Nikon D90 vs Red’s Scarlet – Specdown!

Nikon is about to ship the D90, a 12MP DSLR that shoots HD movies. Wait a minute, doesn’t that sound a lot like Red’s Scarlet?

If you aren’t familiar with the D90, check out my post on the neat little camera.

And more importantly, what will this trend mean for the industry at large?

Read on to see how the about-to-be-released Nikon D90 stacks up against the current specs for next year’s Red Scarlet.

There are bunch of factors worth comparing between the two.

First, lets start off with the very basics. But very first, keep in mind this is comparing a finalized, about to ship product with one we haven’t even seen finalized images, prototypes, specifications, etc. from. This is just a fun early look at how they’ll compare. The Red One shifted specs considerably before (and after!) it shipped, so keep in mind alllll this can change. At least half the reason why I’m writing this is to help folks get a clearer understanding of what it takes to make a usable/professional/sufficient solution, much like my earlier article on MacBook Air: Why it isn’t for Us

What They Are

The Nikon D90 is a 12MP DSLR stills camera that happens to have a video mode allowing for HD movies

The Scarlet is a “3K” (exact specs not defined) video camera with stills capability

Similar, but coming at the common grounds from opposite ends.

Ship date

-there are already unboxing pics out for the D90, which publicly is supposed to ship approximately in October last I read.

-Red’s Scarlet is set to ship in “early 2009” Red does many things well and admirably, but meeting earliest anticipated ship dates is not one of them. That is as kindly as I can put it.

WIN: D90


D90 body only is about $1000 online, with the 18-105 lens about $1300.

Scarlet will include lens, base price $3000. Based on how Red typically equips their cameras, I’m betting a shootable config will be in the $4-$8K range, depending on what you think you need. To match what the D90 comes with in terms of kit to shoot video, I’d speculate in the $3500-$4500 range

WIN: D90


From Nikon’s site:
Approx. 620 g (1 lb. 6 oz.) without battery, memory card, body cap or LCD monitor cover

Red Scarlet: none stated, but destined to be more. Considering current prototype doesn’t show a flip out LCD, expect the “necessary” accessories to add up substantially on weight as well.

WIN: D90 for lighter weight, which isn’t always a plus (for stability of shooting)


OK, I’m getting bored with too much detail here, suffice it to say the D90 is smaller

WIN: D90 for small size, but that isn’t always best

Sensor specs

OK, now we’re back on more interesting ground (to me, anyway):

Nikon D90: DX format, 23.6 x 15.8 mm, 12.9 megapixels total, 12.3MP usable, 4288×2848 max (that’s 1.70:1 aspect ratio),

Red Scarlet: Remember, this is NOT the same chip in the Red One, which is very nearly the same size and resolution as the D90’s. Scarlet’s sensor size, however, is stated stated at 2/3″, but we don’t have exact specs. Take a look at this handy cheat sheet Stu Maschwitz built, however, for a good idea of relative sizes:
ProLost: Sensor Size Cheat Sheet
Scarlet’s sensor is MUCH smaller than the D90s.
We don’t have exact pixel count, but PRESUMING 3K is horizontal, a 16:9 sensor (as Red One’s Mysterium is) should be 3072×1728. Considering they are fitting 3/4 of the pixels into a MUCH smaller area than the D90, the pixels will be smaller on the Scarlet. AOTBE, therefore, D90 has a better shot at having better low light performance and better dynamic range. Of course, all other things are rarely equal…

WIN: D90 for stills applications, depth of field, and I’m GUESSING for low light/dynamic range, but again, that is speculation based on specs, not facts

Stills specs

the D90 can shoot full quality (4288×2848) at up to 4.5 frames/sec. I don’t know if that is RAW, JPEG, or both, but it can do 4.5 fps.

