stargazer78 wrote:
EXIF data on those samples indicate that all the Sigma photos were taken with 1/3 stop longer shutter speeds than their Canon counterparts. Inspite of this, the Sigma samples still look like they're about 1/6 stop darker than the 85L samples. By my guesstimate, the Sigma is exposing about 1/3 to 1/2 stop darker than you'd expect.
This is consistent with what has been discussed before: the lens isn't quite as bright as it should be. So not everything is sunshine and roses for the Sigma 85mm f/1.4.
Bear in mind that F-stop and T-stop are two different concepts. The T-stop indicates how much light is actually transmitted through the glass. Large quantities of glass and non-optimized coatings can absorb light, reducing the overall amount of light that hits the sensor.
The T-stop rating is typically expressed as an f-stop of a lens if 100% of the light were transmitted. For example, a lens may have a wide open f stop of f/2, but a T-stop of T/2.2, meaning that the glass in the lens absorbs about 1/3 of the light passing through the lens.
That being said, I'm sure many of the wide open apertures are fudged, but considering the precision of the aperture mechanism, I'm sure both manufacturers are able to get fairly precise F-stops. So, when you set your camera to f/2, I'm sure both lenses have apertures that are f/2 or very close.
If you look at the specifications of the lenses, the 85 f/1.2 has a design with 8 elements in 7 groups, whereas the 85 f/1.4 has a design with 11 elements in 8 groups. I'm fairly certain that the greater quantity of elements reduces the T-stop of the 85 f/1.4, resulting in the consistently slower shutter speed.
PhilDWedding wrote:
My issue with Sigma has always been focus speed & accuracy (accuracy being more of an issue if I'm honest). If I could get hold of one to try it, I might be interested, but without really really trying it out in anger at a wedding I wouldn't be comfortable...
Alundeb:
Yes it does; but pupil magnification is only variable within a certain ratio in a certain base type of construction. I'm certainly not a "reference" in this area, but i would think that the maximum "span" of available pupil magnification ratios could be maybe about +/-20%. As far as my understanding goes, DoF increases/decreases with the sqare root of the pupil magnification ratio change.
Both macro shooters and landscapers often use this effect to their advantage if they can. Portrait shooters tend to want it the other way around; a DoF as short as possible with a given F-stop.
Most wideangle retrofocal lenses (most lenses <40mm on a DSLR...) give quite a lot of this effect. They give you a lot more DoF than the calculations would imply. DoF-calculators often suppose a pupil magnification ratio of 1:1 (symmetrical) to simplify the calculations. Most zooms also have this effect, their shorter zoom settings give a lot more DoF than the calculators say when you only consider the F-stop (front pupil).
A quick estimation of the P ratio can be seen if you hold a lens at arms' length and first look through it through the front, then turn it over and look through it from the back. A "normal" ratio" seems to be around 1.5.
The 85L will then show a larger rear pupil when at close focus than at infinity focus (since it's a unit focusing lens with a positive fixed rear lens) - the Sigma is rear focusing, and this probably shortens the apparent focal length the closer you focus. The L is the other way around - the apparent FL gets longer as you focus closer - just as with an old unit focusing macro lens. IF you could use enough tube with the L to make it go to 1:1, then it would be an effective 170F/2.5.
theSuede wrote:
Not necessarily... It explains why the Sigma has a lot less falloff vignetting and no edge clipping of the bokeh circles - it has a different pupil magnification, its' exit pupil is further from the image plane. This is not necessarily linked to the size of the front pupil; which decides the F/no and light gathering efficiency. You can do this quite easily by building a positive Galilaean system after/into the main lens' rear geometry. You wait until the beam-packet has tapered off to a smaller point, and then spread them back out again towards the sensor. If your base is a planar/Xenotar, then you need at least 9 lens elements to keep symmetry though. A very small cost compared to the large, and expensive front elements.
I actually think Canon knowingly sacrificed some vignette/bokeh symmetry/CA correction to gain what is the trademark of the 85/1.2L - the very steep falloff from sharpness to 'blurredness'. You get this by increasing pupil magnification ratio, but you loose out on the points I mentioned earlier. ...Show more → alundeb wrote:
It is great to learn something new here!
Question:
If the pupil magnification affects the diameter of the OOF highlight disc like it does here, will it not also affect the amount of background blur and hence our sense of subject isolation in the same way as a smaller entrance pupil does?
Yes.
A very important point here is to remember that the relative f-number as specified by the manufacturer is a property of the lens being focused at infinity. With an internal-focus design, the pupil ratio can vary significantly between near and far focus, and this is part of the reason why the Canon and Sigma lenses behave differently even though they are shot at the "same" f-number.
