It seems to me the Photozone tests are somewhat meaningless with regard to actual performance at infinity since they are not done at infinity. Behavior at close distances is often vastly different than at long distances, depending on the lens. All we can go on with regard to objective measurements are the Zeiss MTF's that I posted earlier and which indicate the Z* 35/2 will outperform the CY 35-70 at similar F-stops. It would be nice to see a back to back test of these two lenses at longer distances to settle it. I bet they would be closer in performance than anyone might expect.
See the distant hills, I estimate they are maybe 4-5 kms away, at a guess. That is a typical far distance for me, important for reasonable detail to be retained and well-presented, even in poor light conditions.
The 35/2 looks like it has a dirty and colour-affecting UV filter on it. Contrast is way down, the whites look sterile, the palette looks weak and 'digital', like we saw from early DLSRs. It might be more 'accurate', but this is about photographic images. I bet the histos are quite different too.
For most landscapes, colour and contrast equal depth; and the subjective quality I call 'authenticity'. They also serve (somewhat) to cover a loss of detail from focus error, focus fade etc. f8 at 35mm on good lenses brings this issue into play very often, and colour tonality/contrast helps disguise it.
The zoom peaks at f8, so f11 is likely more viable than for a faster lens that peaks at wider apertures - the maybe questionnable photozone test shows the 35/2 at f11 to have fallen apart, worse overall than f2, but at least it's now consistently bad across the frame - a pyrrhic victory of sorts.
Given the caveats, there is nothing much between them in sheer resolution (might be on a 24Mp), but the images look a world apart. And, as with many landscape images, corners here are totally a non-issue - the main reason this happens so often, I think, is the distance from the focal plane that corner detail typically lies at, so it's often a defocus issue mixed with lack of contrast (res).
I'd much rather work with the CY image in post, as it's most of the way there already, which is my general experience of the lens. It lets you see the colour and the light falling on it, both. When they pop out looking as crook as the 35/2 does here, you can be tempted to overlook the photo, knowing how much work lies ahead. All subjective of course, but those are some random observations.
Zeiss report a min FL of 35.7mm for the zoom with an AOV of 63 degrees; and 62 degrees of AOV for the prime. So they shortchange you there a bit also.
Thanks for posting the comparison MP. It would be nice to see the 100% crops of the distant hills at the right of the frame that Philip mentions. It is always amazing to me to see exactly how much influence subtle color balance differences can have with the look of an image. When I open both the reduced images and the crops up in PS, and do a quick "match color" of the ZE image to the CY image, they look almost identical (and the "digital" look Philip mentions magically goes away in my opinion (partially due to the warmer colored whites and neutrals). There appears to be the slightest amount of over exposure in the whites of the ZE as compared to the CY which does result in some slight differences mostly visible in the crops (though, of course, the white walls in both are blown out quite a bit and underexposure would have probably helped out). I would love to see how well the CY at 35mm holds up with a higher resolution body such as the A900. It looks great at 12MP.
Per requests, here is a right-side crop. The mountains in the background are 6+ km away, and there's a bit too much haze for fine-detail visibility at that distance. I have no filters on my lenses, just a dusty sensor .
This part of the scene has fewer high-contrast features to highlight the difference between the lenses, though there is a bit more muddling of the branches detail on the left due to CA in the ZE35.
If the participants in this discussion do not know, or distinguish, the difference between lateral CA and longitudinal CA, the discussion is amateur hour.
Keith B. wrote:
If the participants in this discussion do not know, or distinguish, the difference between lateral CA and longitudinal CA, the discussion is amateur hour.
The ZE 35 has both, but in near-infinite-depth-of-field stopped down situations like this, lateral CA is the dominant concern. For wide angle distant landscape use, I have rarely (never?) run across a case where longitudinal CA is problematic (or even noticeable).
Thanks for posting the crops. If the color balance of the ZE crop, which is slightly colder/ bluer, is matched to the CY, here is what we get. They are close. I do see the CY advantage which could be due to more contrast in the lens and/or a very slight exposure difference (different lenses even at the same F-stop will show slight differences).
mpmendenhall wrote:
The ZE 35 has both, but in near-infinite-depth-of-field stopped down situations like this, lateral CA is the dominant concern. For wide angle distant landscape use, I have rarely (never?) run across a case where longitudinal CA is problematic (or even noticeable).
I'm guessing that he means the crops of the purple house further up this page. The tree branches are in the front of the DOF and show uniform blue longitudinal CA, not lateral CA. In the rear DOF they'd show yellow LoCA. If it was lateral CA the blue fringes would be accompanied by yellow fringes in the same plane of focus.
alundeb wrote:
Agree, but not only for infinity, also for the foreground detail.
The odd thing is that I get incredible detail with the ZF.2 35/2 using the 24MP Sony a900 at both infinity and mid range distances (and what we are seeing here is with the 12MP 5D). This would lead me to believe something else is at work here - lens sample variation, exposure, etc. but the CY certainly looks better in this comparison.
AhamB wrote:
I'm guessing that he means the crops of the purple house further up this page. The tree branches are in the front of the DOF and show uniform blue longitudinal CA, not lateral CA. In the rear DOF they'd show yellow LoCA. If it was lateral CA the blue fringes would be accompanied by yellow fringes in the same plane of focus.
If you mean this, then I would agree, except that I see purple fringing in tree branches, not blue fringing. I bet its not LoCA, seen similar stuff accros whole frame even when it was in focus on high contrast areas.
Tariq Gibran wrote:
The odd thing is that I get incredible detail with the ZF.2 35/2 using the 24MP Sony a900 at both infinity and mid range distances (and what we are seeing here is with the 12MP 5D). This would lead me to believe something else is at work here - lens sample variation, exposure, etc. but the CY certainly looks better in this comparison.
Lens sample variation could certainly be a factor. I bought this ZE35/2 from lensrentals.com, who claimed this was an anomalously sharp copy of the lens (and they really do have the data to back up claims like that). However, I suspect their routine testing is primarily at mid-range distances, and the wider apertures of the lens (and the lens does indeed shine for wider aperture, MFD-to-mid-range work, which is the application I foresee using it for). Perhaps whatever fortuitous combination of factors made this copy stand out in those conditions was a tradeoff for some infinity/f8 quality, and other copies might do better there. Nonetheless, I suspect there are enough other copies with similar infinity focus qualities to mine that I am not the only one likely to see "lack of magic at infinity" from the lens, especially compared to cult classics for this like the CY35-70.
I would guess I was the one who is supposed to be ambiguous about which type of CA, but I know the difference. To be fair, the picture of the purple house is just a snapshot out an open car window, so it's relevance to a shot with perfect exposure and focus using a tripod is questionable. HOWEVER, exif show f4 and magnification shows both purple and green-blue fringing of single areas. There is also longitudinal ca, which is a separate issue. The ability to see typical different fringing on different sides of high contrast areas on lateral areas is complicated by low pixel count of sensor and rectangular individual pixels (unlike modern sensors).
I wondered who bought that copy of the lens from LensRentals...