This posting is a rough translation of several comments I wrote in diverse internet fora about my first impressions with the D7000. The camera arrived on Nov 5th and in the following 3 weeks, I’ve basically “blogged” in the german NFF forum my experiences with the newly introduced camera. Others, from the english speaking community, asked about a translated version of this. A good opportunity to start my small project over here. I love to learn and I love to share. Maybe some people find the content useful.
I am not a reviewer or tester, neither do I pretend to be one. Heck, I am not even a professional photographer. I just enjoy a great hobby. Big time. Please don’t take my personal impressions as absolute or even relative statements about the qualities of a camera or other equipment, they are nothing more than my humble view as a simple user. I won’t invest much additional time to carefully review my own statements. I’d rather prefer to share findings as they come along, which sometimes mean that I will contradict myself, based on new experiences.
One additional note:
Many of the images are available in original size JPEGs or NEFs files (about 2 GB in total)
Most 900×600 images are here: (www.pbase.com/andrease/d7000 )
Most original JPEGs and NEFs are here.
With that, enjoy my first blog page about the D7000,
Please consider that my initial comments on the D7000 only covered a period between Nov 5th and Nov 27th, 2010. A period well before the highly respected reviews from dpreview and Thom Hogan were published.
Nov 5th, 2010
Got my D7000 today. Only kits arrived and my dealer was so kind to remove the kit lens (18-105mm) – I have already 2 of them. The sun was going down and I did not have much time to shoot a few pictures. Living in Vienna, Austria – Schönbrunn palace was close by so I stopped and shot there. These first photos were not taken on a tripod, not even careful (which pixel peepers could and would analyze), just some shots to visually support my initial elaborations.
- Feels like a D90. From a weight, balance perspective it is more in the the D90 class. The D300 feels massively different – just massive.
- Nikon introduced again a new battery with the D7000. Of course with new chargers, I have currently 7 different batteries and chargers just for my Nikon D-SLR lineup. (update: according to Thom Hogan, the decisions to change the batteries are based on new safety regulations)
- The D7000 has a screen cover (like the D300 and others). Don’t know if the glass is tempered or not.
- With proper lenses, AF seems to be quick
- Rather annoying for me is the default settings of the camera – fully super duper automatic everywhere. 39 active focus fields, AF-C on by default. The D7000 does not have a menu entry to change the AF mode to AF-S. It took me a while to figure out that Nikon introduced a new operation on the AF switch to change it – which I did not know in my first hour of using it. Verdict: Any camera with these default settings has to be a camera in Nikon’s consumer department, its just so disturbing not to know which AF focus field will be selected. No wonder that many users complain about erratic behavior and unsharp pictures with their D-SLRs. (Too much default “automatic” for me)
- I changed the storage option to 14bit, lossless compression and JPEG large/fine. Expect each photo to consume 20-30 MB on your card. RAW files only are between 17 and 20 MB.
Optical properties (Where to start? Some will moan, but potentially more will be quite happy)
- Resolution is good, very good. Don’t know, why the D7000 and CNX2 have such a strong default noise suppression with accompanying loss in sharpness. Turning off the default settings in post processing solves this and images look sharp.
- The high pixel density doesn’t cope well with fast primes. I had my AFS 85mm/1.4G in the bag and wide open, basically every photo had significant color aberration in OOF area’s. In many cases CNX2 can remove most of it. Recommendation: Take care of it and please check if your preferred post processing software can deal with it.
- The bigger files put more demand on your computer in post processing compare to your D90 and D700 files. People interested in the D7000 shouldn’t overlook this part if your current PC is a bit aged.
- This camera with its advanced high res sensor will clearly show more weaknesses in existing lenses vs. other cams. Will be interesting to follow.
Here is an index of some of my very first photos shot with the D7000. Full size images can be downloaded here (Please use the # to identify)
Nov 5th, 2010
The viewfinder doesn’t seem to be a real 100% viewfinder. I was surprised to see more of Schönbrunn palace on my photos vs. what have been visible in the viewfinder.
To compare sharpness of the D7000 with other Nikon D-SLR bodies. Unfortunately, my box of autumn leaves aged a bit (a week) but you can find the original NEF files of now 21 Nikon D-SLR cameras here.
(Comment: I’ll describe the setup of this comparison in a different blog page at a later time.)
Nov 6th, 2010
A little teaser on the D7000 vs. D90 game
D7000 – ISO 100
D7000 – ISO 6400
D90 – Hi1 (ca. ISO 6400)
All noise suppression turned off in CNX2
100% crop of the D7000 / ISO 6400 picture
100% crop of the D90 / Hi1 picture
The visual difference grows by resizing the photos to the same output size and default noise suppression in CNX2.
Nov 6th, 2010
Question: How would you compare the D7000 to the D300? Anything to note?
