Following are the pair of 100% crop image quality comparisons I have been including in the recent Canon EOS DSLR Reviews. Read about the Camera ISO noise tests in the help section to learn more about the tests and how they are conducted. A key take-away from that page is that noise reduction is completely off unless otherwise specified.
There are many MB of files required to be downloaded to make all of the mouseovers and mouseclicks on this page function properly - please be patient while they load.
While this comparison is primarily designed to show noise (and color), I am again struck with the obvious difference in the T4i's sharpness compared to the other APS-C bodies including the T3i. Results for the 1D X are not included here, but the T4i compares similarly in sharpness to this high end 18.1MP DSLR. Has Canon begun normalizing image sharpness across the EOS line?
The next obvious difference in the comparison above is an unusual one - the newer EOS model has more noise than its predecessor, the T3i.
The T3i delivered a slightly brighter image (using the same camera settings) that makes comparing results slightly more difficult. I could normalize the brightness between the two, but changing brightness affects apparent image noise. Comparing the T4i's results with the 60D results is also interesting - and the brightness from these two cameras is similar.
I definitely welcome the T4i's increased sharpness, but increasing sharpness often increases the visibility of noise. The set of T4i Sharpness=0 results were obviously processed with a sharpness setting of 0. These results are very similar to the T3i's results, both in terms of sharpness and noise levels. Essentially, the T4i's image quality matches the T3i's when the same amount of sharpening is added to both. The T3i has great image quality, so matching image quality is good.
In-camera noise reduction is standardly available in EOS DSLRs, and noise reduction is of course also available during post processing. The bottom three rows of results in the comparison above (and below) show examples of T4i noise reduction. All three NR examples utilized the Standard Picture Style in-camera with the default sharpness setting (so that the RAW NR results would match the JPG NR results). The default sharpening applied to these results examples the destructiveness of sharpening (look for halos around the low ISO color block examples). The noise reduction process is also destructive - I usually apply only light NR in my noisier images.
Multi-Shot Noise Reduction is new for EOS - and I was quite excited to see how the merging of multiple (four) exposures taken in a full-frame-rate burst could be used to reduce image noise. The concept makes a lot of sense. The bottom two NR examples provide a direct comparison with the standard NR and the Multi-Shot NR. There is definitely improvement with MS NR - a full stop or more at some ISO settings - including lower ones.
Some of the downsides to Multi-Shot Noise Reduction include: MSNR is currently available only with JPG output - I want this feature to be added to DPP for RAW capture - perhaps as another HDR preset? Multi-Shot Noise Reduction will not be so useful with moving subjects. Long exposure NR must be off to enable MSNR. The T4i reverts to Standard NR in Basic zone modes, during video recording, in Bulb mode and when the camera is powered off. Flash is not supported in MSNR mode. And the camera remains "busy" for a brief period of time after the 4 shot burst - while processing the merged image.
New with the T4i is highest-ever-in-a-Canon-APS-C-DSLR ISO setting of 25600. As I see it, this setting is useful for marketing purposes only. Just my opinion of course.
ISO 12800 results are still ugly. I avoid ISO 3200 and above on APS-C DSLRs right now, but 3200 and 6400 can be used for extreme situations - or when Multi-Shot Noise Reduction is used.
Here is another noise comparison that includes fine details in a piece of fabric.
The fine details in the fabric better-hide high ISO noise.
Back to the Canon EOS Rebel T4i / 650D Review.
Bringing you this site is my full time job (typically 60-80 hours per week). Thus, I depend solely on the commissions received from you using the links on this site to make any purchase. I am grateful for your support! - Bryan