Purple flare: Causes and remedies (continued)

Anders W said:

Anders W wrote:
The idea I am toying with here is to add something behind the last lens element so as to allow, if possible, only light belonging to that part of the image circle that is covered by the sensor to proceed any further.

Click to expand...

I think the problem will be that a baffle any practical distance from the sensor will be unlikely to be effective. To eliminate the reflection it will be small enough to cause severe vignetting. Basically unless the baffle is almost sitting on the sensor the penumbra of the baffle is too large. I haven't done the geometry for this case with an F/4 lens but I did do it for F/1.4 50mm lenses and a baffle between the lens and sensor just wasn't practical. Still, you might try out the geometry for the 7-14 - be sure to use the actual exit pupil distance and size. There is a chance with the smaller aperture and relatively distant exit pupil you could get something to be effective.


Looking in the camera I of course couldn't be sure what causes the suspected reflection but my first suspicion was that it was the sensor shaped baffle itself that sits just a few mm above the sensor. I wondered if the edge of the baffle was reflecting/scattering. If that is the case a very daring person my try putting something on that edge - I certainly won't be trying it! Also one could try shining a bright light in that region while looking as off axis as possible (or even using the sensor glass itself as a poor mirror) to see if there is anything obvious happening.

Sadly losing time to keep up with this thread, but your most recent example is certainly interesting!

Anders W said:

Anders W wrote:

Anders W said:

Macx wrote:

Anders W said:

One other thought occurred to me here: If Ken W is right in his conclusion here
http://forums.dpreview.com/forums/post/50965381


as I think he might well be, that purple flare resulting from the mechanisms I outlined in the previous thread come from reflections off the area inside the camera that surrounds the sensor and on to the sensor with a very large angle of incidence, then we might be able to get rid of that type of flare (which I would say is the most bothersome type if the diffuse purple blobs that the 7-14 tends to produce belongs to that category) by putting a baffle behind the rearmost element of the 7-14 such as to prevent these reflections. The idea I am toying with here is to add something behind the last lens element so as to allow, if possible, only light belonging to that part of the image circle that is covered by the sensor to proceed any further.

Click to expand...

You mean on the glass of the rear element itself? Because the "mirror box" seems designed to minimize any such reflections in the first place, short of covering it in pristine black velvet.

Click to expand...

Not on the glass. Behind the glass. If you look at the area surrounding the sensor on the E-M5, you can see that it is designed such that light rays belonging to parts of the image circle that are outside the sensor may reflect on to the sensor and arrive at very large angles of incidence. Even if the baffle we add is no less reflective than the "mirror box", that wouldn't matter as long as we can prevent whatever it reflects from hitting the sensor. Think of a piece of black cardboard placed between the rear element and the sensor with a slit of the right dimensions so as to let only those rays that are destined for the sensor pass and the rest being absorbed by the cardboard or reflected back on the rear section of the lens. I don't know yet whether this idea is realistic or whether it will run into unsurmountable problems of one kind or another but at least it might be worth exploring.

Click to expand...

Ah, I see. Yes, worth exploring. Baffles inside the lens would probably be more effective, but not practical except by stopping down. Here is a pretty extensive presentation about analysing and eliminating stray light: http://www.photonengr.com/files/2011/05/20110510StrayLightWebinar.pdf

I also ran into this, that I thought you might find interesting. On page four it shows how memory cells on the face of the sensor can work as photodiodes and capture stray light: http://www.aptina.com/products/technology/Aptina_Global-Shutter-WhitePaper.pdf Now, this is supposedly a weakness with sensors with global shutters because of their particular design, but...

Anders W said:

Anders W wrote:

Anders W said:

Macx wrote:

Anders W said:

Anders W wrote:

But what do you have in mind when you say that BSI would eliminate the problem? That the CFA could be placed closer to the photodiodes thus reducing crosstalk due to diffusion? You might well be right but is this just an idea of your own or do you have a source discussing the issue?

Click to expand...

Simply that the main source of reflection is the wiring in front of the photo diodes on the front-lit designs

Click to expand...

Yes, that makes sense. Where did you find the illustration? I can read German without much difficulty

Click to expand...

That was just a random piece of Sony press material about their Exmor R sensors. I was thinking of you mentioning iPhones having a similar problem, and they are supposedly back-lit, so I'm not quite sure what to make of that.


