douglaurent said:
Having witnessed how Zeiss test their lenses including on F Stop / T Stop machine I would say in that typical Germanic way they strive to be accurate, very accurate.
Yes, Zeiss lenses are great performers and have little sample variation, but they also have their downsides that people who don't own them don't see, or people who own them ignore (Disclaimer: I own the whole top lineup myself). Most Zeiss lenses for example are not to useful for landscape photography, as their focus infinity already starts at around 50 meters (like their 15/.28), while with Canon lenses you can easily distinguish and adjust the exact focus point among 50, 250, 500 or 1000 meters. Autofocus users and users who stop down or don't have a large monitor don't realize this Zeiss limitation, but it's a fact and i have it even confirmed in written form by Zeiss.
In general, all these DXO Mark values might also be correct based on their test system, but real world shows different results like regarding transmission/brightness. Example: DXO Mark lists the Red Dragon sensor as best performer. I own 2 Red Epic Dragon and know that the 5D3 will definitely deliver better and less noisy results after raw processing, which is no wonder as a Canon raw file is much larger than a Red raw frame. The cool thing about the Red is to have 100 frames a second in 6K, but the Canon image potential is way ahead, which makes the DXO Mark numbers pretty useless again.
Canon also has many fields where they outperfom competitors. Stabilized Canon lenses for example are a pleasure on MFT cameras like the GH4 when it comes to stabilization, which is much better than the Lumix' own system. I think if Canon wants to, they could be ahead of any competitor, and the forthcoming Canon expo will show what they are capable of (but unfortunately spread over 5-10 years).
Knowing this, the lack of bringing the quality into the products is what can make you angry as a customer. In the 5DsR they have finally built in an internal timelapse timer feature. But is it useful like in the competitors menus, where you can set a start time and a 4-digit number of images, so you don't have to wake up at night twice to start and stop your timelapse? No, Canon has implemented a 2(!) digit number only of photos your timelapse can do automatically, without a dedicated start time - which means you still need external equipment to do it right.
Same with the world premiere of the video autofocus function in the 5DsR, which lets even all non-STM lenses focus in video - other manufacturers would have cherished it, but Canon wants to avoid that anybody who is doing video buys a 5DsR and instead buys a second and more expensive dedicated video product. All this makes you think that Canon is not really interested in you as a customer at all, at least not like the feeling you can have with Sony or Panasonic at the moment.
[/quote]
Your missing my point so let me if I may explain it differently. We have a T Stop machine, a Euro 350K MTF, which we can use on and off axis as well as multi-point, a specially designed projector, housed in a specially designed room. Every lens we purchase we test at each T stop and at each distance on the scale. With the projector were looking for chromatic aberrations of different types as well as field curvature issues and even field illumination. Zeiss do not engrave each distance they do infinity and six feet & extrapolate the other distances but to get to your point if we were repeatedly seeing Zeiss lenses with inaccurate T Stops we would reject the lenses with a set of Master Primes coming in well over Euro 250K and shows with multi-million budget shooting with them we would not except second class and neither would the DOPs testing them, this applies to all their cinematography lenses.
Historically the Canon K35 lenses were actually faster than their engraved T stops and varied far more than Zeiss, and Canon still have a larger variation albeit were talking small variations. We test all cameras with the same carefully selected 50 mm lens using a device to test the dynamic range of all digital movie cameras that works fine also testing DSLRs and were able to see the variation in individual camera sensitivity within the same type.
All this is important because you can get a wider error depending which side of the optimum setting the lens or camera is. I doubt most people are able to test in the same conditions week in week out to give repeatable results I know Zeiss can as we can.
