-Up mode must be used), but not a fully electronic shutter, which is a crazy limitation.
Be sure to use a remote release of course, choosing the appropriate one for your particular camera, which is ridiculously hard to figure out in some cases, with ambiguous descriptions like “for select Nikon cameras”. There are also annoying two-part ones that require use of the hot shoe for a receiver—fine for a studio but crappy for field work. I use the Nikon MC-36A with the Nikon D850 (the Vello RS-N1II Wired Remote Switch is cheap and smaller and I should probably get one). I am annoyed at having to plug it into the camera both for the hassle and the risk of affecting the camera slightly—Sony’s Sony RMT-P1BT Wireless Remote Commander is far superior in that the receiver is built into Sony cameras. Maybe I’ll get the Vello Free Wave Plus Wireless Remote Shutter Release and just live with the little receiver in the hot shoe.
The problem with self timer and/or bracketing mode is that shutter bang-bang-bang-... bang—so some shutter delay between frames is needed. You might get away with it much of the time, but you cannot count on vibration-free results and that means it is a non-starter for good shot discipline.
But much worse is that a delay of a second or longer between frames can be the kiss of death if external conditions vibrate the camera or move the subject—a half pixel or even quarter pixel movements are not OK. Plus, if there is wind, a strobe effect is ugly and there is also a strong risk of tiny vibrations in the tripod and thus the camera, so even if the shutter speed is high, individual frames might wiggle around so as to not align perfectly, which will slightly blur things.
The best way to do it but one that does not allow for in-camera averaging is to use silent shutter (full electronic) and bang out the exposures at 6 fps—no vibration, and far lower chance of subject movement or lighting changes. A full electronic shutter has its own downsides, namely subject movement and the jello effect. But if the subject matter is moving, the strobe effect is worse problem.
Upscaled to 183 megapixels
DIGLLOYD: the crops below are actual pixels from a 2X upscaling using Gigapixel AI to 183 megapixels (16576 X 11040).