Based on the few hours with the new camera and the usual disclaimers:
- The D300s seem to have more rubber in its coating which seems to be more “sticky” while holding the camera. With many bodies, I don’t have any straps and keep the body with the lens attached only on my finger tips. The more sticky the rubber there is, the easier and less strength I need. Having the camera and the 70-200 for an hour on your fingertip make a difference here.
- Not sure with the left mode dial, which one is the better one. With theD300s, it is easier and faster to change ISO settings (which I do very often), basically I never change the image quality and white balance, so I don’t need it there. With the D7000 the change of the PSAM modes is faster.
- The operation of the release mode dial under the mode dial is better on the D300. The reason: The lock is on the front of the body and it is very intuitive and easy to unlock with the index finger while turning the dial with the thumb. With the D7000 the lock button is on the back on the body. Due to the relatively high mode dial, my thumb has a harder time to press the lock and the index finger has to turn the release mode dial. Unfortunately, the D7000 has less space there, which complicate operations additionally. I can basically operate the D300 settings blind (i.e. setting in the dark to MuP or CH), can’t do it on the D7000.
- On the right side the handle bar around the shutter button is shaped differently, steeper with the D300, less with the D7000. I’ve got used to the D300 style, so I feel more comfortable there.
- The speed of storing a picture seem to be on par.
- On the back of the camera bodies. The 5 buttons on the right side of the screen of the D7000 are more overloaded. 2 buttons have 3 modes, 2 buttons have 2 and one button has only one mode. On the D300, 2 buttons have 2 modes and 3 buttons have one mode.
- The upper display of the D300 is about 50% bigger. Easier to read.
- On the D300, the AF mode can be set on the back of the camera with a simple lever. The D7000 has its AF mode set on the front close to the bajonett. The D300 variant is faster, but the D7000 is not bad either.
- The remote on the D300 is cable based, the D7000 has an one with infrared.
As said, just a quick run through the external differences.
Nov 6th, 2010
Question: Please post some photos with the AFS 18-200mm VR.
Here we go – D7000 and the AFS 18-200mm VR
The original JPEG is here – please download file #372
100% crop. So much for the possible resolution. Check the cover of the cooler.
Nov 6th, 2010
Had to walk with the dog this morning. Took the D7000 and the AFS 60mm/2.8 Micro and the AFS 18-200mm VR type I. The higher resolution is clearly visible. It is fantastic to see thing with fine details and structures. Especially the 60mm Micro allows the D7000 to display its performance.
Considering the price, it looks like Nikon landed a winner with the D7000. Depending on the feedback and gossip in the communities, the demand for the D7000 will most likely increase over time.
Nothing new from the handling side. As with all small bodies (D40, D40x, …) the left wheel quite often gets changed when taking the camera out of the hip-bag. I don’t use every time the rear screen, so I can’t see immediately if my mode changed from A to M. Weird. Would be good to have a control wheel with more friction.
Full sized images can be found here (Please use the # listed below)
Photo (#274), only resized
Photo #245, only resized
Photo #342, only resized
Nov 7th, 2010
Some high ISO experiences.
I did capture the same scene with all ISO setting and the following cameras – D7000, D300, D300s, D3100 and D90.
Looking at the the ISO 3200 images, my personal ranking would be:
The first 3 cameras are relatively close together, 4+5 are following with a visible gap. The surprises for me are the D3100 and the D90. The D3100, as the first camera which is according to internet gossip supposed to be equipped with the first “pure" Nikon sensor (design and manufactured). The “old” D90, which is clearly better than both its sensor cousins D300 and D300s.
To state it clearly: The D7000 is clearly better at ISO 3200 than the D300 and by a smaller margin to the D90.
I can not confirm that the new AF module in the D7000 is clearly and markedly slower than the D300 AF at low light. If there is anything slower, than most likely caused by the respective lens. With a fast AFS lens, both seem to be on par. No issue here.
Nov 7th, 2010
More info’s to my statements above on high ISO performance.
- I took only camera models which were of some relevance to the D7000
- D7000 – the new kid on the block with the highest resolving sensor. At same output size, the higher resolution is an advantage in terms of visible noise, which is more “hidden” by bigger size reduction.
- D3100 – The new entry camera. Presumably the first camera with a pure Nikon sensor. 14 MP.
- D90 – The former shooting star. The D7000 is it logical successor.
- D300/D300s – As there has been so much interest just before the release of the D7000, I’ve included both pro models to be compared against the D7000.
- As usual with the halogen lights from my ProFoto D4 gear. The D4 can only cover 8 f-stops with flash settings – not enough to cover ISO 100 – ISO 25.600
- All bodes were set up to; WB=incandescent, noise reduction = OFF, manual exposure, aperture=8, exposure time varied from 1/15 sec (ISO 100) to 1/4000 sec (ISO 25.600), best file size option in RAW and JPEG, lens used was an AF-S 70-200mm/2.8 VR II mounted on my Wimberly II head, 10 sec self timer used, VR = OFF.
