Nikon D850: Announcement of an Announcement of Development

[davidgp]

FYI: 20 questions about D850 answered by two Nikon technicians: https://www.dpreview.com/news/6772782345/exclusive-nikon-answers-20-popular-questions-about-the-nikon-d850

[Michael Erlewine]

FYI: 20 questions about D850 answered by two Nikon technicians: https://www.dpreview.com/news/6772782345/exclusive-nikon-answers-20-popular-questions-about-the-nikon-d850

Very comforting interview. Answers a lot of questions I was worrying about and makes this new D850 way more than just an update of the D810.

[shadowblade]

Very comforting interview. Answers a lot of questions I was worrying about and makes this new D850 way more than just an update of the D810.

The D850 is to the D810 what the 5D3 was to the 5D2 - similar name, but a completely different kind of camera. Since the previous iteration, it's become slightly better at what its predecessor was good at (the D810 and 5D2 both being slow, resolution-focused cameras, and the D850 and 5D3 improving on the resolution and base-ISO IQ to a small degree), but a whole lot better at everything else.

[Hans Kruse]

I have made a preorder on the camera. The electronic shutter is for me a single reason to upgrade. The D810 EFCS implementation was so restrictive and annoying. EFCS is really essential for shooting landscapes in low light with 70-200 lens. So I look forward to the D850.

[BernardLanguillier]

I have made a preorder on the camera. The electronic shutter is for me a single reason to upgrade. The D810 EFCS implementation was so restrictive and annoying. EFCS is really essential for shooting landscapes in low light with 70-200 lens. So I look forward to the D850.

Hans,

I believe that the implementation of the D850 is very similar, they have just gotten rid of the remaining bits of movement as far as I understand.

Cheers,
Bernard

[Jim Kasson]

I believe that diffraction will begin at large apertures for this kind of resolution in a 35mm sensor. Isn't that correct? Stopping down to a mild f11 will already be challenging, if you are planning to get the most out of the sensor.

A few relevant points to the above.

1) Diffraction is already apparent at wide f-stops with today's sensors and the best lenses. The Otus 85/1.4 shows a loss in central resolution on the a7RII when stopping down from f/2.8 to f/4.

2) Diffraction is one of the blur sources that is easiest to deal with using deconvolution sharpening.

3) Deconvolution sharpening works best when there are lots of samples.

4) False color and aliasing impede sharpening, so lots of samples are good there.

Jim

[Hans Kruse]

Hans,

I believe that the implementation of the D850 is very similar, they have just gotten rid of the remaining bits of movement as far as I understand.

Cheers,
Bernard

You can shoot continous in live view so it is rather different from the D810 where you can only use EFCS in the MUP setting.

[Jim Kasson]

You can shoot continous in live view so it is rather different from the D810 where you can only use EFCS in the MUP setting.

Great! At long last.

Jim

[Jim Kasson]

Hi Jim,

My understanding of deconvolution sharpening is limited to say the least. When you refer to samples, are these samples taken by photographing specific targets at different apertures to create the necessary profiles for the deconvolution algorithm to work?

One more question, does any of the usual photo/RAW processing software (Lr, Ps, C1) are using this method currently, or is it only implemented in more "niche" type of software.

In this context, when I say "samples", think "pixels". Each pixel is a discrete sample of the image.

Lr does some deconvolution sharpening. Other tools such as Topaz afford more flexibility.

IMHO, at this stage in the evolution of digital photographic technology, more pixels are a good thing. Lots more pixels.

Jim

[Bart_van_der_Wolf]

Hi Jim,

My understanding of deconvolution sharpening is limited to say the least. When you refer to samples, are these samples taken by photographing specific targets at different apertures to create the necessary profiles for the deconvolution algorithm to work?

Hi Raul,

As Jim explained, more pixels, a denser sampling. Since the diffraction pattern at a given aperture has a fixed diameter, more pixels/samples allows getting it to be more accurately sampled. More samples will, therefore, allow to Deconvolve with better result. 

