Nikon D3 First Impressions

This is a quick review of the Nikon D3, the first full-frame Nikon DSLR,* from the point of an amateur/hobbyist. The D3 has been around since 2007, but at around $5,000, was priced beyond the reach of most amateurs.  With the introduction of the latest generation of Nikon full-frame DSLRs, the D3 has for the first time come within reach of a greater number of hobbyists.

*Not counting the Kodak DCS-14n, which is technically the first full frame DSLR with a Nikon F-mount. :)


BACKGROUND
For the longest time, I had wanted to upgrade to a full frame DSLR.  I like to study photos by photographers I admire and I have often seen photos with a very shallow depth of field that I found mesmerizing.  My friend and co-author mshafik had once dissuaded me from getting a full-frame camera, providing me with several examples that showed that even with a relatively inexpensive telephoto lens, a crop sensor camera was capable of producing images with a similarly shallow depth of field.

For a while, I was content with using my Sigma 50-150 to get portraits with shallow DOF, although I had pretty much decided on getting a full frame and was closely monitoring D700 prices.

At one point, I became impatient enough at the stubbornly high prices for the D700 that I started looking into switching to Canon, which has less expensive full frame cameras.  Even the original 5D seemed tempting to me.  However when I calculated the cost of switching systems, including the loss on selling gear, it was just as expensive as buying the D700.

This past January, Nikon finally announced the D4 and a few weeks later, the D800.  Not long after that, Nikon also announced a $500 price drop for a new D700.  I was thankful that I had sold my D300 just a few days before the D4 was announced.  Now, I was salivating at a realistic opportunity to upgrade to a full frame.  Surely the price of a used D700 would drop to about the same price as a high-end APS-C such as the D300S?

I certainly found some D700s that were sold for just a little more than a new D300S, but they were quickly snapped up.  There were plenty of other D700s that were being offered but they were around $2000 or a bit less, which to me seemed very unappealing, given that the price of a new D700 was officially $2200.  On the other hand, the stocks of the D700 at $2200 never seemed to materialize, and used D700 prices stayed high.

Meanwhile, I started looking into getting a used Nikon D3.  It turns out that the price of a D3 had decreased significantly.  Whereas a new Nikon D3 used to be almost double the price of a new D700, a used D3 was now just a little more than the price of a used D700.  Moreover, when the cost of an MB-D10 battery grip is factored into the cost of a D700, the price is almost the same.  So, although a D3 is slightly more expensive, I thought it was a better value and got one instead of a D700.

D3 VS. D700 KEY DIFFERENCES
Here are the biggest advantages of the D3 over the D700.

• Much more ruggedly-built, as in this example and this.
• 100% viewfinder.
• Longer life shutter (rated for 300,000 instead of 150,000)
• Built in vertical grip (with a vertical shutter, customizable AF-ON button, and command dials)
• Dual memory card slots.  The second card can be used as overflow or for backup.  For me, this is the biggest advantage of the D3 (I have had a corrupted memory card before when I took my Exilim skiing, and if I take on a project for a client, I would want to have the safety net of a backup card).
• Longer battery life (27.75 Wh vs. 11.1 Wh).
• a D3 is smaller than a D700+MB-D10 combo
• Continuous shutter: 9 fps vs. 5 fps (8 fps with grip)
• Voice memo.  I thought I would use this a lot but I haven't used it much.
• Possibly slower depreciation when bought used compared to a D700.

On the other hand, the D700 does have some advantages over the D3:

• Popup flash with commander.   I had used my popup flash a lot and had always chosen bodies that had built-in commander capability.  However, lately I have been using a TTL cord more often than the popup flash and I now have more than one wireless-capable speedlight, so losing the popup flash is unfortunate but not disabling.
• Self-cleaning sensor.  It seems my D300's self-cleaning sensor wasn't that effective, so this one is not a big deal.

NOW WHAT?
Having gotten a D3, did it live up to my expectations?

There were only two reasons I wanted to get a full frame camera: a shallower depth of field and better high ISO.  These advantages are discussed further in the Full Frame FAQ.

Depth of field
In the real world, I believe the difference in depth of field between an APS-C camera and FF is not immense.  It's not like the DOF difference between a point-and-shoot vs. an APS-C camera, which is easily noticeable even by non-photographers.  Between an APS-C vs. FF, the difference in DOF when using lenses with equivalent field of view (e.g. 75mm for FF vs. 50mm for APS-C) should be around 1.3 stops (i.e., the FF aperture would have to be stopped down by 1.3 stops to yield a similar DOF).  In my tests, the difference seems to be a little more than that but I will have to do some more tests to confirm (click to zoom in):

APS-C at 50mm, f/2.8

Full frame at 75mm, f/5.0

In real world shots, the difference in DOF is immediately noticeable to me.  I think it's because I had gotten used to seeing the shots from my APS-C cameras.  In any case, I was not disappointed.  In the shot below for example, the focal length was 31mm (about 20mm APS-C) and the subject wasn't super close, yet the DOF is noticeably shallow (aperture was f/2.8).

High ISO Noise

The other reason I wanted to upgrade to FF was because of the high ISO capability.  The ability to shoot high ISO helps me use a higher shutter speed to help make my shots sharper.  It's also useful for bouncing flash at farther distances.

This wasn't as significant of a reason as the shallow DOF because technology is improving all the time and the current generation of DSLRs such as the Nikon D7000 have exceptionally low noise.  According to DXO, for example, the low light ISO noise of the D7000 (30db signal-to-noise at 1167 ISO) is close to that of the first generation of FF DSLRs such as the Canon 5D Mark I (1368 ISO).

The D3 was a significant leap from my previous camera, the D300, but it can't do miracles.  The D3 certainly can provide low noise even at very high ISOs.  However, in shadows, luminance noise is quite visible at high ISOs, especially when those shadows are lifted in post-processing.