- 2 directed lamps with 45 degrees from left and right
- The different bodies seem to have a left-right skew in sensor position. All are DX cameras, but on some cameras with the lens set up solidly on the tripod, frames aren’t identical. Would be great if someone would look into this as well. At least an interesting observation, need to check this later.
- At default settings, the D7000 and D3100 seem to have less color saturation at ISO 3200 vs. the two D300 twins.
- CNX2 (ver 2.2.6.) by default set on all ISOSettings with all cameras NR to ON. Don’t know how to turn this off by default. Especially the D700 suffers from these default settings. I suppose that Nikon was willing to to “use” some of the higher resolution of the D7000 to improve its ISO noise handling. It might be that Nikon assumes that most pictures in the future will be anyway resolution capped (by lenses and diffraction limits above an aperture setting of 8.)
- Both, the D300 and D300s have significantly higher dark noise levels than the other 3 models.
- With this scene, the D7000 is about one level better than the D300. The ISO 6400 image of the D7000 looks like the ISO 3200 image of the D300.
- IMHO, Hi2 is a marketing setting – unusable, expect in emergencies. Hi1 is not so bad (ISO 12.800 equivalent), especially used in combination with smaller output sizes (i.e. 6 x 9 inch).
- The D3100 and D7000 are similar at the ISO 3200 level. At ISO 6400 the D3100 looses more vs. the D7000. The D90 has a good first impression, but looses significantly contrast.
Based on similar assumptions and image expectations, the D7000 can be used until ISO 12.800, the other bodies somewhere between ISO 3200 and ISO 6400. YMMV.
All NEF files are available here (36 files, approx 560 MB in total)
This is the overall scene (sorry for the lack of creativity …). D7000/ISO 3200 image, just resized
100% crop, D7000, ISO 3200
100% crop, D90, ISO 3200
100% crop, D300, ISO 3200
100% crop, D300, ISO 3200
100% crop, D300s, ISO 3200
100% crop, D3100, ISO 3200
Nov 7th, 2010
Question: Could you please post some images comparing the sharpness of the cameras with the AFS 60mm/2.8?
3 camera bodies: D300, D3100, D7000
- Settings: tripod, JPEG and RAW at highest settings, 1/125 sec, f8, all NR=OFF, in CNX2 manually NR set to OFF (again), D7000 and D3100 with timer, D300 with remote and mirror lockup, WB set to flash, ProFoto studio flash (D4)
- Again and again. There is no bad camera around anymore. All bodies are able to capture sufficient sharpness for most users. The D300 is surprisingly good. Seems that the quality of the elements in front of the sensor (AA filter, Bayer filter, micro lenses) have an impact on the result. Not too much difference in 100% view – don’t forget we have 3 different sensor resolutions.
All JPEG and NEF files are here.
The overall scene:
100% crops of D300, D3100, D7000; USM set to 12 & 5 in CNX2
Nov 7th, 2010
Ok, one more thing. Had my AFS 28-300mm VR laying around. I was interested, how well the super zoom cope with the high pixel density of the D7000 sensor. Will the lens be a severe limitation?
From a usage perspective, the combination D7000 and 28-300mm is not well balanced. The lens seems to be significantly heavier than the body. In my view, this combination has its center of gravity too much towards the front end. What a difference the heavier D300 makes – well balanced ….
Ok, interested about the sharpness?
Same setup as above, about 4m (13 ft) distance, the lens set to 300mm, VR =OFF. D7000 used with mirror up.
I don’t know what the initial batch of US-based users did with their D7000, but from my perspective, the camera is able to deliver sharp images. The 28-300mm does not seem to limit the capabilities of the camera too much.
The original NEF files are in the same directory as in the post above.
100% crop, D7000, AFS 28-300mm VR
Nov 8th, 2010
Question: How many images will fit on a 32GB card?
My first 400 images used 7,9 GB for RAW and 2,5 GB for JPEG. Which gives you approximately 1200 images on one 32 GB card in the highest JPEG+RAW settings. If you take mostly highISO images you have to reduce these numbers (they are usually bigger).
The D7000 is very energy efficient. Potentially it is possible to capture 2 cards with 64 GB each with one charge. For those with a desire to travel light an interesting aspect. Like with the D3, where I don’t have issues to capture 5.000 images with one charge, which is again the capacity of 2 x 32GB CF cards I have in the camera.
Ok, let’s do some quick back of the envelope calculations:
- D7000 battery: 7 V, 1900mAh = 13,3 Wh
D3 battery: 11,1 V, 2500mAh = 27,8 Wh
Based on the average image size, the D7000 can write 9,6 GB /Wh and the D3 is able to deliver 2,4 GB/Wh.
According to this very unusual performance metric, the D7000 is about 4 times as energy efficient as the D3 – not bad for 3 years of development time. Kudos to the Nikon engineers.
Nov 8th, 2010
Another indicator (can’t call it a test at all)
Shutter lag of the D7000 as explained here. Did it multiple times and averaged the results. With pre-focus it took about 0,1 sec and without pre-focus about 0,13-0,15 sec. Surely it is imprecise, but lots of fun anyway. Give it a try.
