The D850 Blog


Vello Grip Again

From a reader of my previous blog message:

"I am happy to share that with some tricks & hacks now I am hitting that magical 9 fps. It was quite clear from your article that bottleneck is loose contact. I removed top cover on battery grip & connected camera without top cover and it easily started hitting 9 fps even when it was half tightened. This proved that somehow shape of top surface cover is not exactly confirming to camera bottom plate contour. Moreover, portion of battery grip with contact points has little bit of play. Both these issues stopped contacts on camera & battery grip to be firmly locked. I opened top cover of battery grip & placed small piece of plastic beneath this contact point assembly as a "packing material" to stop it wobble & also it gets positioned little upward towards camera contacts with this. Further, there is a thin projected portion on top surface right above "Vello" logo. I felt this to be a hindrance for grip to make deeper enough contact with camera & it had no functional role to play. I filed out that projected portion to make it flat enough. Voila..!!! I am hitting 9 fps without any issue now with Vello grip, Nikon BL-5 & third party EN-EL18 battery."

I just tried something similar on my Vello grip, and yes, it appears that it is the way the Vello conforms to the bottom of the D850 that's the issue. 

Vello BG-N19

Updated again: it's still more complex than it first seemed

B&H sent me a Vello BG-N19 vertical grip to try with the Nikon D850 [disclosure: B&H is this site's exclusive advertiser; some of the product links in this post direct to their site].

bythom vello d850grip

I've done some preliminary testing with it, and it really seems as near a clone to Nikon's own MB-D18 Battery Pack as you're likely to find. If anything, on my samples the Vello's thumbstick press-switch has a little better feedback than the Nikon one, which is mushy. Nikon's Command dials seem a little more like the D850 camera bodies, while Vello's seem to have a less Nikon-like loose feel to them.

The Nikon grip seems to have a bit more heft to it. But visually and in feel otherwise, essentially the two grips appear identical.

What everyone wants to know is whether the dollar problem has been solved. 

What dollar problem? Nikon's full kit to provide 9 fps on the D850 costs US$950 (MB-D18, EN-EL18b, MH-26A charger, BL-5 door). With the Vello, you can buy the Vello for US$100, the door for US$25, and if you shop around, third party EN-18 batteries for US$50 or so. The difficult bit has been the charger, though there are now affordable chargers for the EN-18 batteries you need (keep reading).

So does it or doesn't it? Does the Vello hit 9 fps, that is.

Yes and no. With an official Nikon EN-EL18a battery I found no difference between performance with the Vello BG-N19 and the Nikon MB-D18: both seemed to let the camera hit 9 fps. Update: as it turns out, firmware has nothing to do with it. After a great deal of close examination and continued testing, I find this: if I press hard upwards under where the electrical connection is made on the grip, the Vello does 9 fps on all my EN-18 batteries, regardless of manufacturer. If I stop pressing upwards, I get 7 fps. Clearly, there's a deep-seating contact issue involved here. 

One reason why this was difficult to track down was the camera was originally secured on a tripod when I first tested, then tested later handheld. When handheld, there's never enough "force" on my sample to insure the proper electrical connection. When on the tripod and gravity is working, there often is, thus I was getting intermittent results. When I step in and push hard to pull the grip and body together more at that point, the connection is always made that allows 9 fps. 

No, tightening the thumb screw that goes into the tripod socket doesn't do it. I can't tighten it enough to guarantee 9 fps. This has been one of my issues with the add-on grips—including Nikon's own—for a long time: build quality and eventually rough handling can make for less than perfect connections, and then troubles brew.

If you touch the electrical connector on the grip gently, you can see that on both the Nikon MB-D18 and the Vello BG-N19, there is a little bit of "play"—that connector moves a bit on both models, probably to ensure that it goes into the camera's slot no matter how you start to align the camera and grip when putting the grip on. Left/right, forward/back movement is okay, but any up/down movement basically leads me to the not quite secure connection that precludes 9 fps. 