We have no details on Scarlet’s stills capabilities. I HOPE they can record in RAW as well as Redcode (which is a compressed RAW)

WIN: I think it is a reasonably safe assumption, based on resolution, past history, sensor size, etc., that the D90 will take nicer stills than the Scarlet, but I’d be thrilled to be wrong


D90 is interchangeable with a DX format (about same size as Red One’s Mysterium X)

Scarlet is fixed lens of unknown quality, but still going to that smaller sensor

WIN: overall, regardless of Scarlet’s lens quality, I’d lean towards D90 due to interchangeability factor

Video specs


D90: 1280×720

Scarlet: Here’s where the Scarlet gets to stretch its legs: 3K (and probably 2K and 1K)

WIN: Scarlet

Frame rates

D90: 24 fps only – and I can’t ascertain (yet) whether that is 24.0 or 23.976 fps, which is important for making “real” video deliverables. otherwise, you get into audio sync problems, and/or dropped frames trying to sync frame rates, etc. UPDATE – sample footage is 24.0 fps – so that does create some issues, since it doesn’t divide evenly into 29.97fps the way 23.976 (aka 23.98) does. This means you can’t play it back without dropping a frame every so often while maintaining audio sync.

Scarlet: 1-120 fps in 3K, with burst mode up to 180 fps. Of course with all the fractional video rates fully supported, as well as over/undercrank for delivery at the standard frame rates, etc. etc. etc.

WIN: Scarlet, in a full-on smack down


D90: AWESOME controls for stills. Concerning controls for video – you CANNOT autofocus while shooting. Debatable as to whether you can control effective shutter angle while shooting, and Stu alludes to this in one of his posts:

ProLost: More D90

Read on to get it, but the controls are An Issue.

Scarlet: We don’t know much, but suffice it to say proper control of focus, iris, etc. are all on the To Do List, with as much manual control as they can fit in there.

WIN: D90 for stills, Scarlet for video. Duh.


D90: M-JPEG, presumptively 8 bit. This is a crucial “bit” as it were. M-JPEG is a DCT based codec – established and supported, but inefficient compared to modern codecs. Hopefully it is at least 4:2:2 not 4:2:0 color sampling after processing to M-JPEG from the RAW bayer data (more on that below). Also, what data rate are we talking about here? UPDATE: looking at the duck sample footage, data rate appears LOOOOOOOW – under 3 MB/sec (22.85 megabits/sec for the duck clip). Compare that to about 21 MB/sec from current 3K Red One as documented here.

Scarlet: Redcode, which is a wavelet based, 12 bit, compressed version of the RAW, meaning you have control over white balance and ISO AFTER the fact, which yields a FAR greater amount of malleability in post. Redcode, as Jim has stated, is one of Red’s key developments.UPDATE Based on the duck sample clips data rate, comparing that to some 3K I shot on a Red One, the Red One’s footage was about 7x higher datarate, at 12 not 8 bits, at 3K not 720p. Even though there are 5.3M pixels at 3K, and 0.92M pixels for 720p, that is only a difference of 5.7. And with 12 not 8 bits, and wavelet not DCT, and RAW not RGB….quite the advantage.

WIN: Scarlet, again by a wide margin


D90: single SD cards

Scarlet: dual CF cards, hot swappable for constant rollover shooting

WIN: SD is smaller, but CF offers better capacity/dollar, and with dual slots and infinite recording (and no 5 minute limitation as the D90 has), Scarlet wins


D90 is limited to 5 minutes at a time, claimed because of sensor overheating, but also that conveniently puts it in a bracket of still camera not video camera which affects tarriffs

Scarlet has two CF slots. Red is about to ship 16GB CF cards, so lets presume that as a standard for Scarlet. Using current Redcode 28, 3K at 16:9 at 24p, two 16GB cards would hold about 40 minutes of footage.

WIN: Scarlet


D90: while it has a large sensor which should do well in low light and have nice depth of field, it is hampered by that MJPEG codec – it won’t “bend” very far in post. Dunno about CMOS shutter roll – this isn’t something Nikon has had to deal with for video before. Most concerning, however, is the possibility of funky debayering artifacts. Bayer pattern sensors are notorious for video artifacts depending on the algorithm used to bebayer. The good algorithms involve beefy math and a lot of computation. The standard chipset on the market that most companies use doesn’t use these better algorithms, and image quality suffers. Throw on MJPEG compression on top of that….hmm.

Also, take a look at the duck sample from dpreview – look at the stones at the upper right, and the odd shimmer/judder they do. Also, the duck is way blown out, but that could be user issues/lighting too. Not much detail – I think the datarate/compression is smooshing it out pretty darn flat. And I checked – the datarate is 22.83 megabits, or about 2.85 MB/sec.