Nov 8th, 2010
Question: Is there any impact on the performance using 14bit RAWs?
Using high speed mode (CH), both settings, 12 and 14bit sound similar fast. After filling the internal buffer, the continous shooting speed is reduced by the respective increase in data to be stored on the SDHC card (approx 2 images/sec in both cases – on a 32GB 30MB/sec Sandisk card)
Haven’t checked the difference between 12 and 14 bit RAW images on the D7000. As the camera does not hit the speed brake like the D300 and the D3x with 14bit RAW, I see no reason to use 12bit RAWs. The 4way “oversampling” and respective slowdown Nikon put into the A/D stage of the D300/D3x seems not necessary with the new sensor.
The value of 14bit RAW is most visible by color correcting images under narrow spatial light sources, like high temp mercury-vapor lamps.
My D7000 is set to 14bit RAW and lossless compression.
The capacity of the internal buffer is:
– 12-bit RAW: 11
– 14-bit RAW: 10
– 12-bit RAW: 15
– 14-bit RAW: 12
Nov 10th, 2010
Somebody asked for a low light comparison between the D7000 (DX) and the D700 (FX).
A very quick and simple comparison with only one image:
- With low light the D7000 tend to favor the red channel. Either the red filter in the RGB filter is more efficient or the camera’s software settings is more tuned towards reddish images.
- Noise reduction with long exposures takes as long as the D700 and D300 – no improvement. On 30 sec exposures, between 10-15 sec NR post processing in the camera have to be added before the image is stored on the SD card.
- The D7000 can’t compete with the D700 in night and highISO settings. The D700 is able to preserve much more detail vs its newer silbling.
- The crown is still with the FX camp. The D7000 challenger does not de-throne the D700
It was really dark. The human eye could not see these details. Exposure time: 20 sec, f4, ISO 3200
D700, Auto WB, 100% crop
D7000, Auto WB, 100% crop
D700, WB set in CNX2, 100% crop
D7000, WB set in CNX2, 100% crop
Nov 13th, 2010
(For my international readers some background info: In this week, the german photo magazine Color Foto published a review of the D7000, titled “D7000 – A step back”. Basically rating the D90 higher than the D7000. You might imagine the level of chat and noise this generated in the german speaking photo communities. The major point of contention was the choice of the magazine to base its rating on the OOC JPEGs both cameras produced at factory default settings)
Landed back in Vienna, bought a copy of ColorFoto at the airport. Later that night, I took my D90 and the D7000, made a photo and compared the 2 images carefully. Observing the development of the digital camera arena in the last 10 years it seems to be obvious that a so called “final” image is the sum of many components in the image chain.
What kind of components are we talking about, which influence the “final” image a user get on his screen? (does not claim to be complete):
- ambient light (especially considering spectral distribution, the WB setting in post processing can’t completely compensate for it)
- the quality of the lens (with more than 10 different attributes impacting quality)
- the AA filter (quantum efficiency, spectral profile)
- micro lenses (transmission, …)
- the sensor in itself (multiple factors)
- first and second stage of (analog) amplification
- A/D converter
- WB circuit (its software and hardware interaction)
- the creation of a RAW file, considering ADL and WB settings
- the demosaic process, considering many settings, like camera model, lens model, picture control, …
- Nikon’s secret sauce in their proprietary JPEG software algorithms
- A “final” JPEG arrives
After this initial thoughts about the image pipeline, lets go the nuts and bolts of this discussion. What is the difference between the D90 and D7000?
- One of both contenders does not interpret ISO correctly. When I took the pictures with my quite precise studio flash system (Profoto D4) I could assume that doubling the amount of light should balance the 2 different base ISO settings (ISO 100 , ISO 200). Unfortunately this is not the case. Both camera’s were set to 1/125 sec and f11, the D7000 to ISO 100 and the D90 to ISO 200. The D7000 image was 1/3 f-stop brighter than the D90 image (or vice versa the D90 image 1/3 f-stop darker). I don’t own any scientific equipment to be able to find the reference values. Anyway, put simply, 1/3 f-stop difference.
- Both camera models seem to have different interpretations what the WB setting “flash” should mean. Both were set to “flash” but the color balance is different.
- Using both camera models with default picture control settings produces different results (color balance, …)
- The image process leading to the image in CNX2 seems to be different. Even after calibrating the 2 images on brightness and white balance with the reference card, the images display visible differences in parts of the spectrum. Even the histogram curve looks different. Don’t know if this is the result of the different image sensors or caused by different software algorithms – who know’s.
- There is a right/left skew between the 2 cameras. Mounting the cameras on my lens which was mounted on my sturdy tripod, I often observed that the images have some pixel lines recorded on one sensor on the left side, which the other didn’t capture and vice versa. About 20-30 pixel lines.