I found the illustration in this thesis that gives you a pretty good overview of the different components of a modern CMOS sensor: Design of a high DR CMOS...

kenw said:

kenw wrote:

kenw said:

Anders W wrote:
The idea I am toying with here is to add something behind the last lens element so as to allow, if possible, only light belonging to that part of the image circle that is covered by the sensor to proceed any further.

Click to expand...

I think the problem will be that a baffle any practical distance from the sensor will be unlikely to be effective. To eliminate the reflection it will be small enough to cause severe vignetting. Basically unless the baffle is almost sitting on the sensor the penumbra of the baffle is too large.

Click to expand...

Yes, I fear you might be right about that. This is the main problem I could foresee. For practical reasons, we would like the baffle to be close to the rearmost element but to get the job done properly it might have to be very close to the sensor.

kenw said:

I haven't done the geometry for this case with an F/4 lens but I did do it for F/1.4 50mm lenses and a baffle between the lens and sensor just wasn't practical. Still, you might try out the geometry for the 7-14 - be sure to use the actual exit pupil distance and size. There is a chance with the smaller aperture and relatively distant exit pupil you could get something to be effective.

Click to expand...

Could you please elaborate a bit on what exactly you mean when you say "use the actual exit pupil distance and size" and how you would go about measuring them.

kenw said:

Looking in the camera I of course couldn't be sure what causes the suspected reflection but my first suspicion was that it was the sensor shaped baffle itself that sits just a few mm above the sensor. I wondered if the edge of the baffle was reflecting/scattering. If that is the case a very daring person my try putting something on that edge - I certainly won't be trying it! Also one could try shining a bright light in that region while looking as off axis as possible (or even using the sensor glass itself as a poor mirror) to see if there is anything obvious happening.

Click to expand...

Yes, that might be worth a try.

kenw said:

Sadly losing time to keep up with this thread, but your most recent example is certainly interesting!

Click to expand...

Glad you found it interesting. Let me know of any comments/ideas you might have.

Macx said:

Macx wrote:

Macx said:

Anders W wrote:

Macx said:

Macx wrote:

Macx said:

One other thought occurred to me here: If Ken W is right in his conclusion here
http://forums.dpreview.com/forums/post/50965381

as I think he might well be, that purple flare resulting from the mechanisms I outlined in the previous thread come from reflections off the area inside the camera that surrounds the sensor and on to the sensor with a very large angle of incidence, then we might be able to get rid of that type of flare (which I would say is the most bothersome type if the diffuse purple blobs that the 7-14 tends to produce belongs to that category) by putting a baffle behind the rearmost element of the 7-14 such as to prevent these reflections. The idea I am toying with here is to add something behind the last lens element so as to allow, if possible, only light belonging to that part of the image circle that is covered by the sensor to proceed any further.

Click to expand...

You mean on the glass of the rear element itself? Because the "mirror box" seems designed to minimize any such reflections in the first place, short of covering it in pristine black velvet.

Click to expand...

Not on the glass. Behind the glass. If you look at the area surrounding the sensor on the E-M5, you can see that it is designed such that light rays belonging to parts of the image circle that are outside the sensor may reflect on to the sensor and arrive at very large angles of incidence. Even if the baffle we add is no less reflective than the "mirror box", that wouldn't matter as long as we can prevent whatever it reflects from hitting the sensor. Think of a piece of black cardboard placed between the rear element and the sensor with a slit of the right dimensions so as to let only those rays that are destined for the sensor pass and the rest being absorbed by the cardboard or reflected back on the rear section of the lens. I don't know yet whether this idea is realistic or whether it will run into unsurmountable problems of one kind or another but at least it might be worth exploring.

Click to expand...

Ah, I see. Yes, worth exploring. Baffles inside the lens would probably be more effective, but not practical except by stopping down. Here is a pretty extensive presentation about analysing and eliminating stray light: http://www.photonengr.com/files/2011/05/20110510StrayLightWebinar.pdf

I also ran into this, that I thought you might find interesting. On page four it shows how memory cells on the face of the sensor can work as photodiodes and capture stray light: http://www.aptina.com/products/technology/Aptina_Global-Shutter-WhitePaper.pdf Now, this is supposedly a weakness with sensors with global shutters because of their particular design, but...