In the shot below, I used my Sigma 10-20 on the D3 and used 12,800 ISO.  It looks very usable to me even though the noise here is essentially magnified because of the DX crop (the Sigma 10-20 is a DX-only lens).

12,800 ISO

I took a shot of my family in the same scene at the same ISO.  However, unlike the display, we were mostly in the shadow and I had to lift the shadows in post.  Noise is apparent on those previously shaded areas, even without zooming in.  The noise here can't be easily removed by noise reduction, unless the NR is so aggressive that the subjects look plastic.  In my opinion the noise is sufficiently distracting that shots under these conditions would be usable only for personal albums.

12,800 ISO

In general, when selecting the ISO for the D3, I would observe whether any important compositional element is in shadow.  If nothing important is in shadow then I don't mind using 6400 ISO or if necessary 12,800 ISO.  If on the other hand, something important is in shadow, I would probably limit myself to 3200 ISO or I would use flash.  In either case, it's a significant difference from the D300, when I would usually limit myself to 1600 ISO or when necessary, 3200 ISO.

OTHER DIFFERENCES
Coming from the D300, I actually didn't find many other differences with the D3.  I would say the D300 has almost all of the features of the D3.  Here are my observations:

Autofocus.  The D3's autofocus is not magical the way Ken Rockwell makes it sound, but it is reasonably good.  When I try to focus in what I consider to be challenging conditions - very thin depth of field plus rapidly moving subject, the focus is not always on the perfect spot but is close enough that the shot is usable.  In the shot below, for example, the camera focused on the far eye instead of the near eye (or was not able to focus fast enough before the subject moved), but in my opinion the focus is acceptable.

105mm, f/2.8

It also seems that the D3 has more reliable continuous autofocus, although I don't have the D300 anymore to do a side-by-side comparison.

Continuous AF.  Again, the focus isn't perfect but it's close enough so that the shot is usable.

One big difference between the D3 and D300 is that the D3 can autofocus in really low light even without an AF-assist light.  I find this useful, because an AF assist lamp can often cause subjects to blink.

The D3 can focus well in dim light even without an AF assist light. 

Handling.  One feature I really like in the D3 is that it's very easy to check the focus.  When reviewing images, the middle button in the directional pad can set to zoom in at 100% at the part of the image where the camera was focusing.  I can quickly determine if the D3 picked the wrong spot to focus on, or if the spot was correct but the focus was not accurate enough.

I also like the ease of selecting AF points.  The D300 has the same 51 AF points that the D3 has but selecting the right AF point required quickly tapping the directional pad.  With the D3, using the directional pad is more similar to controlling a joystick.  Whereas with the D300 I would sometimes switch to 11 AF points for speed in selecting the AF point, I feel no such need with the D3 and always keep it in 51 AF points.

I am less enthusiastic about the D3's dedicated buttons for ISO, Quality, and White Balance.  I prefer the approach used in Nikon's smaller cameras such as the D300 and D700 where ISO, Quality and White Balance were assigned to the buttons on the side of the LCD where they performed other functions (e.g. zooming in or out) depending on the context.  On the other hand, I find the extra LCD display below the main LCD screen to be useful.  It helps me see the ISO more conveniently than looking at the top LCD display.

Battery Life.  One of the useful features of the D3 is the battery life, which is rated at 4,200 shots.  However, in real life,  it seems that with the way I use it (always shooting in raw, reviewing the images frequently), I get only about 1,200 shots or so from the battery.  I've only exhausted the battery once (after a few days' worth of shooting) so I might get a better estimate with future use.


Size and Weight.  One thing that I find a bit off-putting with the D3 is the size.  In the few weeks that I've had the D3, I still haven't gotten used to it.  All my previous DSLRs were about the same size.  Even the D700 is about the same size as the D300.  On the other hand, the D3 is much larger and heavier.  It feels like a phone book.  It's large enough that it feels awkward in casual settings, like watering your lawn with a fire hose or lighting a birthday cake with a flamethrower.  Anyway, it's probably just in my head and I'll get over it.  (My brother-in-law used his D3S everywhere and it didn't seem unusual to me.)

BEST OF BOTH WORLDS

If you have been using DX lenses like I have, you may be concerned about the cost of upgrading to full frame lenses.  The Nikon triumvirate (14-24, 24-70, and 70-200 VR II) and 1.4G primes (24, 35 and 85) are indeed quite expensive.  I don't think I could afford to cover all the focal lengths I want with these lenses.  However, I don't believe there is a need to.

First, there are decent alternatives to some of those lenses.  In particular, a reasonable alternative to the 24-70 2.8G ($1889) is its predecessor, the 28-70 2.8 AF-S (around $1100).  I will be reviewing that lens as well as another constant 2.8 lens with a one-of-a-kind focal length range, the Tamron 28-105 2.8.

As for a fast telephoto zoom, there is no real alternative to the Nikon 70-200 VR II, except perhaps the Sigma 70-200 OS, which is not quite as sharp as the Nikon 70-200.  This brings me to the second point, which is that I don't plan to abandon the APS-C format.  Instead, I will continue to use my APS-C Fuji S5 and Nikon D70 to cover wide angles and telephoto focal lengths, using the Sigma 10-20 and Sigma 50-150 respectively.

With ultrawide angles, I won't really have a shallow depth of field anyway so I don't see a need to use full frame.  In fact, I would probably want a deep depth of field in most cases, making an APS-C camera a good alternative.  With respect to longer focal lengths, the depth of field will be shallow in any case.  Indeed, a 70-200 2.8 on a full frame might have a DoF that is too shallow.  On the other hand, the 50-150 on APS-C can render the background as a blur while still keeping the subject sharp.  See here.  BTW having a second or third APS-C camera is possible at a very reasonable cost.  An older APS-C camera is not expensive (there are used D70s on ebay for around $200 or sometimes even less, and even a Nikon D3100 which has a relatively current sensor can be bought on ebay for around $350 or less).