If you want to know what you're going to get with a grip/battery combo, there's a shortcut: set the Shooting Method indicator to CH and then press the INFO button. You'll see either 7 fps or 9 fps towards the lower left on the rear LCD. I've found that indicator to be true. 

So what's my official position at the moment? It's hybrid [affiliate links embedded]. I've found that the third party batteries from DTSE and Wasabi work okay in the Nikon MB-D18 and a properly working Vello:

The battery and charger are the easy part: Nikon uses standard cells in the EN-EL18 batteries, so it's pretty easy for third parties to stick such cells into a case and imitate the Nikon battery. The hard part is the information connector into the camera (or grip), and it seems that everyone has reverse engineered that fine.

In my D4/D5 shooting, I have about half Nikon batteries, half third party batteries in my bag. I've never encountered a glitch from any of them. Ditto with the third party chargers. The only thing I'd note is that you're not paying for build quality on those chargers. But at those prices, you can afford to have to replace the charger should it break.

The tougher part is the grip itself. 

I'll be honest, I just don't like the Nikon MB grips, Nikon made or third party. Over many years of using and encountering students and others using them, they just are too prone to cause issues over time with rough handling. I've seen far too many cases where a "jiggle" causes (or fixes!) a problem. And the Vello sample I received is just another confirmation of that. 

Moreover, with a grip "locked on" to the body as tight as I dare tighten it, I can still, with a bit of force, move the grip/body relationship slightly. Indeed, I've noticed that the grip is always a weak point if you're mounting your tripod to it for landscape shooting at some shutter speeds. (Put another way: if you're shooting with first curtain electronic shutter, mirror up, and other stability bits to try to get rock solid pixels, don't shoot with an MB connected to a tripod.)

But here's the thing: at least Nikon's MB's have been consistent over the years. I've never seen one fail out of the box, only after continued use in the field. I can't say that's true of all the third-party versions I've used over the years. I've had about half of them fail out of the box in some way. 

The Vello that B&H provided me took an ever so slight twist to get aligned to my D850 the first time. I think that's because one of the plastic spikes that go into alignment holes on the camera body is ever so slightly tilted on my sample. I use the word "think" in that last sentence because I have no tool that can measure such a small difference; I'm trusting my eyes on this one.

That said, on the camera I see nothing to differentiate between Nikon's grip and a properly working Vello. I can find no performance difference and other than the small control feel differences there's nothing else to distinguish them in use that I can see. 

So, perhaps I'd put it this way: if you're only going to use the grip for 7 fps, then definitely consider the Vello. It'll save you substantive money. If you're going to have the grip on the camera all the time or absolutely need 9 fps, I'd tend to opt for Nikon's simply because of my experience with previous third party grips. At least you'd have someone to investigate any issues that arise with your use of the grip (e.g. Nikon repair). 

Now, do you really want 9 fps? ;~)

I ask that because the buffer takes a pretty hefty hit at 9 fps. Here's what happened at 7 fps and 9 fps second by second in terms of buffer with my usual settings with my fastest XQD card:

  • 7 fps — start 27, 20, 17, 14, 11, 8, 5, 2, buffer full (51 frames total)
  • 9 fps — start 27, 18, 13, 8, 3, buffer full (38 frames total)

If you don't have a fast card, I've seen hits as big as 60-70%.

Sure, 45mp for almost five seconds is pretty remarkable. But remember that if you're in situations where you're shooting a lot of bursts in sequence, you eventually won't get those five seconds. 

For example, shooting football, I didn't have a problem with the buffer at 9 fps. Shooting soccer, I do. Why? Because there was generally 30-45 seconds between plays in football (and often more), but soccer is more continuous in action, and at times, frenzied continuous in action. The camera was still writing to the card when I tried to pick up a new action sequence in a soccer game, and thus I wasn't starting with a full 27 frames free.

This article was updated to reflect continued testing.