Click to expand...

Thanks for the links. Haven't had the time to look at them yet but will when I have a moment to spare. As to baffles: No, we need a baffle behind rather than inside the lens. As Ken pointed out, the problem here is that the baffle may have to sit so close to the sensor in order to do the job properly (i.e., let through all rays that actually contribute to the sensor image but only those) that it becomes practically impossible (e.g., because it sticks out so far from the lens that the rotation required when mounting it is no longer possible).

kenw said:

kenw wrote:

Looking in the camera I of course couldn't be sure what causes the suspected reflection but my first suspicion was that it was the sensor shaped baffle itself that sits just a few mm above the sensor. I wondered if the edge of the baffle was reflecting/scattering. If that is the case a very daring person my try putting something on that edge - I certainly won't be trying it! Also one could try shining a bright light in that region while looking as off axis as possible (or even using the sensor glass itself as a poor mirror) to see if there is anything obvious happening.

Click to expand...

I had a look at what happens when you have a spotlight illuminate different parts of the area surrounding the sensor but not the sensor itself. Signs of purpleness starts to appear as soon as the light hits anything inside the mount, before it reaches the edge of the innermost baffle that you are talking about (whose shininess makes it a particularly strong suspect). So I guess other parts of the interior may be partly responsible as well.

Anders W said:

Anders W wrote:

Anders W said:

I haven't done the geometry for this case with an F/4 lens but I did do it for F/1.4 50mm lenses and a baffle between the lens and sensor just wasn't practical. Still, you might try out the geometry for the 7-14 - be sure to use the actual exit pupil distance and size. There is a chance with the smaller aperture and relatively distant exit pupil you could get something to be effective.

Click to expand...

Could you please elaborate a bit on what exactly you mean when you say "use the actual exit pupil distance and size" and how you would go about measuring them.

Click to expand...

To determine what the shadow of the baffle will be you need to draw the exit pupil of the correct size and relation to the sensor. This link gives a good illustration of the pupils and a few useful equations:


You can see the pupil ratio is the key thing to measure. Once you have that you can determine the exit pupil position from the focal length as well as its size for a given aperture setting. The rear principle plane is of course determined from the focal length and the lens flange.

To measure P you can place the lens a good distance from a camera and shoot the pupils. Measure them in the images and compute the ratio. With the lens far from the camera any errors due to different principle plane locations will be neglible.

kenw said:

kenw wrote:

kenw said:

Anders W wrote:

kenw said:

I haven't done the geometry for this case with an F/4 lens but I did do it for F/1.4 50mm lenses and a baffle between the lens and sensor just wasn't practical. Still, you might try out the geometry for the 7-14 - be sure to use the actual exit pupil distance and size. There is a chance with the smaller aperture and relatively distant exit pupil you could get something to be effective.

Click to expand...

Could you please elaborate a bit on what exactly you mean when you say "use the actual exit pupil distance and size" and how you would go about measuring them.

Click to expand...

To determine what the shadow of the baffle will be you need to draw the exit pupil of the correct size and relation to the sensor. This link gives a good illustration of the pupils and a few useful equations:

http://coinimaging.com/blog1/?p=127

You can see the pupil ratio is the key thing to measure. Once you have that you can determine the exit pupil position from the focal length as well as its size for a given aperture setting. The rear principle plane is of course determined from the focal length and the lens flange.

To measure P you can place the lens a good distance from a camera and shoot the pupils. Measure them in the images and compute the ratio. With the lens far from the camera any errors due to different principle plane locations will be neglible.

Click to expand...

Fine! Thanks!

Anders W said:

Anders W wrote:

Anders W said:

kenw wrote:

Looking in the camera I of course couldn't be sure what causes the suspected reflection but my first suspicion was that it was the sensor shaped baffle itself that sits just a few mm above the sensor. I wondered if the edge of the baffle was reflecting/scattering. If that is the case a very daring person my try putting something on that edge - I certainly won't be trying it! Also one could try shining a bright light in that region while looking as off axis as possible (or even using the sensor glass itself as a poor mirror) to see if there is anything obvious happening.

Click to expand...

I had a look at what happens when you have a spotlight illuminate different parts of the area surrounding the sensor but not the sensor itself. Signs of purpleness starts to appear as soon as the light hits anything inside the mount, before it reaches the edge of the innermost baffle that you are talking about (whose shininess makes it a particularly strong suspect). So I guess other parts of the interior may be partly responsible as well.