RELATED POSTS
Canon 5DII First Impressions
Full frame DSLR FAQ

MORE SAMPLES
Here are a few more samples from the D3, all shot with the Tamron 28-105 at f/2.8 except the last sample.

Is the bokeh smooth enough for you?  Review coming up on the lens I used for this.

CLOSING REMARKS

• If you'd like to upgrade to FX but D700 prices are still high in your area, you may want to consider a D3 instead, which in my opinion is a much better value at current prices.
• A high-end APS-C body such as a Nikon D300, D300S or D7000 already has most of the features of the Nikon D3.
• You may want to use both an FX and a DX camera to get the best of both worlds.

Read More

My First Crush: the Fuji S5 Pro

This is a hands-on review of the Fuji Finepix S5 Pro, a DSLR released back in 2007 that combines the best characteristics of negative film and DSLRs.

Included here is a dynamic range comparison between the S5, the Nikon D3 and Nikon D70.


BACKSTORY
     I started taking photography seriously at the beginning of 2007.  My wife and I just found out that we were having a baby.  It was a time filled with joy, hope and anticipation.  I wanted to capture memories of our growing family beautifully.
     As I shopped for my first DSLR, I learned about the then-recently released Fujifilm FinePix S5 Pro.   I read the glowing reviews about its colors and dynamic range.  There were all sorts of stories about how this remarkable camera could capture images like no other.  Here is one of those famous S5 shots:

The shot that made the S5 famous.  Credit: Ryan Brenizer - blogged here.

FUJI S5 PRO
      The S5 is basically a Nikon D200 with a customized Fuji-designed sensor.  The S5 -- and its predecessor the S3 -- use Fuji's proprietary Super CCD sensor.  Unlike other sensors, the Super CCD sensor has two pixels in each photosite.  One of the pixels is like a regular CCD pixel.  The second pixel is a much smaller pixel that has a much lower sensitivity.  When the S5 takes a shot, the regular S-pixel records the image with the typical dynamic range of a CCD sensor.  At the same time, the smaller R-pixel, with lower sensitivity, is able to record the highlights and brighter parts of the image.  The information from both pixels is combined into an image that has a wider dynamic range than what would otherwise be possible.  It's like the S5 takes two shots at the same time and combines them into an HDR image.
      The other unique quality of the S5 is its colors -- specifically its skin tones, which many say is more flattering for lighter complexions than other cameras.
      These advantages made the S5 the perfect camera for people photos, and it was precisely what I wanted.  Unfortunately the S5 cost $1,899 then.  I couldn't wrap my head around that number, considering that I was looking for a DSLR under $500.  I sighed and mentally categorized the S5 as unobtanium.
      Eventually, like all cameras, the price started to decline.  But even as recently as a couple of years ago, it was still selling well above $1,000.  At the time I upgraded to a Nikon D300 in early 2010, I did consider getting an S5.  However, given the S5's limited resolution and older specs, I felt that spending that much money just to get the benefit of the extra dynamic range was a luxury I couldn't afford.
      Fast forward to a few weeks ago.  I felt it was time to upgrade and I sold my D300.  While researching which camera to get, I checked S5 prices on a whim and found to my surprise that some were selling for much more reasonable prices.  So, five years after pining for one, I finally got one.  (Cue the Carpenters' "Make Believe It's Your First Time" lol ).

DYNAMIC RANGE
I tested the S5's recovery limits.  As with the D3 and D70, I tested recovery from -5 stops below   the camera's metered exposure to +5 stops above.  Just for kicks, I also tested up to +10 stops above the metered exposure.

Here again is the test for the D70:

And the test for the D3:

And now the S5:

Note: for +8, +9 and +10, I only included black and white versions.

To me, the usable results for the S5 are between -4 stops to +5 stops (with the need to remove some magenta tints on highlights).  This is significantly better than the range of an old DSLR like the D70 (-3 stops to +3 stops).  It is not as good as the D3 in the shadows, but is better than the D3 in the highlights, so there is a net improvement of about 1 stop over the useful range of the D3 (-5 stops to +3 stops).  The S5's 2-stop highlight advantage makes sense because Fuji claims that the Super CCD sensor's R-pixels increases dynamic range to 400%.  Interestingly, both the D70 and D3 completely lose highlight detail above 3 stops, whereas the S5 manages to retain detail (but not color accuracy) up to +7 stops.  Indeed, at least one S5 user has been able to recover seemingly all color (with the need for some correction) at up to +6 stops.

PRACTICAL ADVANTAGES OF THE S5
You may be wondering: the S5's highlight range is amazing but you have to be a really terrible photographer to overexpose by THAT much.  Is the S5 only for incompetent photographers?  Is there any practical use for the S5's unique abilities?  I thought about this and came up with real world advantages of the S5 (with some overlap).  Here they are:

1. Correcting overexposure.  I can make mistakes with exposure (ambient or flash) in the heat of the moment.  It's nice to know that with the S5, those mistakes are probably recoverable, particularly for shots where there are no second chances.

2. JPEG shooting.  The S5's JPEG files are unlike other JPEG files.  They have about the same exposure latitude as the Nikon D70's raw files.

JPEG recovery from -3 stops

JPEG recovery from +3 stops

In addition, I find that the S5's automatic white balance is usually correct.  In the shots above for example, I did not correct the white balance from the yellow CFL lamp.  If you prefer to use JPEG instead of raw, or you hate postprocessing, the S5 provides some of the best quality JPEGs I've seen.

3. Raw editing.  If you like having complete control of the image and prefer to work with raw files, then the S5 provides immense (perhaps unparalleled) latitude with exposure adjustments, whether for correction or creative purposes.
.
4. Can shoot like negative film.  Because of the S5's exposure latitude, particularly with highlights, you can shoot with it like negative film.  If you want to shoot by exposing for the shadows without worrying about blowing highlights, just like negative film, you'll appreciate the S5.

