There is a certain website, I won’t say any names, that is known for it’s dynamic range tests. It is really one of the few websites that consistently comes out with these Dynamic Range Test Scores for new cameras, and something about it has always driven me absolutely bonkers; the do not publish their methodology, which in the land of science is just craziness. If you don’t publish your methodology, how in the world can your peers scrutinize what you are putting out there? How can the results be duplicatable or confirmed? They can’t. HUGE problem. I do wayyyyy too many little tests that I shoot down and do not actually talk about or publish because I do not think the method is good. If I can’t publish the methodology, in that sense, the test is basically meaningless because I could just be making stuff up and no one would know the better. Thats my soap box about that certain website for now at least.
For years I have been trying to figure out a reliable and cost-effective way to measure a camera’s Dynamic Range, on my own. I have seriously tried just about everything and while I sometimes had interesting results, the method was either not convincing enough, too expensive, too hard to replicate or too complex. I was looking for something anyone could replicate with minimal equipment, expertise and cost.
Ive been working on and tweaking this one technique..yes, there are still some problems with it that Ill address some below, but suffice it to say, this is how it works:
Maven Dynamic Range Test Methodology:
1. I purchased a $14 transparent strip from a printing calibration company called Stouffer . The strip has 41 increments, of what is essentially 1/3 stops for each increment. It is calibrated and used in printing processes. This particular strip has about 13.7 stops of dynamic range.
2. The strip was installed into a black poster board that I could hang in front of an 1600 Alien Bee, powered at 1/32
3. With my test cameras, I framed up the strip close to edge to edge and did some test shots testing different apertures until I was essentially clipping out the first interval on the preview.
My exposure settings are 1/10, f10, ISO 100 and the camera is on a tripod triggered by a timer, and the flash is triggered by a pocket wizard.
4. I then import the files in to Camera RAW, again, intentionally clipping out the first stop. I then inspect at what point the borders between 2 adjacent stops are no longer visible.
A few of the problems with this technique:
1. It is not a great technique to measure overall dynamic range, there is some more head room on the bright side of the strip that can be recovered. Im not exactly sure where this is, but I do know that the strip is limited to 13.7 stops, so I also have to keep this in mind. I do believe that with more testing, Ill eventually figure this out.
2. Another problem is that once noise starts coming into play, it is difficult to determine at what point noise renders the dynamic range measurement irrelevant, as you will see below. Not really sure how to calibrate a signal to noise ratio for this that would give me a reasonable, consistent cut off point.
I did get my hands on a brand new Canon 7Dii and put it through this little test, comparing it with the Nikon D750
While Adobe Camera Raw doesn’t support the Canon 7Dii at this time, there is a little free app called “Raw Therapee that will allow you to access the RAW files of both the 7Dii and the Nikon D750. (You can download the latest Beta version of Abobe Camera RAW 8.6 candidate for D750 RAW Support in Camera Raw: Camera RAW Plug-in For Nikon D750
Some of you will wonder where the heck all this noise is coming from. Its from stretching the RAW files as much as possible with things like Shadow Recovery. Otherwise, the cut off point is much sooner (because the swatch is so dark).
Take a look for yourself:
At what point can you no longer differentiate between 2 adjacent stops?
At what point does noise ruin the measurement?
Here is the Canon 7Dii:There is a certain website, I won’t say any names, that is known for it’s dynamic range tests. It is really one of the few websites that consistently comes out with these Dynamic Range Test Scores for new cameras, and something about it has always driven me absolutely bonkers; the do not publish their methodology, which in the land of science is just craziness. If you don’t publish your methodology, how in the world can your peers scrutinize what you are putting out there? How can the results be duplicatable or confirmed? They can’t. HUGE problem. I do wayyyyy too many little tests that I shoot down and do not actually talk about or publish because I do not think the method is good. If I can’t publish the methodology, in that sense, the test is basically meaningless because I could just be making stuff up and no one would know the better. Thats my soap box about that certain website for now at least.
