Daniel Browning
Well-known member
Hi Sanjin,
Thanks for the discussion. I'd like to point out a few areas where I disagree with you; hope you don't mind.
"Best" means something different to everyone. I agree that some still lenses have limited breathing, but many lenses are "best" for another reason, while still suffering from breathing.
For example, the Nikon 14-24mm f/2.8 is the best (for corner CA and MTF) super wide on Still35, but I don't know if it has limited breathing or not.
Agreed. Most good lenses are parfocal (such as most, but not all, of the Canon L zooms).
It many areas is certainly will; especially wide angle lenses. For one thing, none of the still lenses are fast (f/1.2) and wide (14mm) at the same time. You can get f/1.4 at 24mm, or f/2.8 at 14mm, but never the two shall meet. If you need f/1.2 for thin DOF or T/1.3 for low light shooting, cine lenses are the only choice.
There are other problem with wides, too. If you use a wide angle lens designed for Still35 on Super35 format, it will necessarily have more flare than one designed for Still35. The crop factor changes what you see in the viewfinder, but light striking the lens outside of that view will still affect the whole image.
However, if your needs are limited enough that all the focal lengths and focal ratios are already covered by still lenses, then such a comparison with a $30k MP would be meaningful. Pawel says he did the comparison and found the MP performance at f/1.2 to be better than still lenses, and I'm not surprised.
It doesn't matter what the performance is if no other equivalent lens even exists. Bicycles get better fuel economy than cars, but it's not a valid comparison because a car can do things a bicycle can't, just like the 14mm f/1.2 can do things a still lens can't.
Again, using still35 lenses on Super35 means there is a greater intrinsic propensity for flare simply due to the fact that there is over twice as much glass that is not even being used, yet still contributes to flare.
Yes, but there is another, more important rule of lens design: aperture rules. It rules performance, and it rules manufacturing cost.
For example, lenses designed for camera phones have 75% MTF at 400 lp/mm! And the cost is almost nothing. The principle reason for this to be possible is that the aperture is very, very small. The incredible amount of R&D and volume manufacturing (more in one year than all still lenses ever made) are also important.
Similarly, digicam lenses have very high MTF at resolutions a DSLR lens can only dream of. And, like camera phone lenses, that much resolution is needed, because the resolution of the sensor on an area basis is so high (over 10 times greater than DSLR cameras). The manufacturing volumes and narrow aperture makes it possible for such a low price.
Unlike digicams, still 35 lenses don't *need* to be very high resolution. Still35 formats are relatively low resolution (36 sq. microns vs 3.6). If a still lens is just barely sharp enough for a 4K Still35 format, then it will be much duller on a Super35 format. That's good, because they need much wider apertures. Plus, there is a respectable manufacturing volume.
Super35 is more exacting. The performance of a lens at 6K in Still35 will be the same as its performance at 4K in Super35. 6K is about 24 MP. Many lenses perform well at 24 MP, so they will also do well at 4K.
APS-C still lenses can be made to a higher resolution than full frame still lenses for the same cost, assuming the same design and volume; however, restricting them to mount with the same flange focal length causes design constraints/compromises. For example, my Tamron 17-50 f/2.8 runs circles around the Canon 17-40 f/4 L at half the cost, and holds up very well against the 24-70 f/2.8 L.
The point is this: if everything were equal, cine lenses would be cheaper, sharper, or wider (just like APS-C lenses). But because manufacturing volume and different design constraints, the reality is that many of them are more expensive for the same level of sharpness and focal ratio.
That is wrong. It is far easier (and cheaper!) to build a T/2.8 and optimize the wide open performance than to build a T/1.3 with the same level of performance. That can't be the main reason Zeiss made them T/1.3.
The Canon EF 50mm f/1.8 is $70. Stopped down to f/2.8, it has more resolution than the $400 50mm f/1.4 at f/2.8 or the $1400 50mm f/1.2 L at f/2.8. Starting with a wider focal ratio didn't help those lenses.