Click to expand...

That perfectly fits the idea that it is very shallow angle light that it causing it. One could use your idea of a baffle to an extreme and add a kind of funnel from the last lens element to the sensor. Yes, it would vignet the image to down to a circular one but if it eliminates the purple blobs we knew that reflects back from the sensor are unlikely to be the cause.

Anders W said:

Anders W wrote:

Anders W said:

_sem_ wrote:

Anders W said:

Anders W wrote:

Anders W said:

_sem_ wrote:

However, try to tell apart the regular on-axis flares (regardless of tint) with the patterned flare which is not on the axis.

Click to expand...

Is it something like this you are asking for?

http://forums.dpreview.com/forums/post/50972762

Click to expand...

Looks a bit more clear, but still not conclusive.

Click to expand...

What is not conclusive? The regular on-axis flare reacts in a completely different way to my filter treatments than the other type of flare.

Click to expand...

Whether this is a diffraction pattern (like the red dot flare) or not. Notice there are other special types of flare, for instance the "sunstars" (reflections from aperture blades), reflections from lens element edges, reflections from lens surfaces sealed very close together (concentric circles) etc.

Anders W said:

Anders W said:

Can you try using a very small light-source? This should turn a pattern of blobs towards a pattern of dots if this is really diffraction.

Click to expand...

It is not clear to me how and why diffraction could explain any of the flares shown in the post I linked to. I suggest two sets of mechanisms to explain the outcome of my little experiments, each of which linked to a particular type of flare as illustrated in that post: color-channel pollution combined with color-channel imbalance as outlined here

http://forums.dpreview.com/forums/post/50958004

and particular sensitivity to certain wave-lengths of blue light. Are you suggesting a third model? If so, could you please elaborate what it amounts to and what makes you think it might fit the evidence better than the two I have outlined.

Click to expand...

No this is not another model. You're thinking about the cause while a diffraction pattern is a consequence. In crystallography and such, folks are studying diffraction patterns to determine the underlying crystal structures.

noirdesir said:

noirdesir wrote:

noirdesir said:

Anders W wrote:

noirdesir said:

kenw wrote:

Looking in the camera I of course couldn't be sure what causes the suspected reflection but my first suspicion was that it was the sensor shaped baffle itself that sits just a few mm above the sensor. I wondered if the edge of the baffle was reflecting/scattering. If that is the case a very daring person my try putting something on that edge - I certainly won't be trying it! Also one could try shining a bright light in that region while looking as off axis as possible (or even using the sensor glass itself as a poor mirror) to see if there is anything obvious happening.

Click to expand...

I had a look at what happens when you have a spotlight illuminate different parts of the area surrounding the sensor but not the sensor itself. Signs of purpleness starts to appear as soon as the light hits anything inside the mount, before it reaches the edge of the innermost baffle that you are talking about (whose shininess makes it a particularly strong suspect). So I guess other parts of the interior may be partly responsible as well.

Click to expand...

That perfectly fits the idea that it is very shallow angle light that it causing it.

Click to expand...

Yes it does. The idea about reflections off the camera baffle is persuasive because it readily explains how the angles of incidence can become large enough for significant color-channel pollution (diffusion crosstalk) to occur. If the light arrived at the sensor directly from the rearmost element, the question is if the angle of incidence would be sufficiently extreme for much pollution to occur, especially if we are considering purple blobs located towards the center of the frame (which is by no means unusual).

On the other hand, one hitherto unanswered question regarding the "reflection-off-the-baffle" part of the theory is why such reflections would be more prevalent with the 7-14 than with other lenses. All lenses have an image circle larger han the sensor. So what's special about the 7-14 here? OK, we have seen that some of the reflections that we think are due to reflections off the baffle, such as the streaks from a light source just outside the frame that I illustrated in the OP of the previous thread. But the larger purple blobs we have seen examples of, e.g., in my second series of tests shots in this thread and in Michael's example in this thread, appear to be more or less unique to the 7-14. So what's special about it in this context? That it has an unusually large image circle? Hmm. If so why?

