Big Ga
Veteran Member
105/2 DC ??
-
Welcome to the new forums! Please read this first. For known issues we are working to resolve, click here.
Primes Comparison Map #2 (m4/3 vs Canon vs Nikon)
- Thread starter zuikodude
- Start date
Hi,
Large apertures on FF cameras result in significant light loss due to inefficient FF sensors. See:
http://forums.dpreview.com/forums/read.asp?forum=1041&message=41616576
-Craig
Large apertures on FF cameras result in significant light loss due to inefficient FF sensors. See:
http://forums.dpreview.com/forums/read.asp?forum=1041&message=41616576
-Craig
Great Bustard
Veteran Member
- Messages
- 45,961
- Solutions
- 17
- Reaction score
- 34,046
It's not simply FF sensors, but APS-C sensors as well, as a result of inefficient microlens coverings:
http://www.dxomark.com/index.php/Publications/DxOMark-Insights/F-stop-blues
Unfortunately, there's no data for 4/3 or mFT sensors. In any case, it's only an issue below f/2, and I'm afraid that favors the larger sensor systems, since they use higher f-ratios for equivalent settings.
That is, where an mFT camera would be at f/1.4 for a particular DOF, FF would shoot f/2.8 where the light loss is non-existent. If the issue affects mFT sensors as well, that would extend the noise advantage of FF over mFT.
On the other hand, assuming that the issue does not affect mFT sensors (and, again, there is not data on this one way or another), it will eat into the FF noise advantage when shooting at wide apertures (f/1.4 and wider).
Rriley
Veteran Member
- Messages
- 21,846
- Solutions
- 1
- Reaction score
- 474
and in the case of 7D, that is a gapless cover yet it still loses 0.94 of a stop, and that isnt the worst.
since only lenses are calculated at the T Stop, the only effective nomenclature is Ef stop.
as DxO put it
“We can suspect,” Guichard continued, “that sensors collect the incoming light all the more improperly, in that this light comes from a more oblique angle. Since faster lens have, by definition, a wider opening, they raise the proportion of oblique light, hence the proportion of lost energy which never lands on the pixels.”
the losses can be seen here
--
Riley
any similarity to persons living or dead is coincidental and unintended
support 1022 Sunday Scapes'
1) I provided a link to the OM-D data: http://forums.dpreview.com/...forums/read.asp?forum=1041&message=41616576
2) This is a chart of lenses; if the chart shows f/1.2 for full frame, you need to temper that with the light loss data which is significant for FF sensors. For the OM-D you can safely ignore it for the lenses in this chart (because it's -0.1EV only at f/1.4).
2) This is a chart of lenses; if the chart shows f/1.2 for full frame, you need to temper that with the light loss data which is significant for FF sensors. For the OM-D you can safely ignore it for the lenses in this chart (because it's -0.1EV only at f/1.4).
Great Bustard
Veteran Member
- Messages
- 45,961
- Solutions
- 17
- Reaction score
- 34,046
Like I said, 4/3 and mFT sensors were not tested, and since the issue affect both FF and APS-C, there's little reason to expect that 4/3 and mFT are exempt. In fact, it's even possible that Olympus was aware of the issue long before, and this is why they chose not to make lenses faster than f/2 for 4/3.
Great Bustard
Veteran Member
- Messages
- 45,961
- Solutions
- 17
- Reaction score
- 34,046
Where'd you get this chart:
It's not hosted on DxOMark. Seems you'd want to link the source, right?
Assuming the chart is valid, we see that at f/1.2, the EM5 reduces the 5D2 advantage by 0.45 stop, and a by 0.33 stops at f/1.4. But, the 5D2 retains the overall advantage.
However, for the same DOF, the 5D2 would be at f/2 when the EM5 is at f/1, where it is a non-issue all together, so there is no disadvantage with FF for Equivalent settings (same DOF and shutter speed), except inasmuch as the EM5 sensor is more efficient than the 5D2 sensor (higher QE and lower read noise), but DxOMark has yet to produce those results yet.