Here's one shot where I got noise-free detail in the shadows by exposing for the ground (which consisted of black rubber tire fill):

This was the unedited shot:

5. High contrast scenes.  The S5 excels for capturing scenes with high contrast lighting.  In scenes like these, the challenge is to retain detail in both the highlights and shadows.  Here, I chose an exposure that I knew would allow the S5 to have sufficient detail in the shadows (increasing the camera's metered exposure by 0.3 stops).  I was confident that with the dynamic range set at maximum, the S5 would retain the highlight detail.

The SOOC shot (cropped)

In post, I merely pulled down highlights further and lifted the shadows to make the detail more apparent.

This is not to say that the S5's dynamic range is out of this world.  It can't capture both the solar disc and the shadow of a black cat in a backlit cave in the same scene.  Instead I would say it's close to what I can observe with my eyes.  If I can see it without squinting, it's very likely within the range of the S5.

6. HDR in motion.  The S5 combines information from two sets of pixels, almost like taking two shots at the same time at exposures that are two stops apart.  That may not seem like much, but unlike typical HDR shots, you have the ability to take HDR shots of moving subjects.

7. Wide angle shots and sky shots.  If you like using ultrawide angle lenses, you know that the sky is often a significant portion of the frame (especially if the camera is tilted upward as is commonly done with UWA), so you don't want it to be just blown out.  The S5 can help you retain color and detail in the sky while still having a workable exposure for the subject.

8. Photos of black and white subjects.  Who really needs 5 stops highlight range?  Isn't 3 stops (like that of the D3 and D70) more than enough?  Let's take a look at what "3 stops" really means. It does not always mean 3 stops overexposure.  If you look at the D3 or D70 recovery test at +3 overexposure, the parts that lost detail were the parts that were slightly brighter than middle gray, because at +3 stops overexposure those lighter areas were actually more than 3 stops above middle gray.  So by 3 stops, we actually mean 3 stops above middle gray.

If your whole scene is middle gray, then yes you can overexpose with the D70 or D3 by 3 stops and still recover it.  Great!  But what if there's something lighter toned in your scene such as let's say... average Caucasian skin?  At a normal exposure, that skin would be at around Zone VI, one stop above middle gray.  Now your 3 stops latitude has been reduced for the skin -- instead of 3 stops, you can only recover 2 stops overexposure of the skin.  Still no sweat so far.

But what if the subject is wearing a white dress that you wanted to look white (not gray)?  The resulting exposure of the white dress would be higher, maybe 1.7 or 2 stops above middle gray.  Now your 3-stop latitude has been reduced to 1 or 1.3 stops vis-a-vis that white dress.

Then what if it's a summer wedding, and there is direct sunlight on part of the white dress?  That sunlit part would probably be at least one stop brighter than the rest of the dress.  Now you are at the limit of your 3-stop latitude and if the sunlit part is significantly brighter, you would have spent all your 3-stop allowance and gotten 255, 255, 255.  And all of this assumes that you were exposing correctly.  We weren't even talking about overexposure yet.

But wait a minute, you say.  If all you're capturing is a blonde bride in a white dress surrounded by snow, then you're fine - you just expose the scene as if middle gray (retaining detail) and push everything in post to the limits of the paper or screen.  Yeah that would work unless of course in the rather "unlikely" scenario that your blonde bride is next to a groom in a black tux. hmmm.

With the S5 in the foregoing example, you could have a white dress that is partly sunlit up to 3 stops higher than the rest of the dress and still retain color and detail in the brightest highlights, while still capturing that detail in the groom's tux in Zone III (and shadows in Zone II).  In fact if you don't need color information in the white dress' highlights, your latitude is up to 5 stops higher than the rest of the dress.  That's why wedding photographers love(d?) the S5.

Black and white jacket, cloudless sky and black rubber tire fill in shade.  Not a problem for the S5.

9. Capturing highlight details.  Because the S5 has such a wide highlight range, it can capture very subtle highlight tones.

     Here is a demo snapshot of a house with white stucco in direct sunlight.  You can still see the texture of the stucco (even without me pulling down any highlights) - feel free to click on the file and zoom at 100% view.  In post all I did was crop it a bit, tweak the white balance slightly, lift the shadows, and increase the contrast.  I didn't have to recover the highlights and although the white stucco was illuminated by sunlight, not a square inch of the house was blown out.  In fact, the S5 still had room left in both shadow and highlight latitude, so I boosted the highlights a bit.  And BTW this was the JPEG file (not raw).

10. Backlit subjects.  Do you love taking backlit shots?  I do.  In terms of exposure backlit shots can be challenging because of the strong contrast.  This is another situation that plays into the strengths of the S5.

11. Flashless photography.  With a regular DSLR, I often use fill flash to lift shadows and reduce the dynamic range between the subject and a bright background.  With the S5 it is possible to omit the fill flash, get a decent exposure on the subject and recover the overexposed background.  Shooting without a flash allows me to work faster and not have to worry about making the direction, quality and color of the fill flash consistent with the existing light.

The original unedited shot:

The edited shot (lifted shadows, recovered highlights, increased contrast):

12. Underexposing the background.  [Caution: this is only hypothetical -  I haven't tried this out yet.]  

There are cameras that have unlimited or very high sync speeds such as the Canon 1D and the Nikon D70 (or most point-and-shoots).  They can be useful for deliberately underexposing the background for dramatic effect, then illuminating the subject with a flash.  However, for these cameras, the limiting factor becomes the flash duration.  For an SB-800 speedlight at full power, it's around 1/800.  For an AlienBee B1600 at full power, that's around 1/640.