- Using default settings in picture control, the D7000 produces softer images than the D90 (also on default settings). Don’t know if the difference is now based on different sensor technologies or software algorithms and respective parameters the engineers selected during the design phase.
- Using the RAW file, the D7000 image is at least as good as the D90 image. I used for both images the same CNX2 settings.
To sum it up: It is possible that the Nikon engineers of the D7000 project chose different standard settings in picture control and the software image pipeline. Comparing D7000 and D90 JPEG images with default settings is rather a comparison of software than sensor technologies.
It would be interesting to find equivalent settings between the different models to get similar visual results. This potentially only possible with the support of Nikon to explain and publish the impact of all those small wheels and knobs.
Here are the RAW files to play around.
D7K, WB and brightness adjusted to D90 levels. (It does not mean that the D90 has the better more correct values. The goal was simply an adjustment between the 2 bodies)
D90. check out the puppet. The color appears more intense there. On my calibrated monitor, the D7K colors look more natural than the D90 though.
Both histogram curves
Nov 14th, 2010
Since the first D7K cameras were shipped to customers, reports of “soft” and “unusable” images came around, culminating in the review of ColorFoto claiming that the D7000 is “a step back” in design. A good reason for a deep dive in the interaction of hardware and software in modern digital cameras.
My thesis: Going forward we wont be able test the capabilities of hardware aspects and parameters of digital cameras, but we will rather see hardware aspects enriched with software functions designed by the respective development engineers. As described above is the image pipeline based on a set of complex interactions.
Back to the topic of OOC JPEG sharpness of digital cameras.
As described at the beginning of my comments, the D7000 has always NR turned on in its RAW files – independent what the user set at the camera. On the other hand, some people claim that the D90 has better sharpness in many cases. etc, etc …
I’ve done some homework and assembled a table describing which NR settings Nikon engineers chose to implement as default settings for different camera models. The table contains data for 4 digital bodies – D7000, D90, D300s and D700. Of course there are some differences.
- ALL bodies have NR set to ON at higher ISO settings – completely independent what the user set on camera. Also valid for NR settings = OFF.
- The D300s seems to be completely independent of its settings. Both NR=OFF and NR=ON produce the same NR settings in the black reference images. (Can somebody verify this with a D300s ?)
- The D90 sensors seems to depend less on software support for its NR compared to the D7000 (just check how long the Nikon engineers assigned the value “0” to the D90 NR settings)
- There is a significant difference in the default settings of the D90 and the D300s, especially starting at ISO 16000. Internet folklore consistently claimed that a) both cameras have the same sensor and b) the professional D300 series usually apply rather less modifications to an image. Leaving it to the knowledgeable user to correct images appropriately. Seems to be a gap between folklore and actual technical settings.
- NR settings are “exploding” with all cameras at settings of Hi1 and Hi2. Imho, this is a nice piece of evidence how the engineers had to implement with crazy settings the requirements of the marketing department.
- FX bodies have a clear advantage above ISO 1600 (from point of view from Nikon engineers. The D700 sensor is the only sensor, which still has a NR setting of 0 at ISO 1600.
- Consequently, enthusiasts will have to deviate more and more from the default settings, which are targeting the mass market customer base.
- To avoid any misunderstandings. Opening a RAW file in CNX2 with NR strength value set to 0 (zero) and NR sharpness set to 5 improves its displayed sharpness when the NR settings in CNX2 are turned OFF.
It is my impression that the strategy of the Nikon engineers for the current generation of sensors and its market positioning of the 16 MP sensor is as follows:
Nikon is using the additional resolution of the sensor to predominately improve the visual noise impression vs. shooting for additional absolute resolution and sharpness. Nikons step is based on the “facts” that a) most shooters can’t leverage and don’t apply the proper techniques to achieve maximum sharpness, b) customers get more and more sensitive for noise and c) that lenses become more and more a limiting factor for modern high resolution sensors anyway and d) diffraction limits approach a practical upper boundary.
Another dimension is the maximum used output size of its customer base which diminishes the benefits of high resolution sensors anyway. Market differentiation based on noise suppression, not on resolution anymore.
here is the table No1, mentioned above
While creating the files for the table above, it was impressive to see the size difference of the different files. By that I don’t refer to the differences between the different models, but the size difference when ISO settings are all the way ramped up from base setting to its highest value. So, I looked into that deeper.
It is based on the assumption, that noise is a form of distortion caused by the sensor and its electronics. This kind of (useless) data has to be stored somewhere – hence the file size increases. Sure. But how much?
The 2 tables below show exactly that for all 4 cameras.
Table No2: D7000, D700 and D300s were set to 14bit and lossless compression, the D90 didn’t have this feature, was set to 12bit and lossy compression. The right side of the table displays the relative size increase of a RAW file vs its base ISO level size. The green bar displays the base ISO level.
Table No3: To mitigate the influence of different file formats and compression algorithms, I ‘ve rerun the test with identical settings for all cameras – 12bit and loss compression. I do think the second table is the better one for comparison, but I had to start somewhere .
