Article – Comparing image resolution of 21 Nikon D-SLR’s
Ages ago I did some comparisons with the Nikon bodies I had at that time and compared their ISO quality in all settings. Was fun and insightful. In the spirit of this old habit, I’d like to share another comparision I did a while ago.
Hadn’t we heard the following statements quite often, or used it some times by ourselves?
- ”The new model has xx% more megapixel. I’ll wait for that.”
- “6 megapixel are more than enough. Everything above is non-sense.”
- “The xx-model has double the resolution. It is truly unbelievable.”
- “I’d rather take a pro-body. The quality of the pixels in pro-bodies is way better, true.”
- “FX is better than DX – or vice versa.”
- etc, etc, …
To put some meat into this debate, I’ve created 21+1 RAW files for everyone to personally download and analyze. Of course, only if you are interested. As the idea came up during garden work in autumn, I took some autumn foliage, put it in a box und used the studio flash to create a setup which is somehow repeatable between the different contenders.
- I tried to create a compareable point of view for DX and FX bodies and also intended to reduced the labour to change bodies. WIth this idea, I took the AFS 70-200mm/2.8 VR II and mounted it on a sturdy tripod. The AFS is currently one of the best and sharpest zooms in Nikon’s Nikkor line. THe focal setting for DX bodies was 105mm and 175mm for FX bodies. Doing the math, it doesn’t equal the 1.5 crop factor, right? AS my room is not infinitly long, I had to adjust for the typical IF lens shortening in closer ranges. Hence 175mm is 105mm times 1.5
- I stayed away from both extremes on the short and long end of the lens.
- Aperture was set to 11 and exposure time was set to 1/125 sec. First, first to keep the lens in its aperture sweet spot and secondly to stay away from diffraction limitations (especially with the D3x and D7000).
- White balance was set to flash on all bodies
- Each body was set to its nominal ISO setting, either ISO 100, 125 or 200. Compensation was done with the studio flash.
- Self timer was used oin all bodies. To be able to compare so many bodies, I’ve conciously avoided to take advantage of a remote and mirror lock-up with those bodies where available.
- All RAW(NEF) files are stored in their best setting. 12 or 14bit and lossless compression or no compression at all.
- There were 2 light sources. On the right side I used a 5ft (1.5m) Octabox and on the left side a honeycomb reflector to increase contrast a bit.
Your (real world) usage
It has become fashionable to watch 100% views of images on the web. Sometimes it makes sense, but I would argue that most people dont view their pictures in this way. My recommendation with the images provided is: Download the images of the camera models you are interested in, create an output file in the required resolution and compare THEN the difference. There is simple no point in comparing a D3X image with a D1H image on a 1280×800 screen, but you WILL see a difference on your 30 inch screen.
In “violation” of my usual habit, I decided not to post any crops of images produced by the cameras. The bodies produce images with significant different resolution. For instance is the difference in resolution between aD1 (2.7 MP) and a D3x (24.5 MP) almost 10:1.
To resize crops in a way that all comparing images on screen have about the same size would require significant up/down sampling of the original images. Either images from lower resolution cameras would display huge corn (and noise) or high resolution bodies could not demonstrate their superior capabilities.
Another approach would be to display 100% crops in native resolution. Cutting the crops to identical size would impact the object size in the image (in our case, the leaves).
Yet another approach would be to crop with all cameras one object in the image . Given the fact that we have to cover a ratio of 10:1, I did not use this approach as well.
Given all the different views, opinions and requests in communities it somehoy looked like a loose-loose proposition to choose either one over the others. So, I took the simple way forward. No cropped images. I’ll make the original – completely unprocessed – NEF files available – to give everybody the chance to take his/her favorite approach, how to compare 2 or more bodies.
The potential observer might be suprised how small the difference between the different camera models actually is, when used in similar conditions – like identical lens, identical light and a somewhat repeatable setup. Provocatively stated, someone could question the size of progress in this area in the last 4-5 years. “How much resolution should I really look for?”