One more question, does any of the usual photo/RAW processing software (Lr, Ps, C1) are using this method currently, or is it only implemented in more "niche" type of software.

LR and ACR/PS use a quick and dirty deconvolution method (with the detail slider at higher values) that tends to produce lots of artifacts. C1 uses deconvolution with a "Diffraction Correction" checkbox that's quite effective for Capture sharpening of Raw files.

Cheers,
Bart

[Jim Kasson]

Hi Raul,

As Jim explained, more pixels, a denser sampling. Since the diffraction pattern at a given aperture has a fixed diameter, more pixels/samples allows getting it to be more accurately sampled. More samples will, therefore, allow to Deconvolve with better result. 

LR and ACR/PS use a quick and dirty deconvolution method (with the detail slider at higher values) that tends to produce lots of artifacts. C1 uses deconvolution with a "Diffraction Correction" checkbox that's quite effective for Capture sharpening of Raw files.

Thanks for stepping in here, Bart. You are the expert on this.

jim

[uaiomex]

Hello BartvanderWolf:
With such credit as this one from someone like JK, I'll dare to ask you for some personalized advice.
I am about to be included as a participating photographer in a photo book about my city. I have several candidates including many shot with two A6000's, one converted to infrared.

The issue is that some these pictures were taken with lenses known to be sub-par in resolution like the Sony's 20mm pancake and the kit lens 16-50.
I was thinking on buying a deconvolution program to help these pictures get sharper.
In my list are Infocus, Focus Magic, Sharpen and Piccure.

Now, in your expertise, which program could be best for my goal? Will this help or will it be a waste of money?  Under $100 usd seems to me pretty acceptable if I can improve the apparent detail without artifacts that may be seen in the two-ink separation printing. The book size will be about 9X13".
Thanks so much in advanced Bart.
Best
Eduardo

 

Thanks for stepping in here, Bart. You are the expert on this.

jim

[shadowblade]

Counteracting diffraction is just one possible application of deconvolution.

With a lens profile, it can also improve a less-than-perfect lens. After all, if you know how a lens is unsharp in various parts of its image circle, you can mathematically compensate for it. It wouldn't be perfect, though, due to sample-to-sample variation. In a way, this would just be an extension of lens correction profiles already used for CA, distortion and vignetting. It wouldn't surprise me if some manufacturers (especially Sony, given their strength in electronics and weakness in optics) already use this approach in RAW conversion.

Also, it can be used to counteract camera motion. Just identify the pattern in which the camera was moved during the exposure (not to difficult on pixel analysis) then reverse that. Photoshop demonstrated it a few years ago, but I'm not sure if it's available in any software package at the moment.

For all of these, increased sampling gives more data to work with and a better result.

[Bart_van_der_Wolf]

Hello BartvanderWolf:
With such credit as this one from someone like JK, I'll dare to ask you for some personalized advice.
I am about to be included as a participating photographer in a photo book about my city. I have several candidates including many shot with two A6000's, one converted to infrared.

The issue is that some these pictures were taken with lenses known to be sub-par in resolution like the Sony's 20mm pancake and the kit lens 16-50.
I was thinking on buying a deconvolution program to help these pictures get sharper.
In my list are Infocus, Focus Magic, Sharpen and Piccure.

Now, in your expertise, which program could be best for my goal? Will this help or will it be a waste of money?  Under $100 usd seems to me pretty acceptable if I can improve the apparent detail without artifacts that may be seen in the two-ink separation printing. The book size will be about 9X13".
Thanks so much in advanced Bart.

Hi Eduardo,

IMHO, FocusMagic is the most stable performer (with good deconvolution and it doesn't exaggerate noise too much, although noisy images always create a trade-off) and it is relatively easy to use. It is a Photoshop plugin, but can also be used with a number of other applications that support PS plugins, so you're not going to be locked into a single application.

