Being rather free after my economics paper, I took the liberty of pitting these two lenses head to head since it seems quite a number of people are interested in comparing these two lenses. Being of different focal lengths, I thought of standing further back at first in order to produce the same image. Unfortunately, my window prevented me from stepping back further with the 50mm. Hence you will realise that the centre crops are from different locations. I kept the crops at the edges to the same area instead. We all know the centre is good, so I compared the edges more directly. Here are the two original images shot wide open. Straight out of camera Jpegs, auto exposure. No corrections or adjustments.
Sigma 30mm
Canon 50mm
I used AF mode and handheld these images on a Digital Rebel XT, as I would do in real world shooting. I also left the filters on, the Sigma had a B+W Kr1.5 MC filter on, whereas the Canon had an old Original Canon 1.0X Coated Skylight filter on. I was forced to shoot through a tinted window stacked with a polariser to allow the camera to expose at 1/4000 @ Iso 100 @F1.4, so image quality may have degraded somewhat. I had considered using the MKII to test the lenses instead, as that would allow me another 2 stops of speed. However, as the Sigma vignettes substantially on the MKII when stopped down, I decided against it. In real world shooting, I have not known the filters to affect the IQ severely(having shot through the same window several times before), but this is a precaution in case it affected the results.
Looking at the test results, I found them very puzzling due to the inconsistencies. The image would degrade at f1.8 all the time, even after I removed the filters and used tripod mounting. I continued to use AF, not switching to MF. Could be a camera or user problem, I have no idea. Both lenses exhibited this behaviour, but in real world use, I have not noticed such abnormalities. Thus far, I have not used the EF 50/1.4 with my rebel XT yet, having just acquired the camera recently. In order to keep objects around the same size, I upsized the Sigma's images by about 20-25%.
100% Centre Crop
The Sigma has an added F4 because I only intended to test up to F2.8. I accidentally shot one more at f4 without realising, and so it's here. As you can see, the results are somewhat whacked. At f1.4 the centre is fair, but at f1.6, there is a marked improvement. At f1.8 the image degrades as mentioned. I have no idea why. This occured even at high shutter speeds of 1/3200 and even when mounted on a tripod with MLU and timer release. The sweet spot of the Sigma lens is at F4.
The Canon performs somewhat better than the Sigma at F1.4, resolving more detail. At F1.8 it exhibits the same anomaly and it does so again at f2.8. Again, I have no clue as to why this happens, but in real world usage, no such problems occur. The 50mm's sweet spot is at F5.6. Though frankly, IMHO there isn't much difference between f2.0 and F5.6.
100% Corner Crop
One of the things I heard about the Sigma was that it gave trashy corners wide open. The people that said this weren't too wrong. The Sigma's corner sharpness at F1.4 is crummy at best(However, people almost never put things in the extreme corners, so this is really a non-issue). The 50/1.4 truly benefits from the cropped sensor and gives nice sharp performance in the corners even wide open. This would probably not be the case on a FF body. At F1.6 however, the Sigma's resolution suddenly jumps, even at the corner of the image. Not surprisingly, it still didn't match up to the full-frame Canon. The F1.8 phenomenon returned to plague both lenses and continued with the Sigma up to F2. At f2.8 the corners on the Sigma sharpened up nicely whereas the Canon's result was simply weirded out. In real world shooting, the Canon performs very nicely at f2.8.
The test probably had some fundamental flaws therein which I overlooked. I strongly suspect that the phenonmenon I noticed is due to inaccurate AF telling me the image was in focus when it wasn't. Whatever the case, I'm too lazy to repeat the test using MF. If it bums out in AF, then it'll bum out in real-world shooting as well.
All this crap is irrelevant in real world imaging, both lenses are great, that's why I hang on to both of them.
For really useful info about these two lenses, refer to my post in the link above.