The recorded image was a dark frame, taken at 1/125sec and NR set to OFF,
Table No2 and No3:
Lots of fun and I learned a lot
Nov 14th, 2010
Question: Could you please explain how to “read” the tables in the article above?
Sure, let my try to explain what I intended to do with these tables. While capturing all the dark frames to extract the NR settings, this pattern of increasing file sizes got my attention. You wont be able to see this in these tables but there is a difference of a couple of Mbytes at highISO images with different NR settings (off, low, norm, high). MY only possible interpretation for this behavior is the smoothening function of the NR algorithm. HighISO files with NR set to OFF are considerable larger than those where NR is set to high. NEF files contain different partial data types (metadata, JPEG preview, RAW data), so be careful about the absolute values in these tables.
Some patterns are noticeable:
- The D7000 files don’t grow in the first 3 ISO settings, the files of all other bodies do. The D7000 start at ISO 100 and still have only 1% more “data” at ISO 800 (3 levels difference). All other bodies start at ISO 200 and have 5-13% larger files at ISO 1600 (again 3 levels).
- This could be interpreted the other way around as well – negatively for the D7000. The noise level at base ISO is so high (naturally or artificially) that there is less increase at higher ISO levels.
- The “bad guy” seem to be the 2nd op-amplifier, which is supporting the main (variable) op-amplifier. Basically all Nikon D-SLRs add this 2nd amplifier somewhere between ISO 800 and ISO 1600 settings. You can check this easily with i.e. Irfanview. GainUp = OFF, GainUp = LOW, GainUp = HIGH. Typically the first stage (LOW) is added at ISO 320. This is in my opinion the root cause for higher noise levels above ISO 800.
- If you really would like to check the sensel values before the demosaic process, you might have to use dcraw with the “-d” option.
There is another interesting aspect between noise and sensitivity. How does the camera “assemble” an image with higher ISO settings? The pure sensel is only capable of delivering a signal at base level, amplification happens thereafter.
That’s the most likely sequence (not all cameras behave identical):
- Sensel – delivers an analog signal at base level (ie ISO 100)
- 1st amplifier: fixed amplifier with 3 ranges. ISO 100-320 = OFF, ISO 320-800 = LOW, above = HIGH (these are sample D7000 values, might vary with other cameras)
- 2nd amplifier: responsible for all steps within the range ISO 100-3200 (if the first amplifier is set to OFF), other ranges appropriately.
- Sampling of the analog signal to a digital value. It looks like that the D7000 has a true 14bit A/D stage. The D300 had a true 12bit stage which measured longer or more often to create a statistical averaged 14bit value (and lower shooting speed at 14bit).
- If the camera is set to Hi1 or Hi2, another step comes in. Both amplifiers are already at their maximum values. According to Emil Martinec, the Hi1 and Hi2 amplification is applied after the A/D step to the digital value with clipped highlights and reduced dynamic range. Basically, Hi1 and Hi2 can be applied either in camera or in CNX2 – it’s the same process.
More details about noise can be found in the excellent paper by Emil Martinec – highly recommended reading.
Nov 14th, 2010
There had been intense discussions in internet boards about sharpness, or rather the lack of sharpness of the D7000. Similar to the further development of the camera technology, there seems to be a need for further development of the photographers to understand and capture the increased capabilities. I leave it to the reader to see this trend as positive or negative.
The term “careful work” has been mentioned quite often and I’d like to take this opportunity to present my personal interpretation of this term in a kind of checklist. It is of course not always necessary to apply all steps, but if your goal is to get images with maximum sharpness, it might be beneficial to apply some of those recommendations below. The more aspects we have under control, the less likely we will end up with frustration.
Here is my list:
- Use of a stable tripod and an appropriate head. Has been discussed in epic dimension, but interesting to see how often it is ignored. “Hey, I had 1/1000 sec shutter speed, my hand hold picture has to to be sharp”
- Check out if your tripod has rubber feets or is based on a vibrating floor. Especially high res cameras like the D7000 and D3x suffer here, even more when not working with mirror lockup. Using a timer normally does not address this vibration.
- Use a remote control. Don’t mind if infrared or cable based. If used with mirror lockup, a key requirement. Using a timer is no replacement.
- Mirror lockup. It is great that Nikon equipped the D7000 with MUP. Let the whole system settle for some seconds, or wait until the wind settles. For top end performance, the D7000 is critically dependent on this feature.
- Software pipeline. As described above, Nikon chose quite useful settings for the average user. Unfortunately, they are not optimal for sharpness enthusiasts and have to be set accordingly.
- For JPEG users. Don’t forget to set JPEG fine/large. JPEG compression to optimal quality. Turn off noise reduction.
- For RAW users. Don’t forget to turn off NR
- Turn off Auto Distortion Control in the camera and CNX2 – especially with kit lenses with strong distortion this software transformation reduces sharpness.