Scarlet: of course, we haven’t seen ANYTHING from this sensor, all we have to go on are specs from Red, and extrapolations based on Red One’s performance and Red’s track record to date. Based on that, however, the signs are good. Continuing the comparison to that duck shot mentioned above, with a 2.85 MB/sec datarate (with mono sound), consider a 3K Redcode shot I did on build 15 – 3K, 16:9 aspect ratio, Redcode 36 – all which should be possible with Scarlet – generated about 21 MB/sec, or about seven times the datarate. Even accounting for the difference in resolutions – 3K is about 5.7 times the resolution of 720p – and it is STILL a relatively higher datarate than the D90’s. Then include the comparison of RAW vs RGB, and wavelet vs DCT, and 12 instead of 8 bits, and you realize how much better Red’s compression will be. Redcode is a MAJOR piece of improvement over MJPEG.

WIN: Scarlet on paper, but I’ll bet money it will be in reality


After looking at some of the sample D90 movies, I look at it as a nifty gimic. An improved version of the movie mode on my pocket still camera. It shoots higher resolution but still highly compressed movies. They have better depth of field, better dynamic range, but still are limited. It’ll be great for reference material and home movies and impromptu fooling around, but it lacks the detailed controls to control getting what you want, and the compression is severely lacking to do any serious work.

So it is pretty much a foregone conculsion that the D90 will probably take nicer stills than the Scarlet, but that is unproven. The specs lean in that direction, as would common sense – a dedicated device does better at its dedicated task. Simillarly, the specs on the Scarlet imply it’ll smoke the D90 in video mode.

More interesting, however, is where this leads us, and the questions it raises:

-When Scarlet ships sometime next year (and based on Red’s shipping history, I’m guessing it’ll be second half of next year before they are out in bulk, sorry guys), will there be any competition for it in the hybrid camera market?

-will anybody else have a competitive codec, maybe something like this but with wavelet/JPEG2000 type compression? (UPDATE: duh, Mike, use yer brain – Cineform anyone?) – thanks David!

-What happens when more cameras come out like this, from companies that make more than just still cameras? Nikon doesn’t make video cameras….until now. What happens when Canon, or Sony, start making hybrids? Will the stills guys be cut loose to do the best they can? Or will they (more likely) feature impair them to not cut (too deeply) into handycam market?

If Nikon could:
-add sufficient manual controls (like shutter!)
-increase the data rate notably, or get a better codec on a chip
-increase the resolution to 1920×1080 for video
-offer a 1080i option (that’s tougher than it sounds)
-have sufficient storage
-keep the price under $2-3K

…they’d have a helluva competitive camera

And unlike Sony/Canon/etc., they wouldn’t have to worry about eating into sales of other product lines.

-What happens when features typically associated with relatively low volume products (HD camcorders) get put into products that typically sell more widely (still cameras). Product price point is strongly affected by R&D and marketing amortization – if you can sell X times as many units, that means X times as many users to amortize research, development, tooling, marketing, and other costs across. Can this notably change the marketplace? Red sure seems to think so, as evidenced by their approach.

-and will the HD camera folks keep adding better stills capabilities to compete? At what point do the two product lines merge for consumers, or will they? Will there be a firm stills camera with video capabilities, a firm video camera with stills capabilities (which usually suck, BTW), and then a squishy “hybrid” category? All boils down to cost/complexity/weight/power of the hybrids, I’d say. Devil in details, indeed.

For companies that DO have other product lines, that have other areas of expertise, I think life gets verrrrrrry interesting if you look at cameras like the Sony EX-3, and start thinking of mating those electronics (compression, storage, etc.) with a DSLR’s capapbilities.

I’d LOVE to see more cameras like that. Which, incidentally, REd is thinking along those lines too: “Mysterium ‘Monstro’ is a sensor program that pushes the envelope past anything on the horizon. It will go into Epic, and another camera aimed squarely at the DSLR market. Epic ships with Mysterium-X and has a free upgrade to Monstro.”

I think in 2-5 years, it’ll be pretty standard/no big deal to have, at least for the consumer market, a combo camera that shoots 12MP or better stills and 720p or 1080p video, recording onto some form of solid state media. And I mean REALLY common, like your soccer mom sister will have one and not think twice about it.

The pros will have some kind of RAW option, moms will have some JPEG type options for stills, and H.264 or whatever for video. NLE’s will either transcode to their preferred format (DNxHD or ProRes variant) or will gripe and add native support, which will require a pretty beefy computer.

Good times ahead, folks, good times.



Mike Curtis