D850 Firmware Update

Firmware version 1.01 includes three fixes:

  • Long Exposure NR no longer produces increase noise or color cast when set.
  • Apertures didn't fully reset quick enough with non-E lenses when shutter speeds under 1/10 was used.
  • After cleaning the sensor via a My Menu choice, the user was returned to the SETUP menu instead of My Menu. 


XQD Cards

Sadly, B&H is now listing the Lexar XQD cards as discontinued. As a reminder, Micron—the owner of Lexar—decided to get out of lower margin consumer products business and sold off their Lexar brand to a Chinese company (the same one that markets PNY cards). We're still waiting to see the results of that change of ownership, but it may be awhile, as Lexar was using Micron chips and it's unclear what the new owner will do. Most likely, they'll try to find lower cost sources, which means redesigns, which takes time.

B&H isn't the only one with no Lexar XQD inventory. Amazon no longer seems to be selling them direct, and the number of "other sellers" seems to be dwindling as inventories get sold off. This is putting pressure on the inventory of Sony XQD cards temporarily, though at the moment most sources seem to be getting resupplied quickly. 

Only a few cameras use XQD cards directly: Nikon D5, D500, and D850, plus Sony AX1 and FS7. Sony also has other professional video gear that can use XQD cards in an adapter. I believe that's enough gear to keep XQD on the market for the foreseeable future. The problem is that your choice now is basically:

  • 32GB, 64GB, 128GB, 256GB Sony G Series
  • 32GB, 64GB, 128GB Sony M Series

You want G series on the D850 (and D5 and D500) if you want to preserve buffer performance. The M series is between 20% and 40% the speed of the G series in write performance, and yes, it makes a measurable difference on the Nikon bodies I've tried them on.

More About Lenses

I've been trying to come up with a generalized way to think about lens quality and how that relates to the increasing pixel counts, as that question keeps coming up over and over again as I try to explain why I think certain lenses are more "optimal" for the D850 than others. 

First, let me repeat what I've written for decades: more sampling is always better. If you take a picture with a compromised lens like the 18-200mm DX on a low pixel count camera and then repeat that with a high pixel count camera, you're better off with the high pixel count camera. The difference is that problems that might have been masked with the low pixel count now become more obvious. The lens looked "good" at 6mp, but it looks only "fair" at 20/24mp.

Overall, there are two primary things that I see over and over as pixel density increases. 

First up is absolute resolution. Most lenses have enough absolute resolution in the center stopped down a stop or two to look really good on a D850. The difference between a lens with more absolute resolving power and one with minimal resolving power is a bit like looking at a D800 versus a D800E image: acuity appears to be higher when you pixel peep with the D800E image because there's nothing blurring edges/detail (e.g. AA filter of the D800). 

The problem is that many people mistake absolute resolution (acuity) with visual sharpness. What happens is that they crank up the values on their raw converter (or in camera Picture Control) and they get what they perceive to be nice sharp, defined edges. The problem is that those edges are polluted with contrast adjustments. 

Those of us with trained eyes can usually immediately see the difference between innate sharpness (great lens) and contrived sharpness (contrast changes added on edges after the fact). It is visually different. And it has to do with how detail/contrast is rendered across all pixels as opposed to some at edge boundaries. 

A really great lens doesn't need a lot of sharpening in post. A really poor lens may need a great deal of sharpening in post, and the end result just doesn't look the same.

The second aspect I see is related to what I call smearing or smudging. You've probably seen this when you pixel peeped in corners. What the heck is going on there? The detail just disappears into ugly blur. And often blur with abnormal characteristics.

With lenses that have astigmatisms and coma in the corners, here's a test you can do to see the problem more clearly: use a projected laser dot, or set a small LED to resolve at distance and then shoot those in those corners. These light sources should be points (or maybe small circles if you have megapixels galore). But they're often not. Instead, they spread and smear and even change shape from round to oblong or worse. 