- Turn off VR – every time. I know Nikon recommends it sometimes differently, but imho I’d turn it off.
- Check your environment. If your camera or lens is exposed to strong wind – bad luck. If your tripod is placed on a parquet floor or heavy trucks are passing by, you might be impacted as well.
- AF fine tuning. Increased capabilities and resolution require higher accuracy. Great that Nikon equipped the D7000 with AF fine tuning. If you are really going hard on extreme sharpness – use it. Big benefit of the D7000 vs the D90.
- If someone is interested in sharp corners then care about the parallel planes between your object and the sensor in your camera. Especially with wide angle lenses this can be sometimes tricky.
You might have seen, that I did not recommend any specific lens. There is a simple reason. If most of the items listed above are considered, many sharpness issues are resolved and most “complaints” vanish. But using better, higher quality lenses of course might help as well.
The reverse argument, that a sharp lens like the AFS 60mm micro with a high resolution body like the D7000 has to produce sharp images without considering or applying partially the above recommendations is not valid.
Nov 20th, 2010
Just wanted to share the difference between OOC JPEG and RAW with one photo taken today in the garden. The picture was taken with the AFS 10-24mm/4 handhold – really not the sharpest image, but that does not matter. It is an example how big the difference between JPEG and NEF with default settings is. The D7000 stored the image in Large/Fine and RAW. Noise reduction was set to LOW and distortion control set to ON.
Left you see a 100% crop from the SOOC JPEG and on the rights side one derived from the RAW file (NR and distortion control set to OFF in CNX2). USM has not been used in both images.
As usual, if any color channel is blown out, it is most of the time the red channel. I’ll look at it later, don’t know what causes this.
100% crop, JPEG left, RAW right
One additional comment. The AFS 10-24mm cannot keep up with the resolution of the D7000 in the corners. Even with f8 and independent of the distortion control setting.
100% of the lower left corner
Nov 21st, 2010
Used today the D300s and the D7000 together with the AFS 17-55mm72.8 and the AFS 70-200mm/2.8 VR I. The handling of the D7000 with the AFS 17-55 is excellent – I really like it.
With macro lenses the D7000 is capable of producing very sharp images, better than the D300s. Today, using both zoom, I was not able to get the D7000 work at least at the level of the D300s. Looking at the images in 100% there seemed to be little haze on all images – don’t know where it is coming from. Clear victory for the D300s.
Let me explain with some samples. I though that the 17-55 is an ideal companion to the D7000. It’s fast aperture allow excellent subject isolation and its range covers many typical scenarios. edge sharpness is better than with the 10-24mm. It is just this little haze which is so irritating.
Disclaimer: You won’t see much in web sized images, but printed in A4 I wouldn’t be so sure. I can see it clearly on my 30 inch monitor.
Same situation with the AFS 70-200mm/2.8 VR (Typ I) – one of my most often used lenses, which usually excel on DX bodies – too bad.
The images below are based on the RAW images, NR, USM turned OFF. Both bodies (D7000 and D300s) had the following settings: HighISO = off, distortion control = off, picture control = standard. All pictures were taken from the tripod and remote control and mirror lock-up. EXIF data are in the images.
For better comparison, all 100% crops are in this sequence:
- D7000, in original resolution
- D300s, in original resolution
- D7000 resized to be as close as possible to the output size of the D300s
The scenery, only resized. Some clouds passed by the sun, which caused some color shift, but did not disturb the results.
D7000, 100% crop, original size
D300s, 100% crop, original size
D7000, 100% crop, resized to fit the D300s output size
This crop is from the center focus point
D7000, 100% crop, original size
D300s, 100% crop, original size
D7000, 100% crop, resized to fit D300s output size
Surely, every one is entitled for his own opinion, but imho the D300s images are significantly clearer and more defined. Can’t currently explain why this is the case. It could be one of the following:
- The D7000 is more sensitive to the light path of zoom lenses used here
- The lenses might hit the wall in terms of their resolution abilities
- AF setting of the D7000 with my copies of the lenses need fine tuning
- The AA filter of the D7000 is stronger than on the D300s
- May be Nikon used cheaper micro lenses with lower quality for cost reasons in the D7000
- The quality of the pixel read out might be different (quantization noise)
- The sky looks much more smooth in the D7000 images (software?)
To be on the save side, I repeated the D7000 images multiple times – to avoid single image mistakes. I did not improve the results.
Nov 26th, 2010
Looked again at the AFS 17-55mm/2.8 images. I could not find any image with better sharpness than this.
100% crop @24mm, no USM in CNX2
Gee, the AFS 17-55mm lens is such a nice lens on the D7000. Just thinking about subject isolation. I took the combo to Dublin for a night/day trip. Below are some shots taken after midnight.
Some images are available as full size version – click here.
