ISO Shoot-Out: 7D v. 5D Mark II v. 1Ds Mark III v. Phase One P25+ v. 1D Mark IV

just kidding only the first 4

Ed. Note:

Just as we were about to publish this post, the Canon 1D MarkIV was announced. By all accounts, it has a 2+ stop noise advantage over the 5D resulting in a functional sensitivity beyond human vision.

The creative possibilities ultra- low light cameras open are very exciting. We look forward to the opportunity to test this out this thrilling new addition to the pro shooter’s arsenal, and will share our results just a soon as we can.

By Mark Rutherford and John Champlin, LUX-SF, Inc.

The Set-Up

Since the release of the Canon 7D last month, we’ve gotten lots of questions about its merits relative to the 1Ds Mark III and 5D Mark II.  There has been a ton of excitement about this new camera, and just as much confusion, possibly brought on by their seeming abandonment of their own naming conventions: not just a successor to the 50D, the 7D is like the 5D, only 2 better… right?  The promise of being able to shoot high-megapixel still images and 1080p cinema on the same camera, for under $2000, has got the whole photo community giddy.  But how much of a value does this new camera truly represent? Does its output stand up against that of a camera $1000, $5000, or even $15,000 more expensive? In the right hands, could the 7D be anywhere near as good?

Download uncompressed hi-res files here

Thus questioned, when we got our production model of the 7D, we figured it was time for some testing. Accordingly, we called our good friends at Pro Camera Rentals in San Francisco and schlepped on over to put 4 cameras through the paces: the 3 mentioned above, plus a Phase One P25+ for good measure. We shot all cameras on a Horseman view camera with a Schneider Digitar 90mm lens; processed the RAW files in the latest software of their respective manufacturers: Digital Photo Professional 3.7.1.2 for the Canons, and Capture One 4.8.3 for the P25+; and evaluated the cooked TIFFs back at the studio.

The Question

We decided for the purposes of this test to confine ourselves to the question of how each camera performed at different ISO settings, which is to say, “How noisy do these photos and movies get as we raise the ISO, and what effect does this have on their resolution?”

Results

All of the Canons are functionally noiseless at speeds up to ISO 800, that is to say the amount of noise in the files would be undetectable for most forms of reproduction. From ISO 800 to ISO 3200 some additional noise reduction would likely required in post, particularly if the file needed to be enlarged. At ISO 6400 noise is noticeable at most magnifications with the 5D showing the least and the 7D having the most noise. While the 7D shows higher noise than the other Canons, its files are still quite usable up to ISO 3200. Surprisingly the movie files appear very similar between the 5D and the 7D with the 7D fairing nearly as well in overall image quality.

The Phase back is virtually flawless at ISO 100, distinctly cleaner than the 5D when viewed at 100%. It exhibits excellent rendering of the finest details before any sharpening was applied. In fact the Canons needed some pre-sharpening to come close to the Phase and even then the very finest details, such as the smallest type on the red camera in our test, was soft compared to the Phase. At ISO 400 the Phase begins to show some odd banding in certain shadow transitions. This is apparently an artifact caused by the noise reduction method in Capture One.

Overall color fidelity of the Canon files is very good when processed under the “Faithful” setting in DPP. The newer Phase backs coupled with Capture One offers greatly improved color accuracy over Phase’s older software and matched difficult colors such as the wood tones and saturated blues in our test files better than the Canons.

Viewing

We have posted all of the full-res TIFF files, and our original MOV files in a wide range of ISO’s; please download any you’d like to review through this link. Also, here are some recommendations:

- View in Photoshop or other color-managed image editing program.
- View on a calibrated display in a dimly lit, color-neutral environment.
- To approximate what you would be likely to see on press or other printed media, view the still files at 25%.

Tech Notes

- The motion files were shot under the Canon “Neutral” picture style with Sharpness, Contrast and Saturation at their lowest settings. Noise reduction was set to “Low”. Please note that the visible noise in the movies will be more apparent in this test then you will likely see in typical use due to the lack of camera or subject movement. Since the camera is locked down and the subject is motionless there is nothing to distract your eye from the subtly moving noise. Ideal for a test comparing camera noise but exaggerated compared to most footage.
- The full resolution still files were processed with the “Faithful” setting in Canon’s DPP and “Extended shadow detail” in the Capture One software which gave the most similar results between the Phase and Canon files with no further post processing. All Canon files at 200 ISO and above have “Highlight Tone Priority” enabled for extended highlight detail. Noise reduction was at default settings. No sharpening was applied to the RAW files.
- To account for the 7D’s 1.6x focal-length multiplier, we simply moved the camera away from our set and re-framed to approximately the same crop.
- Both applications automatically apply a higher amount of noise reduction as the ISO is increased; when a higher ISO setting is selected, the NR control also shifts. We did not interfere with this when we processed our images.
- The sample JPEGs below are crops of the ISO 100 shots @ 100% showing a small detail of the original file. A small amount of pre-sharpening has been applied (with Photoshop’s Unsharp Mask filter) to the Canon files to approximate a real-world usage and to allow for ease of comparison. (The hi-res, downloadable files have no sharpening at all applied.)
- As noted, we chose to use the same high-resolution lens for each camera in order to eliminate optical quality as a variable.
- We acknowledge that some of the higher ISOs Mark III files have a slight motion blur which we did not catch at the time of capture; they are nevertheless useful for evaluating noise.

7D- ISO 100

5D Mark II ISO 100

1Ds Mark III ISO 100

P25+ ISO 100

Conclusion

Perhaps not surprising for a camera that costs twice as much as the rest of them put together, the Phase P25+ at ISO 100 is the clear winner in a pixel peeping contest at 100% magnification in Photoshop. On the other hand, viewing at 25% is a more realistic representation for the majority of reproduction at normal viewing distances. When evaluated at 25% in CS4 the differences in overall quality among all four cameras up to ISO 400 are negligible.

As the competence of our tools increases, the choice of which to use becomes less obvious, and becomes more a question of which feature set is most appropriate and what level of quality does the final application require. All of the cameras in this test are professional instruments capable of meeting a wide range of requirements, with the nod for these criteria going to the 5D Mark II for its superior low-light performance.

Do you have any questions or comments on this test or suggestions for other tests? We’d love to hear from you; mail us at studio@lux-sf.com.

Download uncompressed hi-res files here

All test files © Lux Digital Imaging Services Inc. 2009.  Usage is permitted with link to: www.lux-sf.com

All editorial photos © 2009 Robert Render Harrison (with his 5D Mark II), except group shot, © 2009 Keith Ross