Update #4 (10/3/2007): Today I tested in the store the latest HSM version of the lens, fully intending to buy it for my D40 camera. Autofocus was very quiet (more quiet than the non-HSM version). Not very fast though - comparable to 18-200VR lens (which is not slow but not the fastest to focus either). This is not "true" HSM in the sense that, while it is quiet, the focus cannot be overriden without first engaging the manual focus mode on the camera body. The focusing *ring* also rotates during focusing (but the front threads do not) - just like it does on the non-HSM version.

Unfortunately, that particular sample was terrible wide open at 50mm: lots of glow that was almost one by F4 and was gone completely by F5.6. Compared to my 18-200 VR at 50mm (also wide open) this Sigma was *bad*. My first sample (non-HSM had none of this softness and glow!).

See this image:
"http://www.pbase.com/kocho/image/86648391/original"

Update #3: I had a second copy of the Sigma lens. I posted some new photos from it. This copy has uneven aperture blades that do not form a symmetric 7-sided shape but a skewed one. Not sure if it matters or not. Comparing this sample with the previous at the same scene (Barnes & Noble) reveals that this second sample is very sharp at 18mm (a little sharper than the previous one) but noticeably shofter at 50mm at all apertures. It also has a very slight extreme edge left-sided softness (which is not there on my Tokina 12-24 - both sides are similar). At 50mm it has a misalignment as well - left side is very sharp, center is bad (soft + dreamy + CA) and right side is not very good but better than the center. At 18mm at long distances the Sigma is marginally sharper than the Tokina at all comparable apertures, except for that extreme left edge. I have also uploaded some close-up photos with it that demonstrate the macro function.

18mm is again confirmed a tiny bit less wide than 18mm on the Tokina and 50mm seems a little shorter than 50mm on my Nikon zooms. So, a 18.5mm-45mm lens (guessing)...

This second sample is going back due to the above issues. I think if you get a good sample, then it should be a great performer!

See also the comments on the bottom of the page where I shared my impressions of the 28-75 Tamron and 35-70 Nikon.

Update #2: Jim Nguyen was kind enough to provide the Canon XTi images taken with the Sigma 18-50 Macro lens. ACR default settings, USM 300/0.2/5.

Update #1: I've returned both and am currently using Tamron 28-75 as it suits more shooting situations for me. It is a tough choice b/w the two - they are both very good and offer IMO a better alternative of the kit lenses in the same price range. See below for more detail. I suppose the focusing issue can be fixed on either the lens thru warranty, but I did not like any of them enough to go thru that. I liked the wider end and handling of the Tamron better but did not like its curved focused field - it is impossible to get sharp corners at wide apertures if the target is flat. It is excellent (as is the Sigma) when stopped down and both are very good wide open. I preferred the IQ of the Sigma overall due to less field curvature and less CA. But I'd rather have something a little wider than 18mm. With the Tokina 16-50 coming soon, I will try that before making-up my mind. Also would like to try the Nikon 17-55 but the cost and weight/size are too much for me, so I might settle on one of the three "alternatives". I would take either the Tamron or the Sigma over my kit lens (18-70, 18-55 or 18-135), provided focus can be fixed, but have to try the Tokina before making up my mind. Thanks for whomever looked and provided input thru my postings at DPReview or Nikonians forums.

======== More detail ======

17-50 Tamron vs. 18-50 Sigma Macro (the new design for 2006). Here is my quick summary of my findings so far:

Build: Both are more somewhat better built than comparably priced kit lenses (18-70 or 18-135 or 18-200 VR). Nothing to complain about here.

Usability: The Sigma feels a little more "rough and tight" in use and the fact that it zooms in the opposite direction of Nikon and Tamron. Focusing and zooming on the Tamron feel a little nicer to me. Tamron focuses a little faster than the Sigma, which is to be expected as the Sigma is designated a "macro" (not a huge difference). Both focus reasonably quietly though not as quiet as AF-S lenses.

Focus: Both show relative front-focus compared to my other Nikon lenses (18-55, 18-135 - three copies, 18-200, 50mm AF/D F/1.4) on D80 (which is calibrated to focus correctly with the Nikon lenses). The Tamron seems to miss focus occasionally, where I did not observe this with the Sigma that much. The Tamron front-focuses more than the Sigma and both are significantly front-focusing relative to my kit lenses or the 50mm prime.

Sharpness: Generally the Tamron seems to be a little sharper towards the edges and the Sigma is a little sharper in the center. Both are sharp enough across the frame for most purposes, except perhaps the extreme corners on the Tamron on flat targets (see below). Both lenses get a little softer towards the edges but maintain sharpness there better than the 18-135.

Macro: Both lenses focus fairly close. I forgot the specific ratios for the Tamron, the Sigma is 1:3. Far from true macro but very usable if you can get to 2 inches or so from the subject (hint - need external flash). Center sharpness on the Sigma in macro mode is very good with no CA. Off-center there is lateral CA that remains visible. Out of focus in Macro mode is rendered very nicely, with blacks remaining black (unlike with some other lenses where they turn red/blue/purple/greenish). Have not tested the Tamron for close-up much so I have no impression there.

CA: The Sigma is a little better than the Tamron. Both are OK, compared to lenses like 18-135 which are pretty bad in this. Not a big issue IMO as it is not very obtrusive even at 100% without correction. The CA from the Tamron is visible in everyday shots when enlarged enough or on difficult contrasty scenes (I would not use if for weddings without correction as there is a lot of black/white contrast and the blue fringe shows there. The Sigma has less of it and it needs no correction in almost all cases.

Field curvature: The Tamron has strong focus field curvature much like the 18-135 or 18-200 VR lenses do. The Sigma is less curved. This plays a role in corner sharpness for the Tamron: while its corner sharpness is very good, that unfortunately usually happens on a plane that is farther back than the focus plane in the center, so it is hard to get both sharp center and sharp corners until you stop down to F8 or F11. This is not a big issue as the middle of the borders are good - it only affects the corners (unlike the 18-135 which seems to get softer towards the edges faster).

Exposure: (update 12/10/06) I did not observe any problems non-flash shots. Many have reported flash overexposure in TTL-BL modes with the Tamron and I found this to be true in my case as well - it would render the same scene with 2/3 to 1 stop brighter flash exposure compared to 18-135 at the same settings and composition. Not a big difference and nowhere near as bad as it could be - I had a Tamron 90mm macro with the flash overexposure issue, so I know when I see one - check my 90mm macro gallery for this. But do check your sample for this in TTL-BL and other flash modes.

Color: I can't tell much of a difference. Look at the outdoors photos from Barnes & Noble, which are processed identically for each pair of shots. Since I have not compared to other lenses for color I can't tell if they are neutral warm or cold. I see no issues there however.

Bokeh: Both seem to have similar out of focus performance. Have not looked too much into this yet. They are not extremely creamy, but are smoother than the 18-135 which may get the "ring" type specular highlights more easily.

Price: The same within $10 dollars for me so not a differentiating factor. Both can be found for about $420 delivered if you shop around (got mine from B&H).

Focal length: The Tamron is shorter than the Sigma at 50mm and both are shorter than the 18-135 at 50mm. So, not true 50mm designs. The 17mm vs. 18mm on the wide end makes a difference - not a huge difference but it is there and is arguably more important than the slight difference on the long end.

Conclusion: (see update above and the comments below).

UPDATE - I have removed the full-sized samples from this gallery. Reviews of these lenses are available now from the "pro" web sites that actually get paid for their existence and bandwidth usage -;).