Where does the S5 fit in this?  The S5 has a so-so sync speed of 1/250.  However, it has 2 additional stops of highlight range.  Assuming the subject is darker than the background, it should be possible to use the same flash power that you could use at a higher sync speed then pull down the background exposure.

Example:  Let's say you have sunny-16 conditions.  The exposure of the background at ISO 100 is f/16 and 1/125 (for simplicity).  To underexpose the ambient by 2 stops, you need an exposure of ISO 100, f/32 at 1/125.  If you could sync at 1/1000, then you could expose instead at ISO 100, f/11 at 1/1000.  To illuminate the subject, you would then need a flash that has sufficient power and distance to give you f/11 at ISO 100 to underexpose the background by 2 stops.

With the S5, our sync speed is only 1/250, which means that to underexpose the ambient by 2 stops, we would normally need an exposure of ISO 100, f/22 at 1/250.  Suppose however, that we nonetheless shoot at ISO 100, f/11, 1/250 and use a flash and distance that gives us f/11 at ISO 100.  The background would come out normally exposed and thus have all the detail.  In post, we could burn the background 2 stops.  In fact, assuming there are Zone VII white clouds in the background, we could place them in Zone X (by exposing 3 stops higher) and still recover color and detail in post thanks to the S5's 5-stop highlight headroom.  So we could instead shoot at ISO 100, f/4, 1/250 (resulting in a 3-stop overexposure), burn the background by 5 stops in post, and the background would appear 2 stops darker as we originally wanted.  A flash that could put out f/4 at ISO 100 is not asking much at all, and the undemanding 1/250 sync speed allows us to use even slower flash duration strobes such as the White Lightning X3200.

13.  Digital vs. Film Highlight Roll-off.  If you've been shooting negative film, you know that one way that negative film looks different from digital is the smoothness of the highlight roll-off.  With digital, the highlight tones get brighter then suddenly becomes totally white, like a car driving off a cliff.  With negative film, on the other hand, highlights get brighter very gradually until it finally reaches total white.

Photo by Le Anh Khoa.  Shot with Kodak 200 film.  Note the amazing highlights.

Here are a couple of S5 demo shots.  In this shot, the sun is in the frame (albeit covered by the plants).  If you look at the sky, you can see that it builds gradually to white, retaining detail all the way.  You can also see this effect in some of the playground shots above.

Here's another demonstration of the S5's smooth roll-off, using a white porcelain sink.

At the 100% dynamic range setting (i.e. with the smaller pixels turned off), the S5 acts like any other DSLR.   There are two issues with the sink.  First, the right side of the sink that is being hit with the light is blown out.  Second, the part where the sensor is able to record highlight detail stops abruptly before blowing out.

Post-processing adjustments (e.g. burning, pulling highlights, etc.) will not restore lost detail.  Here I pulled down the highlights in LR4.  The blown out patch is now a light gray, but still has no detail.  Neither did the roll-off become any smoother.  You can pinpoint exactly where the detail stops and the blown out area begins.

Here's what the sink looks like with the S5's maximum dynamic range activated.  Even without editing, and even with my crappy office monitor, I can see that there was detail recorded everywhere in the sink, even in the patch that was directly lit by sunlight.  Perhaps less evident but just as important, the transition to peak highlight is very gradual.

In post, I pulled down the highlights to make the tonality a little more obvious.  You can see that the tones built up very gradually, just like negative film.  The effect on this particular subject is that it revealed the delicate contour of the sink with finesse.

14. Skin tones.  This one is more subjective.  There's something about the way the S5 renders skin tones that is not 'neutral' but which many people find more pleasant.  It's not easy for me to articulate it, but the best way I can describe it is that the skin tone looks like the hues I usually see on foundation for makeup (before it is applied).

Here is a shot from the Nikon D3, which to me has skin tones that look the most natural among all the cameras I've used.

Here is how the same subjects look with the S5:

And here's another set.

(On the second set, S5's on top, D3's at the bottom.)

FUJI S5 IDIOSYNCRASIES
The S5 is not perfect and has some issues.

1. Speed (or lack thereof).  The biggest weakness of the S5 is that it is a relatively slow camera.  Very low frame rate in continuous shooting, the buffer fills up quickly, and takes a while to empty.  This is not a sports camera, and you can't just keep pressing the shutter thoughtlessly.  [I guess it really is like shooting with a film camera. :) ]  If you tend to pick your shots rather than spray-and-pray, it is not a big deal.

2.  Resolution.  It is true that the S5 has 12 million pixels, but the information from the second set of 6 million pixels is used together with those of the first set of 6 million pixels, therefore the real resolution is closer to 6 megapixels.  I don't mind -- 6mp is enough for me to print 12'x18' without any problem.  However, raw files have a resolution of 12mp.  Since the 12mp resolution is mostly just through interpolation, it is far less detailed than a real 12mp file.  I wish Fuji just kept it at 6mp. Or even better, I wish Fuji makes a successor to the S5 that has real 12mp.

3.  Handling.  One of the S5's quirks is its handling.  Its ergonomics are great - it is physically identical to the Nikon D200 after all.  It also shares with the D200 the ability to change many settings without going through the menus by pressing a button and turning the front or rear dial.

However, the S5's menu system is completely different from Nikon's.  That's too bad because one of the things I like with my Nikon DSLRs is the intuitive interface.  Unfortunately the S5's system is nowhere near as elegant as Nikon's, and the S5 has gotten a lot of criticism for it from Nikon users, particularly with the cumbersome playback controls.  In my opinion, however, the menu is usable once you become familiar with its quirks.  I certainly can't fly through the options the way I do with my other Nikons.  But using the S5 is only slightly harder than using a camera from a different manufacturer.  Seen that way, it's a bit less frustrating.

4.  File sizes.  Raw files with extended dynamic range are around 25MB each, whether you use 125% or 400% dynamic range.  If you use only 100% dynamic range it doesn't take up as much space but the S5's advantages over other cameras disappear.