Of course, we would expect the new EM5 to have a more efficient sensor than the previous generation 5D2, so it makes more sense, in an mFT vs FF "battle" to compare sensors of the same generation, e.g., the 5D3 and D800.
highwave
Veteran Member
Great work
Very informative chart and nicely done as well
It seems m43 is actually lacking in fast primes faster than 1.4
Let's just hope third party companies introduce more of them with autofocus to boot. Certainly recent rumors suggest so.
And for all those discussions on doubling focal length doubling f number blah blah blah. Whenever I read your arguments this picture pops in my mind
http://t3.gstatic.com/images?q=tbn:ANd9GcS3UoTDObNYoME80mPpj9NpILPfyXUOdZ6bEp7y_CRT2bds3M1e
I think it was snapped by Anal Adams.
As said in another thread the equivalency war is over. You lost. Get over it already. We don't want to discuss this topic anymore.
Very informative chart and nicely done as well
It seems m43 is actually lacking in fast primes faster than 1.4
Let's just hope third party companies introduce more of them with autofocus to boot. Certainly recent rumors suggest so.
And for all those discussions on doubling focal length doubling f number blah blah blah. Whenever I read your arguments this picture pops in my mind
http://t3.gstatic.com/images?q=tbn:ANd9GcS3UoTDObNYoME80mPpj9NpILPfyXUOdZ6bEp7y_CRT2bds3M1e
I think it was snapped by Anal Adams.
As said in another thread the equivalency war is over. You lost. Get over it already. We don't want to discuss this topic anymore.
boggis the cat
Veteran Member
Not to mention perfectly incomprehensible to everyone used to thinking of angle of view in 135 format focal length terms.
boggis the cat
Veteran Member
Coroander supplies a table here:
http://forums.dpreview.com/forums/read.asp?forum=1041&message=41616576
There was an engineering decision made. The f/2 limit concerned maintaining telecentricity, as far as I am aware. This was prior to micro-lenses being introduced, and Olympus now seem happy to ignore the f/2 limit on Micro FT.
This problem appears "fixable" to at least some extent, as you can see from the comparison of the EV drop of the Canon 40D and 7D (the 7D showing a marked improvement).
Due to this it is tricky to come up with an appropriate "derating" for very fast lenses.
Rriley
Veteran Member
- Messages
- 21,846
- Solutions
- 1
- Reaction score
- 474
the problem is off axis light which the microlenses are not containing, the very thing telecentric was meant to overcome. The microlenses being gapless means thay have to work on the lightpipe below, as their is no further room on the surface of the sensor, they cant get bigger than they are..
well, perhaps the fix is to do nothing and just nudge the ISO to fool people that all is ok
--
Riley
any similarity to persons living or dead is coincidental and unintended
support 1022 Sunday Scapes'
boggis the cat
Veteran Member
This "advantage" only applies to shallower DOF in practical terms. It could just be that we're looking at relative apertures here, given that the problem being discussed relates to relative aperture.
It appears that thinking of everything in "equivalent photograph" terms causes an inability to think outside that little (and never used in practice) conceptual box.
bobn2
Veteran Member
Every single Olympus DSLR has had microlenses on the sensor, as has every single DSLR made by anyone else, with the exception of the Kodak 14MP FF DSLR's. Same applies to compacts. Microlenses were introduced, way, way back before still cameras were digital.
--
Bob
boggis the cat
Veteran Member
That is indeed the current "remedy".
But my point is that a 40D is affected worse than a 7D, so how would you propose correcting the lenses?
It is, in fact, a sensor-related issue -- just as the "Ef/" vs "f/" difference has nothing to do with the lenses, but is sensor-related.
The problem with "correcting" lenses for sensor problems is that people then "correct" for sensor differences on top of the lens correction. How many times have we seen this "double dipping" from some ride-along clown doing a drive-by on the OSTF?
Best to know what is sensor related, and what is lens related, and understand how to fit those two parts together. (Pretty much what we all do. I know that my E-5 will out-perform my E-1 on almost everything, but if I want a nice solid blue noise-free sky at base ISO from an OOC JPEG then the E-1 has an advantage. Know what you're using and you'll have a better chance of getting the photograph that you are trying for.)
bobn2
Veteran Member
The problem is not 'off-axis light'. The problem is the angle of the light cone projected from the exit pupil. Telecentricity has nothing at all to do with it.