That the light that leaves the rearmost elements contains an unusually large proportion that has already been reflected somewhere and therefore hits the baffle from the "wrong" angle? Note that the staircase structure of the baffle is, I presume, there to prevent reflections that have a relatively small angle of incidence to be reflected on to the sensor. But light that has already been reflected somewhere upstream might come in at a relatively large angle, thus bypassing this obstacle.

One thing that is indeed clear here is that the 7-14, due to its big and bulbous front element, has an unusually large intake of light, only a very small part of which is actually used to form the image proper. While most of the superfluous light never escapes through the rearmost element, what remains may nevertheless be sufficient to create problems.

noirdesir said:

One could use your idea of a baffle to an extreme and add a kind of funnel from the last lens element to the sensor. Yes, it would vignet the image to down to a circular one but if it eliminates the purple blobs we knew that reflects back from the sensor are unlikely to be the cause.

Click to expand...

Yes, you are right about that. One thing that still puzzles me a bit though is how the cross-shaped purple flare (horizontal and vertical but not diagonal purple streaks) around a light source in the frame is actually created. Is that too due to reflections off the baffle? Or is there some other mechanism involved?

_sem_ said:

_sem_ wrote:

_sem_ said:

Anders W wrote:

_sem_ said:

_sem_ wrote:

_sem_ said:

Anders W wrote:

_sem_ said:

_sem_ wrote:

However, try to tell apart the regular on-axis flares (regardless of tint) with the patterned flare which is not on the axis.

Click to expand...

Is it something like this you are asking for?

http://forums.dpreview.com/forums/post/50972762

Click to expand...

Looks a bit more clear, but still not conclusive.

Click to expand...

What is not conclusive? The regular on-axis flare reacts in a completely different way to my filter treatments than the other type of flare.

Click to expand...

Click to expand...

Hi _sem_,

Some comments below. Note that I am asking just to understand your line of thinking better, not necessarily because I think you are wrong.

_sem_ said:

Whether this is a diffraction pattern (like the red dot flare) or not.

Click to expand...

But do we know that the red-dot pattern is a diffraction pattern? And if so, how do we arrive at that conclusion?

If I recall correctly, you said in a previous post, when I asked, that noone has really managed to provide a convincing explanation of this pattern yet.

_sem_ said:

Notice there are other special types of flare, for instance the "sunstars" (reflections from aperture blades), reflections from lens element edges, reflections from lens surfaces sealed very close together (concentric circles) etc.

Click to expand...

Yes, but I think (without being quite sure) that my most recent set of test shots covers all the types that are at issue as far as the purple-flaring of the 7-14 is concerned.

_sem_ said:

_sem_ said:

_sem_ said:

Can you try using a very small light-source? This should turn a pattern of blobs towards a pattern of dots if this is really diffraction.

Click to expand...

It is not clear to me how and why diffraction could explain any of the flares shown in the post I linked to. I suggest two sets of mechanisms to explain the outcome of my little experiments, each of which linked to a particular type of flare as illustrated in that post: color-channel pollution combined with color-channel imbalance as outlined here

http://forums.dpreview.com/forums/post/50958004

and particular sensitivity to certain wave-lengths of blue light. Are you suggesting a third model? If so, could you please elaborate what it amounts to and what makes you think it might fit the evidence better than the two I have outlined.

Click to expand...

No this is not another model. You're thinking about the cause while a diffraction pattern is a consequence. In crystallography and such, folks are studying diffraction patterns to determine the underlying crystal structures.

Click to expand...

Yes, I am thinking about the cause. But if the diffraction pattern is instead a consequence, the question arises what it is a consequence of. I am not sure I understand that.

I did a very simple test with my 7-14 and gh2. I kept the lens aimed at my ceiling where the is a very bright spotlight. walking to the side I noted where the spot disappears from the evf. Then I take the 7-14 without body and a whit paper under it and see when the spot disappears from the paper. It was about the same spot with landscape (not portrait ofcourse) orientation. So my conclusion is the only bright lights outside the vertical view (when in landscape orientation) can cause in body reflections. The Image circle is hardly larger then the sensor AFAICT. So if you indeed can´t cause just out view issues in portrait orientation with the light source in the vertical plane, that would support in body reflections.


BTW I´m still curious what happens to the colour of the flare with systematic underexposure by lowering the shutter speed (in 1-6 or more stops).

