kenw
Veteran Member
Title typo - sorry - as the post indicates I have NOT seen this effect with a legacy 50/1.4.
--
Ken W
See plan in profile for equipment list
--
Ken W
See plan in profile for equipment list
-
Welcome to the new forums! Please read this first. For known issues we are working to resolve, click here.
F-stop blues
- Thread starter JDittlen
- Start date
forpetessake
Veteran Member
Can you post some pictures, like a white sheet? If the theory is correct the periphery of the sensor should get less light than the center due to this large apperture effect, so one would observe noticeable vignetting in uncorrected lenses.
kenw
Veteran Member
But that would be impossible to distinguish from vignetting from the lens itself, so not a useful thing to do. If the DxO theory applies then it applies at the center of the image. As you say, it would get worse the further off axis you get but there is no way to measure this - or rather to separate the effect of the micro-lenses from the effect of light fall-off from the lens (which is rather severe on all wide aperture lenses).
--
Ken W
See plan in profile for equipment list
bobn2
Forum Pro
Unfortunately for most of us here, if anyone acceded to your request it would still count as murder under most jurisdictions. The law is a ass.
--
Bob
Detail Man
Forum Pro
- Messages
- 17,489
- Solutions
- 11
- Reaction score
- 2,745
Oh, man, and I was so enjoying dreaming about having over 16 Million little "active-pixels" charge-pumping away on my dusty planetoid inhabited wonder-widget. Are you saying that they aren't even alive, man ? That ... that would mean that they aren't even "organic", man. Say it ain't so !
Well, even though it's all complimentary MOS, it seems like the moniker "CMOS" really is a "downer"
Thus, it came time for the Marketing Boys to have a long lunch, and blow some pungent hot air ?
Regards,
DM ...
Andy Crowe
Senior Member
I'll give my Voigtlander 0.95 a test tonight, it's always felt like it benefits from the wider aperture but I've never actually done any controlled tests with it
bobn2
Forum Pro
More good practical evaluations of this phenomenon would be welcome. My belief from the theory is that you will see a limitation both in the expected exposure and DOF with respect to nominal. Not a sharp cut-off, but certainly a fall-off.
--
Bob
joergzielosko
Forum Enthusiast
Vignetting should not be a problem with legacy lenses (which are designed for full frame film).
I read few years ago in a french photographic magazine (can't find the reference, sorry) that it's something common to ASP-C sensors too (can't remember if they talked about FF). They came to the same conclusion about wide aperture and light loss in digital photography.
That being said, I believe that even though a part of the light is lost you still get more light and therefore rising shutter speed. Plus, you get more DOF control which is the main reason for me to buy those lenses.
--
L'expérience est une lampe accrochée dans le dos, elle n'éclaire que le chemin parcouru. Lao Tseu
http://yinetyang.deviantart.com/
That being said, I believe that even though a part of the light is lost you still get more light and therefore rising shutter speed. Plus, you get more DOF control which is the main reason for me to buy those lenses.
--
L'expérience est une lampe accrochée dans le dos, elle n'éclaire que le chemin parcouru. Lao Tseu
http://yinetyang.deviantart.com/
kenw
Veteran Member
Actually it is still there, just less severe. All lenses have cos4 light fall off at all apertures (even small ones). Actually many legacy lenses, especially the range-finder lenses, have particularly bad cos4 fall off because of their close exit pupil. You'll never be able to distinguish fall off from the lens and fall off from the micro lens.
For example, the Zeiss 50/2 which is a full frame lens on m43 has almost 0.5EV of vignetting at F/2:
http://www.photozone.de/olympus--four-thirds-lens-tests/469-zeiss_zm_50_2_43?start=1
Since the question is about what happens at very large apertures you'll find all lenses, even legacy lenses, have significant vignetting at the edges of m43 - at least significant compared to the effect everyone is trying to measure.
--
Ken W
See plan in profile for equipment list
More replications would definitely be better. DxO's evidence for the effect seems pretty compelling, but there's also a lot of variance - anywhere from .25 to .8 EV of exposure for an extra stop of lens speed.
Tim F 101
Senior Member
Different sensors have different architecture. Old-fashioned sensors have the deepest pixel wells / thickest wires between pixels and those would suffer the worst. Newer sensors have thinner wires that would block less wide-aperture light. Back-illuminated sensors put the wiring on the non-light detecting side (a more expensive design to make) and would eliminate the problem almost completely.
This is very interesting.
Just a niggle on terminology: the tests don't show no benefit; they show diminishing returns.
If I'm interpreting the axes correctly, you get anywhere from .25 to .8 EV real exposure increase by going from f2 to f1.4 - i.e. less than the stop you'd expect. The next half stop to f1.2 gets you another .15 to .2 EV. Mfg seem to be aware of this, as they up the sensor gain to compensate - which seems a bit sneaky to do without disclosure.
Just a niggle on terminology: the tests don't show no benefit; they show diminishing returns.