Handhold, Auto ISO, ISO 2200, 1/30 sec, f5.6
(image #798) ISO 3200, 23mm, 1/25sec, f5.6, hand hold
100% crop. printed @150lpi this picture would be approx 20×30 inch
(image #801) ISO 100, 3 sec, f5.6, mini tripod, mup
This part of the city was lit up with sodium-vapor lamps (here it was again – my issue with the blown red channel)
(image #803) 10 sec exposure time, f5.6
These sodium lights are a challenge for the sensor. Can’t say if the blooming would have happened with other camera’s as well – didn’t have one with me …
Subject isolation with this combo
Of all the lenses I had a chance to use with the D7000, the AFS 17-55mm is my favorite. Don’t know what happened in the first run, will look at it later.
I’ve captured photos with 30sec exposure time. AF never had any issues. Without any problems and side effects it did, what is supposed to do – focus properly.
Nov 26th, 2010
We had a clear sky in Vienna. The moon was just beautiful. I took my AiS 1200mm/11 lens, mounted it with the D7000 on 2 tripods. Could not get sharp images – mad, mad. The solution was to dampen the whole setup with a heavy blanket. The D7000 body is just too light. BTW, no color aberration on the D7000 image with this 25 year old lens. Fantastic.
For those who want it as 1920 x 1200 wallpaper – click here.
Minimal crop, resized
Nov 27th, 2010
the last 3 weeks had been an entertaining and insightful time. With the introduction of the D7000, Nikon went a significant step in the right direction. New sensor, new AF module, a whole chunk of Pro features in the top-of-the-line consumer camera is less a technical challenge, but rather a business challenge. At the end of the day, Nikon intends to sell way more of D7000 bodies than its Pro bodies and expect that the D7000 line contributes to its overall margin – no simple feat.
How Nikon attempted to bridge these different requirements was for sure also one reason, why I wrote more than usual about my interactions with a new body. Of course, some curiosity was there as well, how this new camera will fit into the existing and upcoming product lineup.
Today, my picture counter hit the 1.000 mark, a good reason to end my “first impressions” series here. As I am not a professional reviewer, nor a professional photographer, I’d like to ask all interested readers to take their own impressions for their personal and final decision making pro/contra D7000.
Today, I had a very special goal – enjoying my day and using the D7000 as my sidekick to capture images. More and more my default lens on the D7000 seems to be the AFS 17-55mm. Its size and weight stabilizes the small, and light camera. Just perfect. In my view, this is one of the biggest drawbacks of kit lenses – too light. I also like the images this combo delivers. It just fits.
The new AF module is a step, exposure metering a big step forward in the right direction. Well done. Matrix metering is mostly spot on (better than the D90). I used to have standard exposure compensation set to –0,7 EV, now it is –0,3 EV. The D7000 matrix metering puts more emphasis on the focus field versus its predecessors, but this usually not a big issue after getting used to it. I’d like the 2 WB auto settings as well- cool idea.
After my initial hassle with the AF setting, changing from AF-S to AF-C is quick (If you need reading glasses, then you really need them for the super small icon on the top display). But honestly, I like the 3-step dial on the pro bodies better.
Don’t know why Nikon decided to have ISO settings only changed in 1/3 or 1/2 but not full stops – very inconvenient. The mode dial gets quite often changed when the camera is taken out of the bag – it could be a bit stiffer. Changing from M to A to P is quick and fast.
Noise with the D7000 increases non-linear:
- From ISO 100 until 320 – first amp is OFF
- From ISO 400 until 800 – first amp is LOW
- From ISO 100 until 6400 – first amp is HIGH
- Hi1 and Hi2 are pure software driven sensitivities
If your shooting situation allows this flexibility, use ISO 320 and ISO 800 more often than other settings.The D7000 seem to like some older AiS lenses – I’ll gradually cover some common lenses in the future.
To finish off, some pictures. As usual, high resolution versions can be downloaded here (use the image numbers to select)
(#905) We had snow in Vienna. My dog tried to be the color-key in this landscape.
(#894) It is impressive how well the D7000 handles cross light, preserving details in shadows and highlights quite long. The AFS 17-55mm keeps its sharpness even wide open.
(#885) ditto @55mm. Please check the resolution in the full size image in the link above
17mm and f2.8
(#962) 45mm and f5.6. Distance about 2 ft (0.5m)
100% crop. decent quality
(#975) used AF-C for the first time. The horse was surprisingly fast.
(#1001) used together, 17mm and high resolution is quite nice in case you’d like to crop
as seen here – 100% crop
With that, I’d like to thank you for keeping up until the end and I hope you enjoyed my early journey with the D7000 and found the content a bit useful.
End of my initial impressions
Dec 28th, 2010
Time warp. Winter vacation in the Austrian Alps. Did translate this blog last night. During daylight, it was just beautiful.
Had to walk the dog again. Took the D7000 and some glass with me. More photos are here.
AFS 10-24mm/4. Used DL a bit in CNX. lens shows sometimes flare and ghosting
100% crop. nice detail preservation
The AiS 180mm/2.8 is a gem