5.  Exposure weirdness.  The S5 has a couple of exposure-related weirdness.  First, its metered matrix exposure tends to underexpose especially when there are bright highlights in the scene.  This is particularly strange because it fails to take advantage of the S5's highlight headroom.  I tend to get better results either by applying a +0.7 or +1 exposure compensation or by using center-weighted metering.

Another weird characteristic of the S5 is that when automatic exposure modes are used (P, A, S) shutter speeds can show up in strange fractions.  For example, before pressing the shutter you might see that the shutter speed in P mode is at 1/60, but when you shoot it might be 1/60 or it might be 1/55.

6.  High ISO Noise.  By today's standards, the S5's high ISO noise is not impressive.  I would say that ISO 1600 is usable.

7.  Film modes.  One of the S5's features is that it has film emulation modes.  My problem with those modes is that you can't control the dynamic range.  Each film mode has its own dynamic range setting.

8.  Postprocessing.  Although SOOC JPEGs from the S5 can look fine, the S5 really benefits from postprocessing.  First, at maximum dynamic range, unedited images can appear flat.  Proper post-processing brings back the contrast while preserving the dynamic range that you captured.

Second, if you really want to squeeze the most out of the sensor, in my opinion you need to post-process the file, otherwise the highlight detail is there but difficult to observe -- as shown in the porcelain sink example.  (That's why if you read Ken Rockwell's review of the S5, he claims that the increased dynamic range is "invisible" -- Ken doesn't like postprocessing.)

9.  File format.  The .RAF raw files from the S5 can be converted into DNG in Lightroom without loss of any information or dynamic range.  As a bonus, the file size is reduced from 25MB to less than 14MB.

One quirk of S5 raw files (whether RAF or DNG) in Lightroom is that the render previews take longer than usual to load.  In addition, they can sometimes show weird artifacts or posterization -- even though the file doesn't have any actual artifacts or posterization problems.  The render artifacts disappear when a 1:1 render preview is made.

SUMMARY
The S5 had been the undisputed leader in dynamic range when it was released in 2007 but two camera generations later, some cameras have caught up with (or even possibly exceeded) the S5's total dynamic range.  Nonetheless, the S5's highlight range remains unsurpassed.  The S5's extreme highlight range and nonlinear response allow it to function and provide shots like a camera with negative film, with tremendous exposure latitude and a very smooth highlight roll-off.  The result is an image that, in my opinion, comes closest to the look of film, a quality that is not necessarily reflected in numbers about total dynamic range, resolution or high ISO noise.


Update: FUJI S5 VS. NEWER CAMERAS
There are now cameras that have greater total dynamic range than the S5.  One could argue that even though the S5 has unparalleled highlight range, a newer camera with greater total DR can match that highlight range if it is sufficiently underexposed (and the shadows are enhanced in post-processing).  I did exactly that in a comparison between S5 and D90.  See here.  You can decide for yourself whether the D90 adequately matched the S5.

FUJI S5 RESOURCES

Other S5 Reviews:

• Ryan Brenizer
• Kirk Tuck
• Jonathan Ryan
• DPReview
• Alexander Kunz
• Collection of S5 Reviews

Other S5 dynamic range comparisons:

• S5 vs D3x
• S5 vs D700 (and a comment from a D300, D700 and S5 user)
• S5 vs D200

Here are some of my favorite S5 shots and galleries:
Leeisaku - note the skin tones in his portraits (as far as I can tell, they're all with the S5)

"On the Stairs" by AJT Images.  I love the dynamic range and the detail captured in the dress.

"Stern Lil' Lola" by Denyer - I like the subtle range of highlight tones captured in the model's face.


"At the arcade" by Alex Brown.  I love the skin tone, delicate tonality, and the detail in the hat.  If you told me this was shot with film, I would have believed you (perhaps Kodak Portra?).


"Rivert y sus ventanas" by Cristina Perea. That shaft of sunlight there isn't blown out.
"tuxedo park Ny" by SKGPhotos.  Nice showcase of the S5's highlight range.

RELATED POSTS
Comparing Digital with Film: Dynamic Range

Read More

Full Frame DSLR FAQ

Sorry to disappoint you guys, but I'm still with the dark side :)

With the release of the Canon 5D Mk III, Nikon D4, and Nikon D800, previous generation full-frame DSLRs have become available at a significant discount, putting them within the reach of photographers considering high-end crop sensor DSLRs.  Should you make the move?

This FAQ discusses the differences between full-frame DSLRs and "crop sensor" DSLRs.  It is divided into two sections: Basic and Intermediate.



PART 1: BASIC

1.  What is a full-frame DSLR?
It is a DSLR with a sensor that is about the same size as a 35mm film frame (36mm x 24mm), which became the dominant still photography format for consumers starting in the 1960s.

2.  What is a "crop sensor" DSLR?
It is a DSLR with a sensor that is smaller than a full-frame sensor.  It's sometimes called a "crop sensor" because the resulting image seems like a cropped version of a 35mm sensor image (even though there is no actual cropping involved).  There are many crop sensor formats.  Synonyms: small sensor camera.

3.  What is an APS-C sensor DSLR?
An APS-C sensor is a type of a crop sensor that measures about 24mm x 16mm, with some slight variation between manufacturers.  It is called APS-C because it is similar to the 25.1mm x 16.7mm size of the Advanced Photo System film format introduced by Kodak in 1996 (like Kodak Advantix, if you remember those).  The APS-C sensor size is currently the most common sensor format for DSLRs, especially entry and mid-level DSLRs.