Macro shooters are familiar with this notion as it relates to the effective f-number. The situation is complex because the way a lens behaves at near focus depends not only on focal length but on the way it is designed--where its exit pupil lies, and even how it focuses.
A very important point here is to remember that the relative f-number as specified by the manufacturer is a property of the lens being focused at infinity. With an internal-focus design, the pupil ratio can vary significantly between near and far focus, and this is part of the reason why the Canon and Sigma lenses behave differently even though they are shot at the "same" f-number.
Macro shooters are familiar with this notion as it relates to the effective f-number. The situation is complex because the way a lens behaves at near focus depends not only on focal length but on the way it is designed--where its exit pupil lies, and even how it focuses....Show more →
Thank you again, guys.
When shooting landscapes (stopped down for maximum DoF, focus near infinity), my concerns are:
Do different lens designs affect the DoF/diffraction ratio ?
With a specific lens design, can I get less light in for a given DoF under these conditions? ( I want long shutter speeds and would love a native ISO 25 sensor)
...
No matter how you look at it, the indications for the Sigma T-stop are that is does not transmit a lot of light for it's nominal aperture.
The OP's shots wide open are taken with shutter speeds 1/10 (Sigma) and 1/15 (85L). The brightness in the image center is very close between those pictures, and there are no hidden DPP corrections for native lenses in play here.
Now, the 85L already has a T-stop of 1.45 (0.51 Ev loss rel nominal)
My estimate for the Sigma then lands on: 1.45 * SQRT(15/10) = 1.78 ( close to 0.7 Ev loss rel nominal)
For reference, the Sigma 50 1.4 has T-stop 1.65 (0.45 Ev loss rel. nominal) ,
The Canon 85 f/1.8 has T-stop 2.0 (0.34 Ev loss rel. nominal)
With some error margin, I would not be surprised the T-stop of the Sigma actually is between 1.6 and 1.8, most probably around 1.7.
So it looks like the Sigma 85 might make for a good in-between for the upgrade of the canon 1.8 to 1.2. Or is the early consensus that it is going to be a direct replacement? I get the feeling that it renders like the sigma 50 and its missing out on what makes the 1.2 special. I've read before that the 1.2 is the closest canon to look like a zeiss.
the highlights from the sigma appear to be more solid....a bit more distracting, seems like the 85L have a slight edge on subject isolation, interesting.
theSuede wrote:
Not necessarily... It explains why the Sigma has a lot less falloff vignetting and no edge clipping of the bokeh circles - it has a different pupil magnification, its' exit pupil is further from the image plane. This is not necessarily linked to the size of the front pupil; which decides the F/no and light gathering efficiency. You can do this quite easily by building a positive Galilaean system after/into the main lens' rear geometry. You wait until the beam-packet has tapered off to a smaller point, and then spread them back out again towards the sensor. If your base is a planar/Xenotar, then you need at least 9 lens elements to keep symmetry though. A very small cost compared to the large, and expensive front elements.
I actually think Canon knowingly sacrificed some vignette/bokeh symmetry/CA correction to gain what is the trademark of the 85/1.2L - the very steep falloff from sharpness to 'blurredness'. You get this by increasing pupil magnification ratio, but you loose out on the points I mentioned earlier....Show more →
Right.... I was going to say the exact same thing......
I'm about to go for the sigma. However, this test and discussion revealed something I was not aware of: 1.4 on one lens does not give the same DOF as 1.4 on the other. The Canon has with its 1.2 both a larger aperture but also a smaller DOF at 1.4.
For extreme shallow DOF portraits (my intended use) it seems realistic to compare the Canon @ 1.2 with the Sigma @ 1.4. I made an attempt to quantify (oversimplified) the DOF by measuring the light dishes. The dish radius the Canon is ~about 20% larger then that of the Sigma, the area about 40%.
Not sure how that translates to portraits. Looking at these images, the difference is quite noticable. I would be really interested to see a one-to-one portrait comparison......
Got the lens and have some quick test shots, the lens for sure worth much more than its price, AF speed is very good. Surface finish of the lens body changed to some kind of rubber spray / coat, more plastic feel.
Nice shots KKFung and interesting conversation everyone.
I've owned the 85L2 for a while and love it. However, I would say that the new Sigma can definitely give it a run for it's money. If I were in the market for an 85 I would strongly consider purchasing the Sigma. That being said, as others have expressed as well, I'm still a little uneasy about Sigma's reputation with focus problems. IQ can be great but, if you can't take an in focus picture what's the point. Anyway, time will tell. It has a lot of potential though.
In a perfect world I would love to own the Canon, but I can live with any small difference in IQ for $1000 or so. I love L, but I'll buy anything that can get close for a fraction of the price. And in this case the lens is also well built and has HSM, so AF is fast (and hopefully accurate).