Looking at the images, my first reaction was that sensor resolution has lost its role as leading indicator of qualifying camera models. Going forward, other camera attributes will get more importance. On the technical side – lens quality are much more important than yet another few MP on the next generation sensors. Other gimmicks will get more attention – the AF module, AF precision, High ISO noise and usability.
Since the introducation of the D1 in 1999, Nikon released 27 digital SLRs as of January 2011. My personal categorization with regards to resolution would be as follows:
- Category 1: Everything between 2.7 and 4 megapixel (4 models: D1, D1H, D2H, D2Hs)
- Category 2: 6 megapixel ( 5 models: D40, D50, D70, D100, D70s)
- Category 3: Everything between 10 and 14 megapixel (18 models)
- Category 4: 16-24 megapixel ( 2 models: D3x and D7000)
If you are interested in all technical details of available bodies, please check out the excellent and comprehensive table at nikonians.org. 6 bodies are missing in this comparison. 4 were borrowed out to family members (D40, D70, D70s, D100) and I dont own the D2Xs and D2Hs.
(Note: When the original comparision was done, the D7000 was still not out. When it became available a week later, I used the existing box of leaves – aged by a week. )
- Nikon D1 (2.7 MP, CCD. No defined color space. The AA filter seems to be four times thicker vs. the D1H, less acuity)
- Nikon D1H (2.7 MP, CCD, Same resolution as D1, but much sharper impression)
- Nikon D1X (5.3 MP, CCD, Such an excellent old workhorse at base ISO level)
- D2H (4 MP, LBCAST, Viewed on a 30 inch display, there is a signifcant visible difference to the images of the 6 MP class)
- Nikon D2X (12.3 MP, CMOS, one of the bast AA filters around – why did Nikon discontinue this filter?)
- Nikon D3 (12.4 MP, CMOS, solid performance, good graduation in the shadows)
- Nikon D3s (12.4 MP, CMOS, imho identically to the D3 in this test)
- Nikon D3X (24.5 MP, CMOS, This setup is its home turf – a class in its own)
- Nikon D40X (10.2 MP, CCD)
- Nikon D50 (6.1 MP, CCD)
- Nikon D60 (10.2 MP, CCD)
- Nikon D80 (10.2 MP, CCD)
- Nikon D90 (12.3 MP, CMOS)
- Nikon D200 (10.2 MP, CCD)
- Nikon D300 (12.3 MP, CMOS)
- Nikon D300s (12.3 MP, CMOS)
- Nikon D700 (12.1 MP, CMOS)
- Nikon D3000 (10.2 MP, CCD)
- Nikon D3100 ( 14.1 MP, CCD; Interestingly, the camera has a base ISO level which can be set to 100, but the real ISO sensitivity is higher – about half a stop versus other bodies. I checked this multiple times, but my current view is that the D3100 has a different effective ISO base setting – approx ISO 160. In this scenario the D3100 exhibited a more limited dynamic range. Check out the lost highlights and lowlights in Capture NX vs. the D200.)
- Nikon D5000 (12.3 MP, CMOS)
- Nikon D7000 (16.2 MP, CMOS, sorry but the leaves are a week older, yet the resolution is quite impressive – never use JPEG out of camera with default settings for a desire of high resolutiuon images)
- Nikon P7000 (10MP, outside of “competition”. Had to use it with ambient light, doesn’t trigger the pockt wizard)
- Please consider sample variability. Don’t conclude from one (my) camera sample to all bodies of this type.
- My setup might be faulty or incorrect
- The results provided are only one aspect of a complex chain of elements and processes. Don’t conclude from one image to a broader assessment of the capabilities of a camera
- Don’t forget that the spectral distribution of my flash light is quite homogenous. Sensors of different type, react differently with other, more limited light sources.
- Thank you, for haveing read until the end. Please follow this link to all mentioned RAW(NEF) files. Enjoy downloading and assessing it. I tried to be as careful as possible in creating them.
This is the scenery – fresh autumn foliage