InFocus also does a good job on many occasions but is much more likely to generate artifacts. It, therefore, needs a gentle hand and more tweaking of its controls. Also, TopazLabs is now rolling out their new Studio suite, so I expect that significant development of the traditional InFocus plugin will likely stall, and there is not a full replacement of it (yet) in the Studio set of tools (although the traditional InFocus plugin can be called up from within Studio). Time will tell what the future will bring for Studio plugins, which are very much faster due to improved GPU aware architecture.

Piccure looks interesting (due to its spatially variant deconvolution) but still has me doubting to purchase it, not helped by its price tag. ColorManagement seems poorly implemented, last time I looked.

I'm not sure which "Sharpen" you are referring to, so I can't comment on that.

Overall, I think that money for FocusMagic is money well spent. It's my #1 Capture sharpener (also useful after resampling, both up and down), and on occasions it has also helped me save some blurred image detail (actually this week, when I shot some forensic type of IR shots, and I didn't quite nail the focus, despite using a correction for the longer wavelengths).

Cheers,
Bart

[BJL]

https://www.nikonusa.com/en/nikon-products/announcement-d850-dslr.page

After all this time, all of this waiting and expectation, what do we get? We get an announcement announcing the development of the Nikon D850, with no details.

. . .

Hopefully more details will be forthcoming.

As a footnote to this thread: it took just under one month to go from "Nikon is peddling evil vapor-ware by giving us partial information about a forthcoming product" to samples of that product getting into the hands of reviewers.

I will continue to say that if you do not like such early information, just ignore it. (As it is wise to do wth 99.9999% of what appears on the internet.) I see no advantage whatsoever to a company keeping a forthcoming product secret until it is ready to sell—except for the company itself, which can then keep selling the previous model at a price that customers will regret shortly aftterward when the new model is announced.

[shadowblade]

Is there any information yet as to whether this sensor uses a three-layer stacked design?

Certainly, Sony would be capable of adding that to the sensor. But it would probably be superfluous for an SLR camera, which doesn't need a fast readout rate for an EVF or AF, except for live view and video. And I'm guessing a standard, two-layer BSI design would be simpler from a manufacturing and heat management perspective.

[davidgp]

Is there any information yet as to whether this sensor uses a three-layer stacked design?

Certainly, Sony would be capable of adding that to the sensor. But it would probably be superfluous for an SLR camera, which doesn't need a fast readout rate for an EVF or AF, except for live view and video. And I'm guessing a standard, two-layer BSI design would be simpler from a manufacturing and heat management perspective.

As far as I know, no evidence so far...

I will be surprised if it is a stacked design, the only reason to use it is to get better FPS, like the A9, or less rolling shutter in video, no benefits for DR or noise... seeing that the camera is doing around 9 fps, good enough at 45 megapixels, I will say it is the typical two layer design without the RAM layer in the middle...


http://dgpfotografia.com

[uaiomex]

Thanks so much Bart for taking the time to answer. I will be re-reading your response until I buy the program (now for sure, lol).
By the way, you made it easy for me because FocusMagic was my strongest contender. The price is right and it looks simple to apply.
Best regards
Eduardo

Hi Eduardo,

IMHO, FocusMagic is the most stable performer (with good deconvolution and it doesn't exaggerate noise too much, although noisy images always create a trade-off) and it is relatively easy to use. It is a Photoshop plugin, but can also be used with a number of other applications that support PS plugins, so you're not going to be locked into a single application.

InFocus also does a good job on many occasions but is much more likely to generate artifacts. It, therefore, needs a gentle hand and more tweaking of its controls. Also, TopazLabs is now rolling out their new Studio suite, so I expect that significant development of the traditional InFocus plugin will likely stall, and there is not a full replacement of it (yet) in the Studio set of tools (although the traditional InFocus plugin can be called up from within Studio). Time will tell what the future will bring for Studio plugins, which are very much faster due to improved GPU aware architecture.