If I'm interpreting the axes correctly, you get anywhere from .25 to .8 EV real exposure increase by going from f2 to f1.4 - i.e. less than the stop you'd expect. The next half stop to f1.2 gets you another .15 to .2 EV. Mfg seem to be aware of this, as they up the sensor gain to compensate - which seems a bit sneaky to do without disclosure.
rare wolf
Leading Member
If what DXO is referring to is about the sensor's reaction to oblique light, Zuiko FT lenses were claimed to be especially designed for digital sensors (i.e., not film) delivering telecentric (parallel) light to the FT sensor. That is, Zuiko was quite aware of the differences between legacy lenses designed for film and oblique light causing problems like vignetting with sensors, .... That said, I have no idea how this might carry over to Pany lenses of Zuiko mFT, but I imagine it does.
--
cheerios from the Avalon Peninsula, Newfoundland
http://www.michael.shaffer.net/albums.html
jimmyfergus
Well-known member
Surely the effect won't be so marked on legacy (SLR) lenses? Since the distance from the rearmost element to the sensor is large, incident light is closer to perpendicular to the sensor.
Which might imply your OM 50mm f1.8 might get significantly more usable light to the sensor than the Oly 45mm f1.8...
Bryan Biggers
Senior Member
As usual, DXO publishes a test that can't be directly interpreted in real life, and it isn't clear what their methodology was. Plenty to argue about here.
1) All lenses have light falloff at the edges, it is just geometry and having a flat sensor. Larger apertures are worse. It has always been this way, but it is usually negligible except at very wide angles or large apertures..
2) As was pointed out, you still get a net light gain, even if it is not exactly the 1 stop gain that you expected, it is still beneficial and not a reason to avoid/ not covet large aperture lenses.
3) Different sensors have different micro lens architectures to try to take advantage of the incoming light angles, especially at the edges. Some might have more fall off than others.
4) I tried a 50mm legacy lens up against a light box on a micro 4/3 camera with fixed ISO. 1/200 at f/2 and 1/100 at f/1.4, just what you would expect with this lens. If there is an effect, it is small with that particular lens anyway.
1) All lenses have light falloff at the edges, it is just geometry and having a flat sensor. Larger apertures are worse. It has always been this way, but it is usually negligible except at very wide angles or large apertures..
2) As was pointed out, you still get a net light gain, even if it is not exactly the 1 stop gain that you expected, it is still beneficial and not a reason to avoid/ not covet large aperture lenses.
3) Different sensors have different micro lens architectures to try to take advantage of the incoming light angles, especially at the edges. Some might have more fall off than others.
4) I tried a 50mm legacy lens up against a light box on a micro 4/3 camera with fixed ISO. 1/200 at f/2 and 1/100 at f/1.4, just what you would expect with this lens. If there is an effect, it is small with that particular lens anyway.
Tim F 101
Senior Member
Thanks to everyone (and me) for doing some seat of the pants testing.
Various people have found that micro 4/3 sensors deliver one stop of light gain for one stop of aperture all the way down to f/1.4. Tests have yet to show that you gain the full benefit between f/1.4 and f/0.95, but we clearly gain some benefit and the shorter DOF effect is clear. Even the few us who own lenses faster than 1.4 shoot at or above 1.4 most of the time since you need to stop most lenses down a bit to get peak sharpness.
Conclusion: people who own m4/3 cameras should not worry about the DXO article.
Various people have found that micro 4/3 sensors deliver one stop of light gain for one stop of aperture all the way down to f/1.4. Tests have yet to show that you gain the full benefit between f/1.4 and f/0.95, but we clearly gain some benefit and the shorter DOF effect is clear. Even the few us who own lenses faster than 1.4 shoot at or above 1.4 most of the time since you need to stop most lenses down a bit to get peak sharpness.
Conclusion: people who own m4/3 cameras should not worry about the DXO article.
He is talking about F-stop =2 or wideris he talking about?Good evidence that life, in all of it's myriad enchantments towards distractions, is indeed more a "state of mind" than a "state of matter/energy". The triumph of form over function is complete
Tedolph
--
MFT in progress
kenw
Veteran Member
Well not quite, but you've almost got the right idea. What matters is not the distance from the rearmost element, but the distance of the exit pupil. This actually has nothing to do with the position of the rear most element, in fact depending on whether the rear most element is positive or negative it can make it further or closer.
"Native" m43 lenses are actually designed to have a very distant exit pupil (also referred to as being "telecentric"). Legacy lenses, even SLR ones, are not necessarily designed as such. That said, you wouldn't expect a legacy 50 to have a particularly close exit pupil. On the other hand, it wouldn't at all be surprising to find that a m43 lens has just as distant or even a more distant exit pupil than that legacy 50!
Also, the exit pupil distance issue is only applicable for what would happen at the edges of the sensor. At the center the exit pupil distance is not relevant and all that matters is the aperture number being used. As referenced elsewhere in the thread you could only reliably measure this micro lens effect at the center of the image anyway, so exit pupils aren't really a factor for what folks are measuring in this particular case. (But they are a big deal at the edges for a whole bunch of different reasons beyond light fall off).
--
Ken W
See plan in profile for equipment list
bobn2
Forum Pro
Keyboard shortcuts
- f
- Forum