Or did Nikon deliver?
As we wait for Nikon's late August announcements (D90, 18-105mm DX lens, and more ;~), perhaps it's time to take a step back and ask what the ultimate pro Nikon DSLR should look like right now.
I'm not going to go all futuristic on you and specify something that probably couldn't have been made today (e.g. 48mp, post Bayer, no noise, solar-powered, focus anywhere), but rather what I think should have been in the queue for our current time frame. In order to do that, I have to put myself into the Wayback Machine and have Professor Peabody take me back to my mindset of 2005, as that's about the right place to be to see what Nikon was capable of (D2x) and give us enough design cycle time to engineer all the fixes we need to make the "perfect" 2008 model year DSLR. So I've popped a bottle of vintage Diet Coke (2005 edition), placed my D2x on my lap, pulled up a copy of Mac OS X 10.4, and have taken myself mentally back to the time when Karl Rove was first revealed to have outted Valerie Plame, and...
- It's the sensor stupid. Anyone who's used a D2x has to know that the thing that is least perfect about that camera is the sensor. Specifically, it just needs to work better at higher ISO levels. It wouldn't hurt to be less constrained by diffraction. And 12mp isn't quite enough to max out the biggest Epson desktop inkjet. These things all point me towards an FX sensor. I can't cram more photosites in and deal with diffraction in the DX space, so I'm immediately thinking FX. This also helps with the higher ISO level, so for a US$4995 camera expected to ship in late 2008, I think this is a no brainer. FX it is. Now for the finesse decisions. Canon's already at 16.6mp and there are rumors they're working on a followup. I could be chasing pixel counts if I'm not careful. My target is 5700x3900 (which would fill a 13x19" print), which is 22mp. But that puts me at 6.3 micron photosites, not very far off where I am with the D2x. Getting good ISO improvement without increasing the area of the photo diode is going to be next to impossible, so compromises are necessary. I'm going to go with 5200x3460, which puts me about 7 microns and 18mp. Diffraction is going to "fall off the cliff" at f/13 or so, which is okay, if not great; at least it's better than the D2x. I've got 13 square microns more light capture potential per photosite (a one-third increase). In order to really push this design over the edge and get me to a tight 13x19" print, I'm going to remove the AA filtration. Any moire that occurs is going to be embedded down at 300 dpi and not really visible. I'm not at the 9 microns of the D2h sensor, so I'm going to have to really push the sensor guys for design optimizations to meet my ISO goals: better microlenses, less power and data line coverage, less read noise, and so on.
What Thom Designed: 18mp, better high ISO, less diffraction.
What we got from Nikon: same 12mp, but better than expected high ISO; diffraction not really an issue.
What people expect on August 27th: 24mp, better than D2x but about D300 level high ISO performance; diffraction will be the same as D2x.
- Performance-enhancing drugs. 5 fps (8 with a cropped frame) is nothing to skoff at, but it's not at the upper end of the performance spectrum. Buffer sizes in the teens to thirties for most useful settings/sizes doesn't exactly cut it, either. With NEFs we can outrun the D2x buffer in three seconds, not exactly a "pro-like" characteristic. The shutter lag and viewfinder blackout are state-of-the-art, though. So my primary focus is going to be on goosing the frame rates and increasing the buffer. Frame rate really has to hit 8 fps. I'd like it to hit 10. I know my focus system works at 8 fps, so the question is whether or not I can do anything that will improve that (that already state-of-the-art viewfinder blackout time starts to loom as a barrier). I've been looking at the proposed final specifications for UDMA and that looks like it can sustain JPEG bursts writing to the card if I can keep the rest of the pipeline from being a bottleneck, so I'm going to target no buffer restraint when shooting JPEGs. NEFs are another problem due to the large size of the file, so I'm going to fall back to what impressed me most when I first handled a D2h: long continuous NEF bursts. Three seconds isn't what I'd call long, so I'm going to double that. At 8 fps, that's 48 frames of goodly size. UDMA will handle perhaps half the speed improvement I need, but it looks like I'll need to double the RAM in the camera to get the rest. RAM always goes down in price, so snuggle up to the feeding trough.
What Thom Designed: 8 fps, unlimited JPEG buffer, six seconds NEF buffer.
What we got from Nikon: 9 fps (11 without focus), almost unlimited JPEG buffer, same teens NEF limit (now doubled with RAM upgrade).
What people expect on August 27th: 5 fps with 8 fps DX crop, buffers back to where we were with the D2x.