4.  What is a crop factor?
It is the ratio of a full-frame sensor length (or width) to the corresponding length (or width) of a crop sensor.  For a sensor measuring 24mm x 16mm, the ratio is 1.5x.  (Take note that this is not the ratio of the area of a full-frame sensor to a crop sensor.  A full-frame sensor has more than twice the area of an APS-C sensor.)  When a lens is used on a crop sensor body, it has a field of view on a 35mm body when the focal length is multiplied by this ratio.  For example, a 50mm lens used with a Nikon APS-C sensor DSLR results in a similar field of view with a 75mm lens used on a full frame DSLR.  Sometimes referred to as "focal length multiplier," even though the focal length is never actually multiplied and retains the same distance regardless of the format.

Check the image below, it represents different sensor sizes, with the crop factor shown at the left edge.

Sensor Sizes Compared - Credit: Wikipedia

Now that you've seen how much difference there is between sensor sizes, let's relate that to real world cameras, some famous full frame DSLRs are the Nikon D3 and the Canon 5D, and famous crop sensor (APS-C) DSLRs are the Nikon D90/D300 and the Canon 60D/7D. The 1/2.3" sensor is your average point and shoot sensor, just for fun, see how many 1/2.3" sensors you can fit inside a full-frame sensor, the answer is 35.

5. What are the advantages of a full frame sensor DSLR over a crop sensor DSLR?

• Shallower depth of field (see below).
• Generally has less high ISO noise.
• Generally has better dynamic range (can record more shadows and highlights data).
• Viewfinder appears larger.
• Lenses with traditional focal lengths can function as designed on a full-frame (where they are often too long for an APS-C camera).  For example, 24-70mm for standard zoom, 70-200 for telephoto zoom, 85mm for portraits.
• All other factors being equal, a full frame camera has a higher diffraction limit (i.e., a narrower aperture before the effects of diffraction cause image sharpness to deteriorate).

6. What are the advantages of a crop sensor DSLR over a full frame sensor DSLR?

• Deeper depth of field (see below).
• The DSLR body is usually smaller, lighter, and less expensive.
• The lenses are usually smaller, lighter, and less expensive.
• A crop sensor DSLR can use lenses made for either full-frame DSLRs or crop sensor DSLRs whereas a full-frame DSLR generally cannot use a lens made for a crop sensor DSLR without heavy vignetting. In Canon's case, you can't use crop sensor lenses (EF-S) on a full frame.
• When a full-frame lens is used on a crop sensor DSLR, you are using mostly only the center part of the lens, which tends to be the sharpest and least susceptible to vignetting, i.e. while your 50mm f/1.4 lens is sharp wide-open on your APS-C camera, it might not be as sharp at the same aperture on a full-frame camera, since now it's using the whole projected image of the lens instead of only using the center area of the projected image (usually sharped than the edges) in the APS-C case.

Part 2: Intermediate

DEPTH OF FIELD

1.  How much shallower is the depth of field of full-frame?
Short answer: for the same field of view, about 1.3 stops (more than twice as shallow). But see below.

2.  If you use the same lens on a full-frame camera and a crop sensor camera, will the depth of field be different?
Answer: There are two possible answers to this question. One way to answer the question is that the depth of field will be the same, given the same actual focal length, aperture and same camera-to-subject distance, regardless of format.

To make it more clear, I made the below illustration quickly using MS Paint, the black circle is the image projected by the lens (remember, we are using the same lens for both sensor sizes), now if we place the full frame sensor in the middle of the projected image, we capture most of the projected image. But if we put the smaller crop sensor in the middle, we would only capture the middle part of the projected image, and as a result it will look as if it is magnified. But will the DoF be different? No, everything is exactly the same in both situations, the only difference is in the magnified look. Simply stated, if you take the full frame shot and crop it to match the crop sensor shot, both images should be identical.

The other possible answer takes into account that the circle of confusion for a crop sensor is actually smaller than that of a full frame sensor.   The APS-C sensor is like cutting the middle of the FF sensor then magnifying it to the same size as a full frame sensor, so any blurriness becomes more noticeable.  It is the same reason why the diffraction limit is lower (wider aperture) for APS-C compared to FF.
You can try this in the DOF calculator.  If you keep all variables constant and the only thing you change is the camera (from APS-C to FF or vice-versa), you'll see that: 1) the circle of confusion is smaller for an APS-C; and 2) the DoF is actually shallower for APS-C compared to FF.  Therefore it could also be said that at the same actual (not equivalent) focal length, aperture and same camera-to-subject distance, the crop sensor camera has a shallower depth of field.

I have tried this out as an experiment, using a 50mm at f/2.0, with a full frame and with a crop sensor, and I don't see much of a difference in depth of field (the text on the far block is just as blurry in both shots).

50mm f/2 on full-frame sensor

50mm f/2 on APS-C sensor

3. If you use the same lens on a full-frame camera and a crop sensor camera, and you crop the full-frame image, will the depth of field be different?
Answer: The depth of field will be the same. Here are a couple of test shots. Both were with a 50mm at f/2.  The top one is with an APS-C camera. The bottom is with a full frame but I forced the camera to crop the image to an APS-C size. The resulting images are essentially identical.

50mm f/2 on APS-C sensor

50mm f/2 on full-frame camera with forced APS-C crop.

4.  If you use a 75mm lens on full-frame and put a 50mm lens on a crop sensor, will the field of view be the same? Will the subject distance be the same (for the same field of view) and will the depth of field be the same?
Answer:  The field of view will be the same and the subject distance will be the same. The depth of field will be different -- the full frame will have a shallower depth of field under these conditions (because of the longer focal length being used on the full frame camera).

Here are test shots that demonstrate this. The top one is with a crop sensor camera at 50mm, f/2.8. The bottom one is with a full frame camera at around 75mm, f/2.8.  Both shots have essentially the same field of view, and I was shooting from an identical distance from camera to the subject. The depth of field is quite different, as you can see from the blurriness of the farther block.