For years I have been trying to figure out a reliable and cost-effective way to measure a camera’s Dynamic Range, on my own. I have seriously tried just about everything and while I sometimes had interesting results, the method was either not convincing enough, too expensive, too hard to replicate or too complex. I was looking for something anyone could replicate with minimal equipment, expertise and cost.
Ive been working on and tweaking this one technique..yes, there are still some problems with it that Ill address some below, but suffice it to say, this is how it works:
Maven Dynamic Range Test Methodology:
1. I purchased a $14 transparent strip from a printing calibration company called Stouffer . The strip has 41 increments, of what is essentially 1/3 stops for each increment. It is calibrated and used in printing processes. This particular strip has about 13.7 stops of dynamic range.
2. The strip was installed into a black poster board that I could hang in front of an 1600 Alien Bee, powered at 1/32
3. With my test cameras, I framed up the strip close to edge to edge and did some test shots testing different apertures until I was essentially clipping out the first interval on the preview.
My exposure settings are 1/10, f10, ISO 100 and the camera is on a tripod triggered by a timer, and the flash is triggered by a pocket wizard.
4. I then import the files in to Camera RAW, again, intentionally clipping out the first stop. I then inspect at what point the borders between 2 adjacent stops are no longer visible.
A few of the problems with this technique:
1. It is not a great technique to measure overall dynamic range, there is some more head room on the bright side of the strip that can be recovered. Im not exactly sure where this is, but I do know that the strip is limited to 13.7 stops, so I also have to keep this in mind. I do believe that with more testing, Ill eventually figure this out.
2. Another problem is that once noise starts coming into play, it is difficult to determine at what point noise renders the dynamic range measurement irrelevant, as you will see below. Not really sure how to calibrate a signal to noise ratio for this that would give me a reasonable, consistent cut off point.
I did get my hands on a brand new Canon 7Dii and put it through this little test, comparing it with the Nikon D750
While Adobe Camera Raw doesn’t support the Canon 7Dii at this time, there is a little free app called “Raw Therapee that will allow you to access the RAW files of both the 7Dii and the Nikon D750. (You can download the latest Beta version of Abobe Camera RAW 8.6 candidate for D750 RAW Support in Camera Raw: Camera RAW Plug-in For Nikon D750
Some of you will wonder where the heck all this noise is coming from. Its from stretching the RAW files as much as possible with things like Shadow Recovery. Otherwise, the cut off point is much sooner (because the swatch is so dark).
Take a look for yourself:
At what point can you no longer differentiate between 2 adjacent stops?
At what point does noise ruin the measurement?
Here is the Canon 7Dii:Here is the Nikon D750:Again both cameras shooting through the same strip, same power flash, same exposure settings.
Anyone can see that the performance of the Nikon D750 is simply phenomenal when compared directly with the Canon 7Dii, which is actually much closer to the result of the Canon 5Diii , which I have also tested:I am excited with this promising result of my new Dynamic Range Test, I know it can be improved and tweaked to get it better, so I invite your criticism, scrutiny and suggestions, Thank you!
Hello Michael ! I didnt get the results. The Dinamic Range in the Nikon is better beause of the quality of the blacks and grays?
About 1-2/3 stops for the D750, possibly 2 stops. A little less than the other website you reference which is gives about 2-2/3 stops.
While I don’t disagree with the results… I think to make the test more fair you should use the same lens. Both of them being 24-70mm f/2.8’s doesn’t really mean they’re the same. There are a lot of lenses that are made for both systems, like sigma arts or something.
Inspecting where two adjacent stops are no longer visible sounds a little suspect too… your eyes can tire… on one day, you might be able to spot the difference between 38 and 39, and on other days you can’t.
One reviewer also noted that with identical settings, under test conditions, the nikon consistently had a brighter, somehow more-exposed image. By almost a full stop. That would of course give nikon a big edge in noise and DR, if they just inherently let in more light. You might need to normalize exposure.
One last thing I’d be curious about… there seem to be two flavors of noise according to LR and every other program, luminance and color noise. Almost every program (including the NR option built into the camera) can kill color noise easily. I’d be curious to see how the results look with 100% color noise removal, and untouched luminance.