Full Version: A couple of suggestions
You're currently viewing a stripped down version of our content. View the full version with proper formatting.


Before I suggest you anything, I'd like to thank you for all your work, there's no site I trust more than yours when it comes to lens reviews, ever since I got my first DSLR (a Sony Alpha 200) you've helped me decide what is worth my hard earned money and what isn't. 


There're only two things that I think you should add to further improve the utility of your reviews:


1. How effective optical stabilization is

2. Light transmission (t-stop)


I think the second point is really important, sometimes the difference is a full stop. Also, it seems relatively easy to test.


Many probably buy wider aperture lenses with the explicit intention of reducing noise or shooting in darker environments.


For example, many probably buy the Olympus 25mm f/1.2 over the Leica 25mm f/1.4 precisely because they want a faster lens, however, they actually transmit the same amount of light (yet the former costs twice as much), and both are only 1/3 of a stop faster than the Panasonic 25mm f/1.7. 


So people might buy it expecting to reduce their ISO from 6,400 to 3,200, only to find out it merely went from 6,400 to around 5,300. 


Of course, they might offer other things that justify their price for any given customer (better build quality, bokeh, sharpness, among many other things), but spending 3 times as much in a lens for the wrong reason is something must of us would rather avoid. 

Measuring T-stops is actually not that simple. How do you propose to go about that? 

Camera sensors contribute to the light loss an image would try to reveal, microlens design and pixel size may contribute to that and vary from sensor to sensor.

Cameras also "cheat" by applying extra gain to the signal with larger apertures, due to that sensor level light loss. Do they apply the same gain with lenses from other manufacturers, which to the camera may identify wrongly?


Lenses may also show varying light transmitting through the focal range, or not.


Buy larger aperture lenses primarily for more shallow DOF possibilities, and worry less about 1/2 of a T-stop more or less.


As to how effective IS is... That is pretty hard to establish in an objective, consistent way. In which use cases? And it should be done by a machine test rig which assures the same conditions for every lens, review to review, through the years. 

Before I reply to your suggestions, first of all: welcome to the forum, and thanks for your kind words Smile

We tried several attempts to measure the effectiveness of optical stabilization in the past, but none of our ideas really worked or gave reliable and repeatable results. The amount of effort we would have to put into creating a test setup that actually delivers results would be huge, and still the result would be questionable, since the effectiveness of stabilization (optical or in-body) always also depends on the shooter and his/her shaking movements.
That's why instead we usually mention the claimed gain on the first page, as well as our impression from shooting with the lens.

Concerning t-stops: just like BC above, I don't see an easy way to measure it. Finding a reliable constant light source would probably already be the first difficult step to master.
And in addition: I question the motivation to get such a fast lens, at least the way you described it. Because besides the possibly lower than expected gain of usability in low light, a buyer of such a fast lens might also come to the (maybe unexpected) conclusion, that at f/1.2 or f/1.4 such a thing as depth of field exists even on m43 Wink And personally I think excessive noise is easier to fix in post-processing than lack of focus, if needed.
Noise is given - more or less, but always visible in dark sitiations, Fooling around with 1/3 or less stops is only theoretically relevant. In reality, there's just much more and much better reasons to buy into fast or even expensive glass (if one wants to afford it), such as
  • making the AF's work easier (although at the same time fast glass = big and heavy glass = slower AF movement)
  • getting a nicer bokeh, abetter 3D-pop
  • getting a probbaly (but not always) better build quality because these lenses are expected to be more costly and they have to deliver more precise and reliable movements (which doesn't mean, all fast lenses do so).
T-stops I saw mostly on cine-lenses. I can't remember to have seen a cine lens tested from PZ. Lensrentals might be a better address for them. They just recnetl tested some on their optical bench.

On making the AF easier... For traditional PD AF, no. If you have a f5.6 sensitive AF point... it makes no difference if you put on a f5.6 lens of a f1.2 lens (the AF module has its own aperture limit). Same goes for f4 or f2.8 sensitive AF points.

On live view AF: Sony, for instance, has the odd habit of closing the lens down to a certain aperture during AF/live view. Not sure why.


And Klaus has tested some Cine versions, like some Samyangs Wink


Thanks for the answers, I thought measuring t-stops would've been simpler (although I know that sensors matter and that cameras somewhat compensate for this, you usually perform your tests on the same body for a long time, so I didn't think it'd matter as much, but I trust you that is in fact hard). 


By the way, I'm not saying I'd buy an f/1.2 lens for that reason (I don't even own a M43 camera, I shoot on a fuji), what I mean is I think many people (perhaps less experienced photographers) think such a lens would let them use a faster shutter speed or lower ISOs, which is just misleading.


I realize things are complicated enough as is, explaining equivalence (especially aperture equivalence) is problematic enough, I just think consumers in general would benefit from knowing this. For example, say I'm trying to decide between the Fuji 56mm f/1.2 and the APD version, the APD version is T/1.7, so that might lead buyers to believe it loses essentially a full stop, but I'd be absolutely stunned if the non-APD version is actually T/1.2. 


Of course, I think manufacturers should report both. 


Again, thanks for all your work and explaining why my two suggestion are not actually easy to implement Smile