So what happens when something that should be a point blurs into a stretched oblong object? Well, if you don't have a lot of megapixels, you don't really detect that blur very well. Indeed, because of the way demosaic routines do their work by looking at a range of pixels, sometimes that plays into things, too, and you just get some pixel blurring. 

But at 45.4mp FX? Yes, we're going to actually see the multi-pixel blurs/smears/smudges in the corners if they're present, and you're going to see their shape and size. You don't see that on a 12mp FX camera. 

So, when I say that a lens is D850-worthy, I'm suggesting that these two problems are not present, ignorable, or minimal. Particularly compared to other optics you could choose of the same focal length(s), where those problems will be present, un-ignorable, and in many cases spreading over multiple pixels. 

I'll be the first one to admit that my lens list is subjective. That's because so many factors enter into the analysis (e.g. different subjects, different subject distances, different apertures, different focal lengths on zooms). What you tolerate in terms of blurring and smearing may not be what I'll tolerate. 

At some point you have to be critical of why you're buying a high pixel density body with lots of pixels, though. What do you really gain by riding the Pixel Express? Well, not a lot if you try to use compromised lenses. That makes the other aspects of the D850 more important to add value—and they may for you—than the pixel count. 

From Hot to Not

Ah, the Internet. First, new tech on announcement gets the pile-on-hype effect, where the product can do no wrong. Eventually the product gets shipped and the measurebators get all upset when their expectations of fantastical, imagined gains aren't met. 

The D850 has gone from Hot to Not in less than two months. At least if we're to believe the Internet. In particular, two things seem to have dulled the enthusiasm, with a third pushing those on the fence off. 

  1. It doesn't have the D5 Autofocus. This all seems to stem from one test done with what appears to be 3D Tracking. I'm still testing, so I can't say anything definitive at the moment. I suspect that there are subtle changes due to the differing viewfinder blackout times of the D5 and D850. But as many of you know, I've been pushing the D850 into areas you wouldn't normally expect it to be, such as sidelines sports. I can tell you that it works perfectly fine, though you may need to pay some attention to your settings and adjust accordingly.
  2. It's not a stop better than the D810. In actual measurements, I'd say that is completely true. But it also wasn't something I was expecting to be true. We don't have that big a generational gain in sensors any more. But there's more to this than just a measurement. Let me describe it this way: same shot, same situation, same settings, in every case I've preferred the D850 files to the D810 files. Every case. It's a subtle thing, and I'm not sure that I can yet ascribe it to a specific change Nikon made. I'm still digging (thanks Iliah, for RawDigger). But the difference is there, though small enough that some might miss it. But then again, I now have 45mp instead of 36mp and I've got a small gain? How can I not like that?
  3. Nikon's greedy. The third item tends to be something to do with the features. The lack of built-in flash. The absurd cost of gaining 2 fps. The second slot being SD. The power-hungry and still broken SnapBridge. Yeah, I hear you. These things can be disappointments on a US$3300 camera. 

Ironically, the more I shoot with the D850, the more I like it and the hotter I get about it. Exactly the opposite of the Internet. 

I think the conclusion in my review of the D850 is going to be very similar to my conclusion about the D810. On paper and in actual measured testing, it doesn't seem like the bar has been moved all that much (D800->D810, or D810->D850). In reality, the sum of the parts add up to a far better experience in doing the things that you want the camera to do. Moreover, the camera now does more things than it used to. For example, the D850 is more than a passable pro DX camera, it's nearly equivalent to the D500 with only frame rate being lower. The D810 wasn't quite that. 

I'm nearly prepared to say that Nikon has done the right thing by the D850 iteration: they've now iterated the best all-around camera you could get twice, and in both iterations they've moved the bar considerably forward. Which is going to keep it at the top of my best all-around camera list. 

This camera should still be hot. Don't believe everything the dogs are writing on the Internet.

Nikon's Recommended Lenses, Maybe

Hmm. Not sure what's going on here, but dpreview re-published a list of Nikon's recommended lenses for the D850 from SLR Lounge, and then amended that list a few hours later when NikonUSA seems to have woken up from a slumber. 