Thanks for the discussion. I'd like to point out a few areas where I disagree with you; hope you don't mind.
"Best" means something different to everyone. I agree that some still lenses have limited breathing, but many lenses are "best" for another reason, while still suffering from breathing.
For example, the Nikon 14-24mm f/2.8 is the best (for corner CA and MTF) super wide on Still35, but I don't know if it has limited breathing or not.
Agreed. Most good lenses are parfocal (such as most, but not all, of the Canon L zooms).
It many areas is certainly will; especially wide angle lenses. For one thing, none of the still lenses are fast (f/1.2) and wide (14mm) at the same time. You can get f/1.4 at 24mm, or f/2.8 at 14mm, but never the two shall meet. If you need f/1.2 for thin DOF or T/1.3 for low light shooting, cine lenses are the only choice.
There are other problem with wides, too. If you use a wide angle lens designed for Still35 on Super35 format, it will necessarily have more flare than one designed for Still35. The crop factor changes what you see in the viewfinder, but light striking the lens outside of that view will still affect the whole image.
However, if your needs are limited enough that all the focal lengths and focal ratios are already covered by still lenses, then such a comparison with a $30k MP would be meaningful. Pawel says he did the comparison and found the MP performance at f/1.2 to be better than still lenses, and I'm not surprised.
It doesn't matter what the performance is if no other equivalent lens even exists. Bicycles get better fuel economy than cars, but it's not a valid comparison because a car can do things a bicycle can't, just like the 14mm f/1.2 can do things a still lens can't.
Again, using still35 lenses on Super35 means there is a greater intrinsic propensity for flare simply due to the fact that there is over twice as much glass that is not even being used, yet still contributes to flare.
Pawel said:
Yes, but there is another, more important rule of lens design: aperture rules. It rules performance, and it rules manufacturing cost.
For example, lenses designed for camera phones have 75% MTF at 400 lp/mm! And the cost is almost nothing. The principle reason for this to be possible is that the aperture is very, very small. The incredible amount of R&D and volume manufacturing (more in one year than all still lenses ever made) are also important.
Similarly, digicam lenses have very high MTF at resolutions a DSLR lens can only dream of. And, like camera phone lenses, that much resolution is needed, because the resolution of the sensor on an area basis is so high (over 10 times greater than DSLR cameras). The manufacturing volumes and narrow aperture makes it possible for such a low price.
Unlike digicams, still 35 lenses don't *need* to be very high resolution. Still35 formats are relatively low resolution (36 sq. microns vs 3.6). If a still lens is just barely sharp enough for a 4K Still35 format, then it will be much duller on a Super35 format. That's good, because they need much wider apertures. Plus, there is a respectable manufacturing volume.
Super35 is more exacting. The performance of a lens at 6K in Still35 will be the same as its performance at 4K in Super35. 6K is about 24 MP. Many lenses perform well at 24 MP, so they will also do well at 4K.
APS-C still lenses can be made to a higher resolution than full frame still lenses for the same cost, assuming the same design and volume; however, restricting them to mount with the same flange focal length causes design constraints/compromises. For example, my Tamron 17-50 f/2.8 runs circles around the Canon 17-40 f/4 L at half the cost, and holds up very well against the 24-70 f/2.8 L.
The point is this: if everything were equal, cine lenses would be cheaper, sharper, or wider (just like APS-C lenses). But because manufacturing volume and different design constraints, the reality is that many of them are more expensive for the same level of sharpness and focal ratio.
That is wrong. It is far easier (and cheaper!) to build a T/2.8 and optimize the wide open performance than to build a T/1.3 with the same level of performance. That can't be the main reason Zeiss made them T/1.3.
The Canon EF 50mm f/1.8 is $70. Stopped down to f/2.8, it has more resolution than the $400 50mm f/1.4 at f/2.8 or the $1400 50mm f/1.2 L at f/2.8. Starting with a wider focal ratio didn't help those lenses.
