Ah, the first technical photo post. This one's on what a Transmission stop is, vs a Focal (length) stop, why you (shouldn't) care, and why it matters (but only a little bit).
We'll start with the caring part. The T-stop scale is derived from the F-stop scale, it still goes f/1.4, f/2, f/2.8, f/4, etc. However, F-stops are geometric ideals while T-stops are measurements of light passed through the lens. All lenses have a difference between their maximum F-stop and maximum T-stop. This means you get slight underexposures with all lenses, but they are usually less than 1/3EV and not significant, unless you're a cinematographer. In a motion picture, slight changes in exposure would be noticeable between shots, which is why T-stops are used, not F-stops, on cine lenses, f-stops are not perfect, thus the actual exposure of f/5.6 across a variety of lenses will vary just a little bit. T5.6 is T5.6, all lenses will be equally bright at T5.6, but not at F5.6.
Right, on to the science. Lens elements absorb light, as everything does. Uncoated elements will cost between 4% and 6% of the total light, each. Coatings improve this to anywhere from 99.4% to
99.9% That doesn't sound like so much light, but remember that's per element. An f/1.8 50mm, such as the one in the cover photo of this post has around 7 elements in it. Uncoated, it would only transmit about 65% of the light a T1.8 lens would, placing it around T2.3-T2.4 for F1.8. This would produce underexposures of about 2/3EV when shooting at f/1.8. If we change that number to 99.7% transmission per lens surface, we get 98% transmission, or about T1.8 with only a slight underexposure. A 23-element lens, such as a 24-70/2.8 or 70-200/2.8, with coatings making it have 99.6% efficiency, would transmit 93% of the light for T2.8, making it a T2.9 or so. At 94% efficiency, it would only transmit 24% of the light it should; making it a T5.6 lens, two stops underexposure at f/2.8.
Some super-bright lenses, such as the canon 50mm f1.2L, isn't near T1.2 at all. The lens' efficiency is probably about T1.4-T1.5, but something called pixel vignetting comes into play, where the aperture is so wide the microlens array in front of the sensor actually 'misses' some of the light from the lens, the camera is programmed to raise gain on the sensor when shooting faster than 1.4, which results in higher than expected noise. However, if you remove this gain change, counting pixel shading you get T1.7, a full stop of light.
We'll start with the caring part. The T-stop scale is derived from the F-stop scale, it still goes f/1.4, f/2, f/2.8, f/4, etc. However, F-stops are geometric ideals while T-stops are measurements of light passed through the lens. All lenses have a difference between their maximum F-stop and maximum T-stop. This means you get slight underexposures with all lenses, but they are usually less than 1/3EV and not significant, unless you're a cinematographer. In a motion picture, slight changes in exposure would be noticeable between shots, which is why T-stops are used, not F-stops, on cine lenses, f-stops are not perfect, thus the actual exposure of f/5.6 across a variety of lenses will vary just a little bit. T5.6 is T5.6, all lenses will be equally bright at T5.6, but not at F5.6.
Right, on to the science. Lens elements absorb light, as everything does. Uncoated elements will cost between 4% and 6% of the total light, each. Coatings improve this to anywhere from 99.4% to
99.9% That doesn't sound like so much light, but remember that's per element. An f/1.8 50mm, such as the one in the cover photo of this post has around 7 elements in it. Uncoated, it would only transmit about 65% of the light a T1.8 lens would, placing it around T2.3-T2.4 for F1.8. This would produce underexposures of about 2/3EV when shooting at f/1.8. If we change that number to 99.7% transmission per lens surface, we get 98% transmission, or about T1.8 with only a slight underexposure. A 23-element lens, such as a 24-70/2.8 or 70-200/2.8, with coatings making it have 99.6% efficiency, would transmit 93% of the light for T2.8, making it a T2.9 or so. At 94% efficiency, it would only transmit 24% of the light it should; making it a T5.6 lens, two stops underexposure at f/2.8.
Some super-bright lenses, such as the canon 50mm f1.2L, isn't near T1.2 at all. The lens' efficiency is probably about T1.4-T1.5, but something called pixel vignetting comes into play, where the aperture is so wide the microlens array in front of the sensor actually 'misses' some of the light from the lens, the camera is programmed to raise gain on the sensor when shooting faster than 1.4, which results in higher than expected noise. However, if you remove this gain change, counting pixel shading you get T1.7, a full stop of light.
Cine lenses aren't necessarily sharper than their photo counterparts, but they're brighter (for instance, the Zeiss 70-200 is T2.9, Canon's is T3.4), tend to have less CA, and have way longer focus throws. Those factors make cine lenses more expensive. You can use photo lenses for video, but at the film/cinema level it's all cine lenses for those reasons. Photographic lenses are not inferior, they're just made to a different set of standards.
No comments:
Post a Comment