APS-C at 50mm f/2.8

Full frame at around 75mm, f/2.8

5.  What if I want a deeper depth of field?  Is an APS-C camera a better tool than a full frame camera?
I have not tested this, but at least one commentator believes that a full frame camera has the advantage for a deeper depth of field because you can take advantage of the higher diffraction limit and better noise to use a narrower aperture to achieve the same depth of field with a better signal-to-noise ratio.  See   http://www.clarkvision.com/articles/dof_myth/

LENSES

1.  Can you use a crop-sensor lens on a full frame?
Answer: It depends on what camera you have.  Usually, the image circle of a crop-sensor lens will not fill the frame of a full-frame image.  (Note: there are a few crop-sensor zoom lenses that can fill a full-frame at longer focal lengths).
On Nikon full-frame cameras, the camera can automatically detect that a crop-sensor lens ("DX") is being used and can crop it automatically. (However, the automatic cropping can be turned off if you wish).
In Canon's case, crop sensor lenses (with the EF-S designation) cannot be used on full frame cameras, other than the vignetting, they have a different design with the lens' rear element closer to the sensor, and if used on a full frame camera, the camera's mirror will hit that protruding part of the lens when taking a picture. Now you don't want to do that to your multi-thousand dollar camera.

APS-C lens used on a Nikon full-frame; automatic cropping disabled

2.  If you use a crop-sensor lens on a full-frame will you retain the high ISO advantage of the full frame?
Answer: Yes but at a lower resolution. If you down-res a crop sensor photo to the same size, the gap in ISO performance will be reduced. The shot below is with a crop-sensor lens on a full frame body, with automatic cropping activated. The shot was made at 25,600 ISO (with +30 luminance noise reduction applied in Lightroom 4).

APS-C lens on a full-frame; 25600 ISO

3.  Is a full frame camera better for taking wide angle photos?
Answer: A lens will appear wider on a full-frame camera assuming it can fill the entire frame. However, there are now several ultrawide angle lenses for APS-C cameras. For example, the Sigma 8-16 (APS-C only) has an equivalent focal length of 12-24mm, which is just as wide as, for example, the Sigma 12-24 on a full-frame. These crop sensor lenses are usually less expensive than their full frame counterparts, and have similar performance.  Moreover, a crop sensor camera has a deeper depth of field which is often desirable for landscape photos.

NOISE

1.  How much better is the full-frame noise performance?
Answer: This will of course vary with the cameras being compared. In general, for the same generation, the difference is about 1.5 stops. For example, the Nikon D300 (announced August 2007) was measured by DXO to have a signal-to-noise ratio of 30dB at 679 ISO. The Nikon D3 (announced together with the Nikon D300) was measured at 2290 ISO using the same standard, a difference of about 1.7 stops. The Nikon D7000 (announced September 2010) has a score of 1167 ISO, a difference of about 1 stop compared to the Nikon D3.

[mshafik: in my case I don't really believe/care about DXO's measurements, noise is a very subjective topic, and what appears bad for me, might look very good for someone else. I find that shadow noise is my worst enemy, so when I am taking a shot at a high ISO, I make sure that I expose the photo correctly or even slightly to the bright side. But to answer the question, comparing my Canon 60D to my 5D Mark II, I find that 6,400 ISO on the 5DMk2 is as good as 1,600 ISO on the 60D].

2.  Does the full-frame high ISO performance make a real world difference in the kinds of shots you can get?
Answer: Of course, previously I was exclusively shooting at 1,600 ISO, and only used 3,200 ISO for emergencies (with not-so-great results), but now I am using 3,200 and 6,400 ISO without a second thought. The benefit is the ability to shoot in darker situations with available light only, and the ability to use faster shutter speeds to reduce blur and freeze action.

DYNAMIC RANGE & COLORS

1. How does the dynamic range of a full-frame compare with that of a crop sensor?
Answer: All other factors being equal, a full-frame should have greater dynamic range because it has greater capacity for collecting photons.  Source: http://www.cambridgeincolour.com/tutorials/digital-camera-sensor-size.htm and http://www.clarkvision.com/articles/does.pixel.size.matter/
I cannot test this scientifically because I sold my Nikon D300, and my remaining cameras have too many other differences between them (my full frame D3 is one generation newer than my APS-C D70, and my D70 camera is mid-level as opposed to high end).  In addition there is a lot of variation in dynamic range between different manufacturers' cameras.  Plus, there is an element of subjectivity in judging dynamic range because people have different levels of what they consider to be acceptable noise or color shifts.

For what it's worth, I tested my full frame D3's and APS-C D70's ability to recover highlights and shadows.  I did a 10-stop bracketed exposure, from -5 underexposure to +5 overexposure, then reversed the under- or overexposure in post using Lightroom 4, with white balance corrected and zero noise reduction applied.  The slideshows below shows the results.
First, the D70:


Next, the D3:

The full album for the D70 is here, and the D3 is here, both with full resolution JPEGs.

In terms of recovery from overexposure, the D70 has possibly usable results from 3 stops overexposure, but there are several areas that were blown out and have no detail or have no color.  The D3 has a better result at recovery from 3 stops overexposure, with less blown out areas that have no color.

When recovering from underexposure, the D70 has banding noise, yellowish tint in the highlights and a bluish/purplish tint in the shadows when recovering from -5 and -4 stops underexposure.  Meanwhile, the D3's recovery from -5 stops underexposure has slight yellow bands and a slight bluish/purplish tint in the shadows.  It may be usable.  Here is what it looks like with a +40 noise reduction in Lightroom applied (click for full size):

I don't have a real world shot demonstrating the dynamic range of FF, but here's a demo of sorts.  Here's the original shot (the light is a bug zapper with bluish light, and nearby is a yellowish CFL lamp):

And here's the shot after adjustments in Lightroom 4 (click for full-resolution JPEG):

RELATED POSTS

1. What's the point of having different lenses?
2. Using Zoom Lens as Compositional Aid
3. Alternative Full Frame Lenses for Nikon

Read More