Piccure looks interesting (due to its spatially variant deconvolution) but still has me doubting to purchase it, not helped by its price tag. ColorManagement seems poorly implemented, last time I looked.

I'm not sure which "Sharpen" you are referring to, so I can't comment on that.

Overall, I think that money for FocusMagic is money well spent. It's my #1 Capture sharpener (also useful after resampling, both up and down), and on occasions it has also helped me save some blurred image detail (actually this week, when I shot some forensic type of IR shots, and I didn't quite nail the focus, despite using a correction for the longer wavelengths).

Cheers,
Bart

[Bart_van_der_Wolf]

Thanks so much Bart for taking the time to answer. I will be re-reading your response until I buy the program (now for sure, lol).
By the way, you made it easy for me because FocusMagic was my strongest contender. The price is right and it looks simple to apply.

You're welcome.

There is a trick to get the most (more resolution without artifacts) out of FocusMagic.

"Image source" set to Digital Camera or Forensic gives good results unless the image is very noisy, in that case another source setting may be required.
Click on image detail that represents the best focused area in the image, or that needs to be restored due to misfocus.
Then set the "Amount" to its maximum of 300% (don't worry it's only temporary).
Now gradually increase the "Blur Width" starting at 0.
There comes a moment where adding 1 to the Blur Width will suddenly not improve sharpness, but instead, it produces 'fatter' details and double contours. Back-off 1 on the Blur Width, and you've found the maximum Blur Width to use. Now set the Amount depending on how much you want to sharpen. Amount values between 100 and 175 are common with small Blur Width settings, maybe a bit more for larger Blur Width settings.

Sorry for being slightly off-topic, but with relatively small photosite pitch sensor arrays, the per pixel diffraction contribution will be more noticeable, and because it also offers better/denser data samples, it makes it a logical candidate for deconvolution Capture sharpening. With sensors that lack an Optical Low Pass Filter (OLPF), one needs to be careful and not over-sharpen the aliasing though. FocusMagic is just a pretty robust tool that doesn't require much tweaking, and it also does well on resampled images.

Cheers,
Bart

[uaiomex]

Thanks Bart, appreciated all the way.
I already read and marked another thread here about deconvolution apps in which you gave this same advised. Interesting way to find the optimum setup.
One last thing before I buy FM: In this thread (if I recall well), it's said that Focus Magic works better for low frequency images and that Infocus might be better for high frequencies.
In my particular case in which I mainly want an app to help boost the capture resolution and the contrast of two lesser optics, which program would work better, FM or Infocus?
They cost the same actually. IF seems to be more complex but more versatile. While FM is simpler and proved to work tho not always.
TiA
Best
Eduardo

P.S. Sorry guys for being off main topic. I promise this is my last intervention.

You're welcome.

There is a trick to get the most (more resolution without artifacts) out of FocusMagic.

"Image source" set to Digital Camera or Forensic gives good results unless the image is very noisy, in that case another source setting may be required.
Click on image detail that represents the best focused area in the image, or that needs to be restored due to misfocus.
Then set the "Amount" to its maximum of 300% (don't worry it's only temporary).
Now gradually increase the "Blur Width" starting at 0.
There comes a moment where adding 1 to the Blur Width will suddenly not improve sharpness, but instead, it produces 'fatter' details and double contours. Back-off 1 on the Blur Width, and you've found the maximum Blur Width to use. Now set the Amount depending on how much you want to sharpen. Amount values between 100 and 175 are common with small Blur Width settings, maybe a bit more for larger Blur Width settings.

Sorry for being slightly off-topic, but with relatively small photosite pitch sensor arrays, the per pixel diffraction contribution will be more noticeable, and because it also offers better/denser data samples, it makes it a logical candidate for deconvolution Capture sharpening. With sensors that lack an Optical Low Pass Filter (OLPF), one needs to be careful and not over-sharpen the aliasing though. FocusMagic is just a pretty robust tool that doesn't require much tweaking, and it also does well on resampled images.

Cheers,
Bart