50D RAW Processing Shootout: Aperture 2, Aperture 3, Digital Photo Professional, Lightroom 3 Beta

In my last post, I complained about how Aperture 2's processing of my Canon 50D's RAW files left in way too much chroma noise and how my hope that Aperture 3 would match or exceed the current champion, Lightroom 3 Beta, had been dashed.

I also promised examples, and here they are.

First up, this shot of jazz saxaphonist Russell McCray performing in a hotel bar.



Shot at ASA 400 with my Canon 50D and trusty 50 f/1.8 lens, lit by a single Canon 550EX strobe firing through a Honl snoot about 30 feet away from the stage, facing the sax player. The flash was triggered by a radio slave, which allowed me to move over to the side of the stage to shoot this photo. Not all places allow this sort of thing, but this one did.

At ASA 400, I expect to see a little luminance noise, but no chroma noise. In Aperture 2, I was seeing a lot of chroma noise - so much that I thought I might have to send my 50D in for warranty service.

And then I did a RAW processing shootout between Aperture 2 (AP2), Canon Digital Photo Professional (DPP) and Lightroom 3 beta (LR3b), and discovered that it wasn't the camera. It was AP2, and it was getting schooled by both DPP and LR3b. Especially LR3b.

This really, really bothered me. As I have said before - I prefer Aperture's way of doing things to Lightroom's. But the image quality difference was too great to ignore.

So I pinned my hopes on Apple finally delivering the much-anticipated Aperture 3 (AP3) with improved RAW processing, before LR3b expired.

And, earlier this week Apple did!

I eagerly downloaded the trial, and redid the RAW Processing Shootout. An damn it, LR3b is still better.

AP3 has visible improvements over AP2 in it's RAW processing for the 50D. But Adobe set the bar damn high with LR3b, and AP3 just doesn't match it.

The proof is in the image below. It contains 100% crops of the same section of the above photo, as processed from the RAW file by AP2, AP3, DPP and LR3b. I simply opened the file in each program, and then exported the result of its default processing as a 16-bit TIFF file. I cropped out the same section of each file and put them into this one composite image to make it easy to see the differences between the noise and detail levels. The image is really big because I saved it at maximum quality when I converted to JPEG to preserve the chroma noise detail.

My conclusion - LR3b did the best, period. Second place goes to DPP, but it also rendered the output very flat, some of which got lost in the JPEG conversion. AP3 did better than AP2, but both left more chroma noise than I find acceptable in the image.

One disclaimer - this test is only applicable to the 50D. Folks who use a different camera, especially one from a different manufacturer, might get different results from the RAW processors in this test, and reach a different conclusion about which program does a better job.

That said, here's comparison file, for your pixel-peeping pleasure.





I'll post an ASA 3200 comparison separately, later.