Note: Part 1 of this review can be found here.

I’m constantly impressed by the speed at which digital SLRs have progressed in making better and better images. All of the comments below should be taken in this light, let me start off by saying that it’s quite arguable that this 10-megapixel camera produces better images in many, if not most circumstances than my 4-year-old top-of-the-line Canon EOS 1Ds. (The 11-megapixel Mark 1, that is.) That is astonishing, the 1Ds originally retailed for about $7,000, this camera can be had for under a tenth of that price.

I didn’t, and don’t, do lab-quality resolution tests, but I did look carefully at many of the images I generated, and certainly in terms of basic resolution this new Digital Rebel is the equal of the 1Ds. Since I use the 1Ds consistently for my 16×11 and 24×16 fine art images–and I’m quite happy with the 48×32 image of Descent into Chaos on my fireplace, I don’t think you’ll find much to complain about with the resolution of the XTi. When resolution is no longer a barrier to good images, questions of image quality move to issues of noise levels at high ISOs, color fringing, and color reproduction, and as you’ll see below I was favorably impressed with the XTi there as well.

This unsharpened 100% crop (right) from a San Francisco skyline image (left) shows good resolution–I hadn’t noticed this flock of birds when I’d taken the image or when I first processed the image, with sharpening the birds are fairly cleanly rendered.

Noise

I took and examined a number of images from the XTi at a variety of ISO speeds, this image of Cirrus, taken at ISO 1600 can be considered pretty representative. Images were processed using Adobe Camera Raw 3.6 with default noise reduction settings and without significant exposure, contrast or saturation settings.

This 100% crop from an out-of-focus area gives a good sense of it. With typical noise reduction settings, there is a noticeable but non-objectionable grain-like noise structure. I found that some noise reduction was necessary in particular to remove chroma noise, but at ISO 1600 this is, relative to my own expectations, an impressive result. I’m fairly certain that this is a better result than I’d expect from the 1Ds at 1250, 800 and I think even 400.

This 100% crop of an ISO 200 image of sky (taken from the San Francisco skyline image above) demonstrates that noise is not entirely gone at ISO 200, but is slight and again non-objectionable. It’s my sense, unconfirmed by formal tests, that the 1Ds retains an edge in noise performance at at ISO 100 and 200, in any case, both are excellent performers when low ISO settings are used.

Fringing

As with most modern DSLRs, chromatic aberration on the microlenses used on the digital sensor can contribute to a type of artifact known as “purple fringing”, the XTi is no exception in difficult lighting, but performs well in general.

These 100% crops tell the story. The eye and whisker detail show essentially no purple fringing, perhaps a bit of chromatic aberration from the lens in the highlight in the cat eye. It’s only in the water highlights from the SF skyline, where vastly brighter sunlight shares an edge with dark water that we really see this artifact, this is the “worst case” for purple-fringing and even here it’s well under control, and compares well with the competition.

Color Rendition

I got excellent color rendition out of the Rebel XTi using ACR 3.6 in Adobe Photoshop 9.0.1. Skin tones were good, colors were saturated but not gaudy. I did see some examples of poor, no, terrible color renditions using the same RAW image files using Phase One’s usually-excellent Capture One software, version 3.7.5. It was poor enough that I was quickly convinced that I was seeing a bug in Phase One’s software, not a camera problem. While I haven’t tested the images using P1C1 3.7.6, I do note their release notes indicate revamped color processing to correct “broken colors” on the Rebel 400D (XTi) and will, based on my previous experiences with Phase One, assume that this problem is likely corrected in 3.7.6. Let me reiterate, that problem is not an issue with the camera or its use in with either Canon or Adobe software, only with the processing of RAW images from that camera using Phase One’s software.

Autofocus Performance

Auto-focus performance was good but not quite as fast as I find on Canon’s highest-end bodies using the same lenses. This image was taken with the Canon EF 70-200L/4, which I’ve used extensively on the Canon 1D Mark II N and the Canon 1Ds, and there was a difference in focus lag in favor of the high-end models. Again, please remember that I’m comparing the Digital Rebel here against cameras that cost six or more times the cost of the XTi.

Dust Protection Measures

I didn’t test this feature directly, but did examine carefully the most recent shots I’ve made with the camera, and see no sign of dust on the sensor yet. Given that I worked with the camera extensively at a dog park, I’m of the opinion that the anti-dust features on the camera are functional and helpful, I very much look forward to seeing them move into the 1-series line in time.

Conclusion

Seeing is believing, and at this point I’d be more than happy to use the XTi as a backup camera for my own landscape work, or even the only camera when weight was an issue (long hikes, etc.) I’d still likely edge toward my 1Ds II for critical wildlife work, not because of the high-ISO performance of the XTi (which is downright excellent) but because of the relatively slower focusing, but it’s not a bad performer even there. At a price point near $700, you can color me quite impressed.