_sem_ said:

_sem_ wrote:

Whether this is a diffraction pattern (like the red dot flare) or not. Notice there are other special types of flare, for instance the "sunstars" (reflections from aperture blades), reflections from lens element edges, reflections from lens surfaces sealed very close together (concentric circles) etc.

Click to expand...

I wonder about the different shapes of flare.

Round or aperture shaped flare, or just general loss of contrast, is relatively common.

On the other hand, I don't recall ever seeing rectangular flare other than the purple flare examples I've seen posted (including my example). Is rectangular flare more common than I've noticed?

--

Anders W said:

Anders W wrote:

But do we know that the red-dot pattern is a diffraction pattern? And if so, how do we arrive at that conclusion?

If I recall correctly, you said in a previous post, when I asked, that noone has really managed to provide a convincing explanation of this pattern yet.

Click to expand...

Well, watching the samples, and comparing to various other diffraction phenomena (I've linked some in my previous posts), I'm pretty much convinced. But mind this can't be pure diffraction (as laser-on-slit samples), because bouncing back from lens surfaces should be involved. I've seen a number of such patterns with different cameras and lenses. Mind some lenses tend to magnify blobs hugely, some produce a "normal-scale" pattern (like the RX100), while others don't magnify as much and only leave a smeared cross with hardly discernible dots.

Anders W said:

Anders W said:

Notice there are other special types of flare, for instance the "sunstars" (reflections from aperture blades), reflections from lens element edges, reflections from lens surfaces sealed very close together (concentric circles) etc.

Click to expand...

Yes, but I think (without being quite sure) that my most recent set of test shots covers all the types that are at issue as far as the purple-flaring of the 7-14 is concerned.

Click to expand...

When one observes flares there always tend to be a bunch of different flares at once, and it is difficult to explain why some are more intense than others. I'm just saying that your samples don't clearly show (or rule out) a diffraction pattern yet, although I think I "smell" it (but I may be wrong). Pls do check my cellphone video if you haven't already, the video says more than my explanations. And try recording a similar video with the 7-14 on the E-M5 (just the flares from a small intense sharp-edged reflector in a dark room, clowns are not required).

Anders W said:

Anders W said:

No this is not another model. You're thinking about the cause while a diffraction pattern is a consequence. In crystallography and such, folks are studying diffraction patterns to determine the underlying crystal structures.

Click to expand...

Yes, I am thinking about the cause. But if the diffraction pattern is instead a consequence, the question arises what it is a consequence of.

Click to expand...

Certainly. But I think it is probably very difficult to establish the cause without having full sensor structure layout and a good background in sensor design. This may be work for folks at Sony. On the other hand, it should be much easier to establish whether there is a diffraction pattern involved as a first step. Then, one could try establishing which lens surfaces are involved in the most offending reflections. Finally, the pattern combined with optics might reveal relevant dimensions at the source of diffraction, if one manages to show that certain elements of the sensor act as a diffraction grating http://en.wikipedia.org/wiki/Diffraction_grating . Though, I think diffraction will not explain where the purple (or red) comes from in the first place.

One more idea one could try is using a point-like light source to induce the flares, then covering the front element of the lens with a dark screen with a small hole as to isolate a narrow beam of rays and watching the impact on the flares.

richarddd said:

richarddd wrote:

richarddd said:

_sem_ wrote:

Whether this is a diffraction pattern (like the red dot flare) or not. Notice there are other special types of flare, for instance the "sunstars" (reflections from aperture blades), reflections from lens element edges, reflections from lens surfaces sealed very close together (concentric circles) etc.

Click to expand...

I wonder about the different shapes of flare.

Round or aperture shaped flare, or just general loss of contrast, is relatively common.

On the other hand, I don't recall ever seeing rectangular flare other than the purple flare examples I've seen posted (including my example). Is rectangular flare more common than I've noticed?

Click to expand...

That's usually called "ghosts", where a mirrored image of bright object appears elsewhere in the frame. Often appears with old uncoated lenses on new digital cameras with reflective sensors. Also commonly appears due to cheap filters used on top of the lens.