First the amended list (original list items shown with *).

Primes

  • 14mm f/2.8D#
  • 19mm f/4E PC-E*
  • 20mm f/1.8G*
  • 24mm f/1.4G
  • 24mm f/1.8G
  • 24mm f/3.5D PC-E
  • 28mm f/1.8G
  • 35mm f/1.4G
  • 45mm f/2.8D PC-E
  • 58mm f/1.4G
  • 85mm f/1.4G
  • 85mm f/2.8D PC-E
  • 105mm f/1.4E*
  • 105mm f/2.8G Micro-Nikkor
  • 200mm f/2*
  • 300mm f/2.8G
  • 300mm f/4D
  • 300mm f/4E
  • 400mm f/2.8E*
  • 500mm f/4E
  • 600mm f/4E
  • 800mm f/4E

Zooms

  • 14-24mm f/2.8G*
  • 16-35mm f/4G
  • 16-80mm f/2.8-4E#
  • 17-35mm f/2.8D#
  • 24-70mm f/2.8E*
  • 24-120mm f/4G#
  • 70-200mm f/2.8E*
  • 70-200mm f/4G
  • 80-400mm f/4.5-5.6G*#
  • 200-400mm f/4G

NikonUSA appears to be totally asleep at the wheel. The 28mm f/1.4E, a very good lens that I just reviewed, isn't on their "final" list but was on the initial list. Likewise, the 8-15mm f/3.5-4.5E was on the initial list but not on the final list. I'm also not sure why the 85mm f/1.8G wouldn't be on that list, other than the fact that NikonUSA really wants to sell the f/1.4 version. Someone is just winging these "lists" off the top of their head is my guess.

You may have noticed some # marks in the list. As far as I'm concerned, those lenses probably do not belong on a D850 optimal lens list:

  • 14mm f/2.8: High degrees of CA, high vignetting, poor corners with clear smearing, poor contrast wide open, most of the zooms with this focal length actually are clearly better (e.g. 14-24mm f/2.8). This is a lens that's needed a redesign for over a decade.
  • 16-80mm f/2.8-4E: Yes, a DX lens that Nikon wants to sell more of is on the D850 list for some reason. It's a very good lens. At DX crop. Why you'd have that in your D850 kit I have no idea. 
  • 17-35mm f/2.8D: Guess Nikon still has some stock of this lens they need to sell. Pretty much all the same complaints as with the 14mm, but with the additional problem that it is pretty poor at the 35mm end. Another lens that's needed a redesign for over a decade.
  • 24-120mm f/4G: Another lens Nikon wants to sell you because they make them in volume. But it was struggling with the D800/D810 and things don't get better with more resolution. I'm very tempted to say that you're far better off buying the 24-85mm f/3.5-4.5, and you'll appreciate the smaller and lighter build. The 24-120mm f/4G is one of the lenses that I've been considering dropping from my list. I'm pretty close to doing so after having looked at a few more images.
  • 80-400mm f/4.5-5.6G: Again, the long end is just weak, and it easily shows on the D850. This is another lens on my "considering dropping" list. Unlike the 24-120mm, the 80-400mm has some redeeming qualities that keep it more in contention, particularly its performance from 80-200mm, and even out to 300mm. I don't think I'll drop it from my list, I'll just add a comment that you'll want better at 400mm.  

In more than a month of shooting now, I've seen nothing that's changing my mind about my best lenses for the D850 list. Indeed, my opinions are solidifying the more I shoot. The only lens I'm still not completely sure of is the 8-15mm f/3.5-4.5, which is currently in my test bag, so I should have an answer soon.

