On Thu, 6 Nov 2003, Rob Studdert wrote: > On 5 Nov 2003 at 9:13, [EMAIL PROTECTED] wrote: > > > Unfortunately most of the primes on this list are older optical designs, > > primarily because pentax hasn't made a lot of new ultrawides I assume. > > The 15/3.5 design apparently isn't great (nor is the equivalent Nikkor), > > the 20/4.5 is generally held to be not the best. > > Virtually all of the Pentax wide primes performed as well as if not better than > their competition of the day from other manufacturers.
That is probably true. They are still not great compared to 50mm lenses. Nikon's aren't either, nor are Canon's likely to be. It's just not as easy to make a good 20mm as it is to make a good 50mm, and certainly not easy to to it small and inexpensive (nikon's first 20mm took 72mm filters!). Even the most modern, expensive ultrawides do not perform as well in terms of distortion and corner sharpness as lenses in the 35-85mm range. I'd love to find a 20mm screwmount lens that is faster than the 20/4.5 Takumar and performs at least as well, but I don't really expect that anyone made a better one back then. The Fujinon EBC and Zeiss Jena Flektogon are faster, but also damn near impossible to find. This was my point, in a way. I suspect that a critical examination might find that these ultrawides, indeed most ultrawides, don't perform well on FILM either. I've always wondered when I read one of those reviews that says "great at all apertures" or "great lens" whether the reviewer means that the performance is great in an absolute sense or simply relative to other similar lenses. > > I'm curious what is WRONG with the images delivered by these lenses on the > > *istD, especially compared to the images delivered by the same lenses on film? > > Sharpness and contrast? (The *istD is known to "undersharpen" its images > > electronically, and the sensor in it does appear to be less sharp by nature than > > the Canon and possibly the Fuji sensors). Distortion? > > Mix the existing subtle chromic aberrations with Bayer colour sensor matrix and > lenslets and you end up with exaggeration of the error, as we've seen in many > of the sample images presented thus far. Why does it affect the ultrawides more? It certainly seems to, since the Nikkor 20-35 and Canon 17-35 have lousy reputations with respect to chromatic aberration and such. The Nikkor 17-35 is apparently much better, presumably because Nikon saw what happened when the chromatic aberration hit the sensor of the D1. Presumably the newer Canon 16-35 is better too. The other question is whether the Pentax ultrawides would produce better results on the Fuji S2, or the Nikkor lenses on the *istD. Unfortunately, it is very hard to test this. How 'bout Sigma or somebody like that? They make the same lens in both mounts. Tamron makes a couple of ultrawides that you could presumably try the same lens on BOTH cameras with the right adaptall mount. Even if the lenses are mediocre it would show if the camera is to blame. It does suggest that unless the Foveon technology succeeds in the market and the Bayer grid disappears the lens-makers are going to have to try to correct better for chromatic aberration in future lens designs. > > I'd actually > > expect Pentax lenses to perform better on a digital camera than most > > because Pentax seems to optimize for center sharpness at the cost of > > corner sharpness and digital of course doesn't use the corners. > > Well a diagonal of AOV 28.4mm across the sensor for the *ist D can hardly be > deemed as "doesn't use the corners" (of the lenses projected image). The sensor > covers almost 66% of the diagonal compared to a standard 35mm film frame. That 1/6th in each corner however is precisely where the ultrawides tend to struggle, especially the aforementioned 20/4.5. The extreme edges and corners of my 18/3.5 Nikkor show some really nasty distortion and lack of 3-dimensionality that don't show in DSLR images. The corner performance of lenses is often tested presumably as a sort of "worst case scenario" for the lens. Depending on how you shoot there may not be anything important in those corners anyway. DJE