The issue is the speed of the microlenses, which needs to be faster than the taking lens by a factor of the linear fill factor of the silicon underneath. The aperture of the microlens can't be larger than the pixel size, so the only way of reducing the f-number is reducing the FL which means a low stack height.
Panasonic sensors have an architecture which reduces the wiring required for a pixel, so can have a lower stack height.
--
Bob
Rriley
Veteran Member
- Messages
- 21,846
- Solutions
- 1
- Reaction score
- 474
40D probably has poorer microlenses on what is an older sensor, the first time I saw Canon talk about gapless microlenses was on 7D. One could assume this makes a difference. Still that doesnt advance the situation from 7D which might be as good as it gets.
wel thats right as the T stop of the lens wont change b/se it is over a smaller sensor
which is sure to continue, denial is the best fit when the facts are no longer observed
--
Riley
any similarity to persons living or dead is coincidental and unintended
support 1022 Sunday Scapes'
bobn2
Veteran Member
Sorry, don't make the Gestapo look like an eccentric, they where really bad guys.
I would say DoF Stasi (to stay with German history): Once you said something not totally in line with The Party (here: The red book of FF) they file the issue and keep watching you and will subvert your credibility.
Rriley
Veteran Member
- Messages
- 21,846
- Solutions
- 1
- Reaction score
- 474
preamble
Our measurements all point in the same direction: as you go further than f 4 – to f 2 and wider, the accrued quantity of light falls marginally onto the sensor. A stronger and stronger part of this additional light is blocked or lost. I am therefore inclined to question the real benefit of faster lenses.”
condition
“We can suspect,” Guichard continued, “that sensors collect the incoming light all the more improperly, in that this light comes from a more oblique angle. Since faster lens have, by definition, a wider opening, they raise the proportion of oblique light, hence the proportion of lost energy which never lands on the pixels.”
oblique angle = off axis light
which would rule out stacked microlenses
the issues have been well known for some time, and quite exactly are meant to deal with the situation you previously describe the angle of the light cone projected from the exit pupil
Joseph W
Ordinary retrofocus lenses move the exit pupil just far enough to clear an SLR mirror. A lens can be made more "digital friendly" if the exit pupil is moved farther than this. For example, the Nikon 17-35mm f2.8 lens has an exit pupil that ranges from 98mm (at the wide 17mm setting) to 78mm at the 35mm telephoto setting. The Sigma DG lenses also feature such "extreme" 80mm exit pupils. Camera and lens companies refer to such lenses as "telecentric". This is a slight exaggeration: a true telecentric lens has an exit pupil an infinite (mathematically, anyway) distance from the sensor. This causes the light to arrive perpendicular to the sensor. It is also costly, and creates problems, so 80-90mm "near telecentric" lenses are considered the norm.
This table shows that a lens with an exit pupil 52mm from the film plane (typical of Nikon or Canon wide angles) has the potential for severe vignetting on a 1.0x or 1.3x crop camera, and noticeable vignetting on 1.5x or 1.6x cameras. (1.7x and 2.0x cameras are essentially immune to such vignetting). Increasing the exit pupil to 80mm means that the 1.0x camera may have objectionable vignetting, but 1.3x to 2x cameras will have no vignetting at all.
higher density sensors just seem to be rejigging the conditions, for which I assume some makers arent going to care. The conditions super fast lenses setup are a minority issue, and it happens where the light is the most abundant anyway, hence where the system is best able to tolerate less light.
--
Riley
any similarity to persons living or dead is coincidental and unintended
support 1022 Sunday Scapes'
boggis the cat
Veteran Member
I seem to recall microlenses being introduced as a new "feature" sometime in the early 2000's. That would have been after Olympus were through the main design phase for the FT system.
In any case, if Olympus designed the FT system based on assumptions about existing microlenses (instead of no microlenses) then that would still make sense of their decision to pursue telecentricity, and eschew lenses faster than f/2.
Whether that turned out to be a competitive difference is another issue, and it seems that f/2 is no longer a boundary for Micro FT. (The reduced sensor to mount distance would make telecentricity a lot more difficult, so I guess it got binned as a design goal.)
Keyboard shortcuts
- f
- Forum