Update: Nikon Rumors found a document on the Japanese Nikon site that lists their recommended lenses for the D850. And yes, it disagrees with the dpreview list in the following ways: no 14mm, added the 28mm f/1.4E, added the other f/1.8 primes, added the 60mm f/2.8 Micro-Nikkor, added the two DC-Nikkors (105mm and 135mm f/2D), added the older 500mm f/4G, added the 8-15mm f/3.5-4.5, added the older 24-70mm and 70-200mm f/2.8G, and added the 200-500mm f/5.6E. It also omits the 17-35mm f/2.8 and 16-80mm f/2.8-4E DX ;~).  So I was right, NikonUSA seems asleep at the wheel again. 

I'm sticking by my list. 

Greg Benz's Suggestions for the D850

I don't usually do this, but my friend Greg Benz has posted some good starting suggestions for using the D850, and I thought I'd cross comment on them. Greg's choices are quite good and very defendable, but I have a few points to make you might want to also consider.

  • In RGB Histogram image review, remember that you need to then set the highlights display via the thumbnail button+pad. Greg's right, you only need this one view to tell you what's going on with your pixels, but the default is to only display the histogram; you have to add the highlights review.
  • If I have any real questions about Greg's settings, it has to do with Auto WB, Flat Picture Control, and AdobeRGB Color Space. What he's trying to do here is get a better histogram that more closely reflects the raw data. Sorry, but that's not a magic combo. I'm not sure that there is a truly magic combo you can set in camera, but the problem is that the raws are underexposed to the JPEGs (especially with the usual Picture Controls), and the WB twist to the red and blue channels can be a second gotcha. Realistically, you need UniWB with this camera (yeah, working on it). 
  • AF limiting: for landscape shooting, these are rarely useful, as Greg implies. What is useful? Single Point and Group. Single Point gives you pinpoint focus control, Group gives you Closest Focus Priority across a small area. 
  • Greg doesn't mention that CSM #D6 appears to also apply to Quiet Mode shooting. That can be useful if you're not on a tripod.
  • I'm not yet sure about his recommendation of 3 for the focus shift. Thing is, Nikon has embedded a CoC calculation into their steps. The actual steps vary with focal length and aperture based upon that CoC. I think I agree with him that 5—which everyone else is suggesting—probably isn't the right starting value to try for landscapes, it's something lower. That's because Nikon appears to be using a not very good CoC value (probably the same 0.033 as they mark the lenses, not 0.025 as the Zeiss calculations would suggest it should be, or 0.016 as some photo site calculations would suggest). Thus, when you step you have to be wary of slight DOF softness at the front/aft of the focal plane. Step too much and you'll get bad data in the merge. Step too little and it'll take you far too long to stitch the focus in post. Nikon hasn't helped us find the right middle ground, and I believe it to be different for macro work than for large landscape work.
  • As I've noted in my work on AF Fine Tune, don't average the results, select the median. 
  • Yes, ISO 64 is where you want to be. Up to ISO 200 is okay, and with those ISO values set 14-bit Lossless Compressed. At ISO 400 and above, set 12-bit Lossless Compressed.
  • I can't quite agree with his loss-of-resolution off tripod comments. Pixel smear happens over time, thus a 1 pixel smear isn't actually a halving of total resolution. True, smear is not a good thing to have in your pixels, but at low levels I wouldn't worry about it, and deconvolution sharpening can help you recover from very low levels of that type of handling error.
  • Which aperture to use is still something I'm trying to come to grips with. Greg suggests f/8 to f/13. But f/8 is already diffraction limited on a D850. f/13 results are absolutely impacted by diffraction. The tradeoff here is acuity versus depth of field: you can't have both past f/5.6 on the D850. This is why focus shift (ugh, Nikon, what were you thinking using a derogatory term already used in optical evaluation?) becomes very important. Technically, to preserve everything the lens and camera can capture, you want to be at f/4 or f/5.6 and focus stacking to get your depth of field. And as noted above, Nikon didn't make it easy to figure out what the optimal settings to do that would be.

Are You Ready for Some Football?

I’m in testing mode at the moment. I’ve got four different cameras and a dozen lenses sitting in the review queue. That means I’m out and active doing a lot of different types of shooting.