- Body building. The current D2x body is perfectly fine, there's nothing I can identify that's functionally wrong with it. We might want to tweak some control placements and feels, maybe even think a little harder about how many buttons we have on the camera, where, and why. But that LCD is my first priority. 18mp is a lot of image quality to put on that dinky screen. So I'm going push as big and high quality as I can with the color LCD. Indeed, I think I'm going to give up small rear status LCD, too. Can we get 4" on there? Certainly 3", but I'd really like to push higher. I want those pros to have no doubts when they're chimping their shots, so I'm going to go with 3.5", and have the dedicated buttons (ISO, QUAL, WB) have current status shown at the bottom of that large LCD instead of having another LCD down there. The Direction pad needs a lot of work, and if we're going to allow a center press, why don't we just use those center buttons we have on the Coolpix? All buttons have to have a unique feel so they can be distinguished solely by touch. It's like we started to do that with a few buttons on the D2x but didn't really complete the thought. Some people are complaining about size and weight, so I see only two paths: use the old F4 different grips approach, or have two variants of the same body, one with an integrated grip, one with an optional grip. I like that last idea, actually, as it gives us more SKUs on the same basic product while giving us a clear answer to the complaints: we'll have a Sr. and Jr. version, with pretty much everything else identical.
What Thom Designed: Two bodies, main difference the added grip. Lots of button tweaking. Large 3.5" LCD.
What we got from Nikon: Two bodies, main difference the added grip. A bit of button tweaking. Large 3" LCD.
What people expect on August 27th: Two more bodies, main difference the grip. No more button tweaking. Same 3" LCD.
- Control freak. Are we really up to 50 custom settings? Five menus, two of which have four banks? And 20+ setup items? How do we really expect users to stay fully in control? The camera's menu system is getting more sprawl than my waistline. And why isn't there a way that a user can switch completely between two different sets of settings across the entire camera (and preferably four)? The menu system needs work. First, we need a unified "bank" system and an easy way to switch between it. This one's a little beyond me in terms of simplifying, so I'm going to enlist some of old UI guru friends (I've mentioned Alan Cooper before and I'm sure to call him, but Tony Hoeber was good at this stuff, too. I'm going to call all the old UI designers I've worked with and get this thing settled. If that requires a touch screen or some other hardware aspect, we need to know about that now so we can get it into the test mules.
What Thom Designed: A unified, simplified, and better menu heirarchy, with the ability for the user to switch "sets" of everything with a single action.
What we got from Nikon: No real change. We did eventually get INFO button access to setting many options, which is a small start.
What people expect on August 27th: No change.
- Bells and whistles. This is a new generation of camera, and everyone expects we Nikon engineers to push technology as far as it can be pushed and to dazzle them with something unexpected. We're already putting a lot of new technology in, but it's mostly invisible (sensor, UDMA, etc.). We need some visible Holy Cows! that will get people talking just like they do about Macs and iPods. Some sort of "see what the sensor sees" system sounds like one candidate. Direct storage connection seems like another (and gives us another SKU to sell, the ProWalker drives). Geotagging looks like it's going to get popular, so let's put the GPS in the camera.
What Thom Designed: Live View, GPS, direct storage connection.
What we got from Nikon: Live View.
What people expect on August 27th: nothing new.
- Bam Pow Flash! How many times have I seen someone write that they switched to our system because of the flash? That i-TTL thing is working pretty well, but it has some warts. I don't like the preflash sequence being where it is. That extra lag is okay for many casual situations, especially for our consumer customers, but the pros want more. FV Lock isn't good enough. I think we need a new approach for when the flash is on the camera. We know the focus distance, which is likely the subject, right? We know the powers the flash can produce. We know the aperture. So why the heck can't we use all that together in precalculations? After all, with flash exposure APERTURE = GN / DISTANCE. So it won't work if the photographers want to bounce or diffuse. We'll just sell a special bounce/diffuser that changes the calculations and make more money off them! For straight ahead flash they can get straight-on precalc and we can even give them a balance setting (more ambient or more flash). We also need a way to set off wireless flashes without visible light; time to do our own radio wireless trigger (and hey, we can sell new flashes! Dang, I'm on fire with adding SKUs to increase the bottom line and hook that user forever with accessories).
What Thom Designed: A way around preflash. Radio wireless trigger.
What we got from Nikon: Same old, same old.
What people expect on August 27th: Same old, same old.
- Focus on the objective. The only thing wrong with the current focus system is that it has big gaps in it (though that will be mitigated somewhat by the change to FX). Dare we leave it as is? Yes, I think we can concentrate our technology elsewhere. Still, I know my colleages are going to be arguing that "every new generation of pro body needs a new system." I'll have to out-argue them.
What Thom Designed: Same old, same old.
What we got from Nikon: A new, arguably better (in most respects) system.
What people expect on August 27th: nothing new.
Okay, so there's the basics. You've got my best shot at what I would have been thinking if I were a Nikon designer back when the current pro bodies were being thought through. We know what Nikon gave us (D3/D700). Now, everyone is expecting something additional (and different) on August 27th. Will Nikon deliver? For the most part, they have so far. Check back for an update next week when all is revealed... |
