The conversation gets a bit absurd outside of the APSc context. The Tokina/Pentax design is limited to 90 degrees horizontal FoV on crop sensor. The Sigma was designed for FF yielding an expansive 122 degree FoV - so the marginal difference in resolution from an exotic design is a given. Yet, despite the greater demands on the lens design, the Sigma is far superior in CA and lack of distortion.
The review gives the Sigma a 9.0 (excellent) rating, Pentax an 8.8 (very good) rating - playing on the latter's turf. The retail price at time of review was $829.95; that price is now roughly halved... So, the review conclusion:
If you have an unlimited lens budget, the Sigma DG 12-24 f/4.5-5.6 is the perfect choice. The lens has the least barrel distortion, vignetting, chromatic aberration, and is reasonably sharp. Why wouldn’t you buy a lens that already addresses some of the major problems of ultra-wide angle zooms?
For me, I can tolerate only so much CA, and software correction always leaves defects when the problem is so severe as in this Tokina lens design. The review sums it up correctly:
The Sigma DG 12-24 f/4.5-5.6 was the clear winner. Not only did it have very limited chromatic aberration wide open but it was consistent throughout the f-stop sample. The Tamron SP AF 10-24mm f/3.5-4.5 DI II displayed some chromatic aberration at f3.5 but improved significantly at f8 and f16. The Pentax DA 12-24mm f/4 did not do as well as the other two lenses, which was surprising. Not only did it have significant chromatic aberration at f4 but it did not improve much as the lens was stopped down.
My experience with the Tokina version of this lens (on Canon) was not so severe, but still is considerably worse than the Sigma (second sample). These UWA lenses exhibit extreme CA with only very minor decentering. This is why you get such varied test results in credible testing (especially apparent with the Sigma). It is quite certain that if you are finding greater CA on the Sigma, you have a slightly decentered copy, unfortunately.
---------- Post added 12-24-2017 at 11:34 AM ----------
Originally posted by clackers Good discussion, and some nice images in that thread. Filters are helpful for the kinds of situations you posted.
Believe me, I use filters when I have the need. Way back when when I was working in a camera shop, we got great deals on fully coated filters (typically Hoya back in the days when their filters were much higher quality) because the retail mark-up was huge, and employees got them for cost+10% - $80 filters for $9, etc.
The problem with even thin filters on UWA is angle of acceptance. That's why you see the bulbous outer elements - they can manage the angle better - closer to perpendicular. Obviously, front filters are flat, causing optical degradation in the UWA field due to the severe angles involved. At least gelatin filters mounted at the rear are very thin and are less of an optical challenge. Frankly, I've never bothered with them, though, so I can't say if they are an effective solution.
A second problem is how polarizers work in sunny conditions at UWA. As is quite obvious, the effect of polarization varies considerably depending on angle to the light source (the sun). This results in increasingly uneven impact the wider the FoV. On UWA, it is almost always unnaturally distracting, and very difficult to fix in post. It can be interesting left alone as a special effect, but not my cup of tea.