This weekend I took the D850 (and a Sony A9) with me to the University of Colorado football game against University of Northern Colorado. This gave me a chance to evaluate the speed aspect of the D850.

I’m not going to go into great detail, but let’s run through just one sequence of shots on a single play and see what comments that generates from me.

bythom us co CU versus UNC football 9-17-2017 42261


This is a punt reception towards the end of the game. Not exactly the way you want to catch the ball, #21. I’m at the far opposite end of the field with a 300mm lens. But I’ve got a 45.4mp camera, so this is about a 15mp crop that would work just fine for most uses. Thing is, I’m at least 60 yards from the action, so what we really want to know is what the focus is going to do. Right here, it’s exactly where I put it and dead on.

bythom us co CU versus UNC football 9-17-2017 42262

 
Still need to get a good grip on the football, dude. I’m firing at 7 fps in this sequence. One thing I can say is that the “slowest card wins” thing is definitely present. For a while I was shooting NEF on the XQD and JPEG on the SD slot. But that limits the buffer to what the SD can manage, and it isn’t as good as the XQD can do. I eventually just pulled the SD card and shot XQD only.

bythom us co CU versus UNC football 9-17-2017 42263


Okay, we’ve got our hands on the ball now. Even though I’m panning across a busy background, the focus is holding right where I want it. (I’m at f/4, giving me a bit of depth of field, but I’m going by where I see the focus plane: it’s where I asked it to be.)

bythom us co CU versus UNC football 9-17-2017 42264


Is #44 ever going to get there? 

Meanwhile, let’s talk about image quality. I’m shooting at ISO 800 here, which is keeping my shutter so high that when the sun occasionally popped out I maxed out the shutter speed at 1/8000. One thing I noticed while converting these images was that I was sharpening them but pulling out Color Noise Reduction in the Adobe converter. There’s no noise reduction in these images, and there’s no real visible noise at the pixel peeping level even though those black jerseys are a good place to generate it. I’ve tried far higher ISO values, and so far I’m pleased with what I’m seeing. 

bythom us co CU versus UNC football 9-17-2017 42265


Everything still going as expected. Let’s widen my crop some for the next shot:

bythom us co CU versus UNC football 9-17-2017 42266


Focus is still right on the punt returner, I think on his left wrist. Note that the background has been getting busier and busier, giving the camera plenty of opportunity to lose tracking. I’m pretty sure my focus sensor is on dark, by the way. Dark arm or the black part of the jersey. 

bythom us co CU versus UNC football 9-17-2017 42267
bythom us co CU versus UNC football 9-17-2017 42268
bythom us co CU versus UNC football 9-17-2017 42269
bythom us co CU versus UNC football 9-17-2017 42270


And finally the answer to our question: does #44 ever get there?

bythom us co CU versus UNC football 9-17-2017 42271


Okay, so a few points:

  • Focus is basically like a D5. Very reliable, great tracking. The one thing that Nikon did that compromises that a teeny bit has to do with the joystick: you can’t push it in, get AF-On and a new autofocus area mode, and still move the cursor. On the D5 you can move the cursor while you’re holding the joystick button in, on the D850 you can’t. Fine, I can live with that, but it’s still a bit of a disappointment. I can see that compromise on the lower priced D500, but the D850?
  • Exposure seems to be very much like the D5/D500: Nikon built in a fair amount of highlight latitude on raw images. So even though -0.3EV exposure compensation was still showing some blinkies and was pushing JPEG exposures about as far as you’d want to go, I shot much of the game at +0.3EV and could still recover all the highlights in the raw files.
  • Detail is incredible. Put an incredible lens on the front of the camera and you’ll get incredible resolution. The sequence above is mostly about 15mp or so images cropped from the full frame. And very credible ones at that. I’d put these up against the 16mp D4 images any day. But with the D4 I’d have needed an 800mm+ lens to get these crops from my position. I didn’t just shoot sports this weekend, and I can say that the full frame images of landscapes and natural scenes I’ve peeked at also look excellent. Definitely a modest but still measurable step up from the D810.
  • Noise handling looks to be quite good. At base ISO we’ve got the D810’s wide dynamic range. But even as I press upwards into the ISO values I tend to use, I’m very pleased with what I see so far. I’m going to hold off judgment against other cameras at specific high ISO values until I have more data and more time to analyze images, but this looks like typical Nikon: as the megapixels went up, that didn’t break anything. Indeed, my first assessment is that the D850 goes far further than the D810 before you get that annoying magenta cast at ridiculous ISO values. If you can tolerate noise, you’re not getting that tint and contrast build-up the way you did with the 36mp sensor cameras.

mRaw and sRaw are Raw, but...

My friend Iliah Borg sent me a message about the D850 medium and small mode raw files and how he thinks they’re created. I have to admit, I hadn’t yet gotten around to looking at anything other than full size images. After all, I’ve got my hands full trying to figure out some basic things about the D850.

My apologies to Nikon. I was wrong when I wrote on launch that they were still using the old “near JPEG” style of smaller raw images pioneered by Canon. It’s clear that Nikon is now doing something different, but it’s also clear that something isn’t quite as we expected.

Visually, the smaller raw sizes just don’t look like they have the “right” acuity. At the pixel peeping level there’s a roughness to them I wasn’t expecting. 

So here’s what seems to happen in the D850: the camera is doing a double re-sampling. And using more pixels than the large size uses! This actually may be the source of the 45.7mp number (45,749,760 is the actual calculation of the total pixels generated initially).

Normally when you shoot at NEF (RAW) Large you get 8256 x 5504 pixel images. When you shoot in NEF though, the camera actually generates 8288 x 5520 pixels. Interesting. 

Okay, so what happens with those 45.7mp pixels when you shoot smaller NEF (RAW) sizes?

A double resampling, that’s what. 8288 x 5520 appears to be resampled to get 7104 x 4728 data, then resampled again to get 6192 x 4128 pixels for NEF (RAW) Medium. For Small, 8288 x 5520 becomes 6216 x 4136 then 4128 x 2752 on the second resample. So we get a 1.15x division followed by a 1.15x division (though the rounding is different for each, so they’re not perfectly equal). This results in a 1.32 reduction in file size: 41.5MB > 30MB > 21.9MB. 

This leads me to something I hadn’t caught before: if you choose Medium and Small for raw images, you will be switched to 12-bit Lossless Compressed; you have no choice of bit depth or compression. 

Quite obviously, someone has discovered an interesting mathematical formula you can apply to data that can get you something akin to new 12-bit Bayer data at a different size. What that is I can’t yet find a patent or description for. 

The strange thing is that the data generated is back in Bayer form, not the double reduction. It’s not clear to me yet if it stayed Bayer all the time (doubtful), or that it was demosaiced into intermediary pixels, then remosaiced back to Bayer in the final pass. There’s some strange resampling routines going on in the camera I haven’t seen or heard of before.

But it also explains why the first time I looked closely at a NEF (RAW) Medium image it seemed just a bit odd to me. Edge acuity changed in some way I can’t yet describe. It just looks a bit fuzzier than I’d expect. I also haven’t yet managed to do enough testing to determine whether white balance coefficients are used at any step in the process.  

I’m sure Iliah is doing more testing. As will I. This is a very interesting discovery, and something we haven’t seem before from anyone else.

Now, the question is whether you’d rather have a 25.6mp 12-bit lossless compressed image that’s 30MB in size or a 45.4mp 12-bit compressed image that’s 34.2MB in size. First approximation: the latter, especially since your buffer just got cut in half. But I need to do much more testing...


text and images © 2018 Thom Hogan
portions Copyright 1999-2017 Thom Hogan-- All Rights Reserved
Follow us on Twitter: @bythom, hashtags #bythom, #dslrbodies