Notice the rays from the image plane are defocused at the aperture plane, then focused again on the sensor plane. But if you focus on a nearby object, spot-like highlights from a distance tend to convert into aperture-shaped bokeh-balls on the sensor plane...

secretworld said:

secretworld wrote:

I did a very simple test with my 7-14 and gh2. I kept the lens aimed at my ceiling where the is a very bright spotlight. walking to the side I noted where the spot disappears from the evf. Then I take the 7-14 without body and a whit paper under it and see when the spot disappears from the paper. It was about the same spot with landscape (not portrait ofcourse) orientation. So my conclusion is the only bright lights outside the vertical view (when in landscape orientation) can cause in body reflections. The Image circle is hardly larger then the sensor AFAICT.

Click to expand...

Yes it is. The diameter of the image circle must at least equal the sensor diagonal (21.6 mm for MFT) and needs in practice to be a bit larger than that to avoid excessive vignetting at the corners. Even if the image circle has a diameter of just 21.6 mm rather than slightly more, the area it covers is 1.63 times the sensor area (10.8^2 x pi versus 17.3 x 13). In reality, it may well cover twice the sensor area. On top of that, you have the possibility of flare that isn't really part of the image circle. The image circle is a matter of how the lens is meant to work. Flare isn't.

And yes, there will be a significant out-of-the-frame but inside-the-image-circle area, not just above and below the frame but also to the left and right of it, although the latter will of course be smaller.

That said, experimenting with how things work horizontally as opposed to vertically is not a bad idea.

secretworld said:

So if you indeed can´t cause just out view issues in portrait orientation with the light source in the vertical plane, that would support in body reflections.

BTW I´m still curious what happens to the colour of the flare with systematic underexposure by lowering the shutter speed (in 1-6 or more stops).

Click to expand...

I have seen no signs at all that this would make any difference. An exception would be when you come close to rock-bottom in any one of the three channels, in which case you are likely to see color shifts for reasons completely unrelated to flare.

_sem_ said:

_sem_ wrote:

Well, watching the samples, and comparing to various other diffraction phenomena (I've linked some in my previous posts), I'm pretty much convinced. But mind this can't be pure diffraction (as laser-on-slit samples), because bouncing back from lens surfaces should be involved. I've seen a number of such patterns with different cameras and lenses. Mind some lenses tend to magnify blobs hugely, some produce a "normal-scale" pattern (like the RX100), while others don't magnify as much and only leave a smeared cross with hardly discernible dots.

Click to expand...

One could also try inducing diffraction directly on a bare sensor with a laser beam, but I'm not sure who would volunteer for this experiment, hehe

Maybe an ordinary LED flashlight from a distance into the open lens mount would do (then observe reflections at different angles).

Most of these are accidents - however the 2 woodland ones were deliberate - the more flared one is at f4, the tighter one is at f8.

I'm still convinced that it's an interaction with the sensor as everything seems to be focussed towards the centre of the image from the light source - but I hope this help those looking at it!




This was an overcast day - just a general orb on her face in the centre of frame


as above but different shaped


This was taken a few minutes later as the sun was breaking through.... No Flare or orbs despite very bright clouds


Here there is an Orb for each lightsource between the light and the centre of the frame - even for the out of shot lights


As above


A Woodland scene at f4 (deliberately looking for it)


and at f8 - more orb and less flare - all between the light source and the centre of frame again




--
James

richarddd said:

_sem_ said:

Whether this is a diffraction pattern (like the red dot flare) or not. Notice there are other special types of flare, for instance the "sunstars" (reflections from aperture blades), reflections from lens element edges, reflections from lens surfaces sealed very close together (concentric circles) etc.

Click to expand...

I wonder about the different shapes of flare.

Round or aperture shaped flare, or just general loss of contrast, is relatively common.

On the other hand, I don't recall ever seeing rectangular flare other than the purple flare examples I've seen posted (including my example). Is rectangular flare more common than I've noticed?

Click to expand...

Richard,

As already indicated by _sem_, I am sure that what we see in your sample here

http://forums.dpreview.com/forums/post/50961227

as well as in Michael's sample here

http://forums.dpreview.com/forums/post/42396372

are mirror images of the light source (the three windows in your case). Essentially, these mirror image are of the same kind as those in the series of orbs extending from the light source towards the lower left corner in my test shots here

http://forums.dpreview.com/forums/post/50972762

The mirror images as such are impossible to prevent in shots like those shown by you and Michael where the light source is in the frame. They are a matter of lens design. But only some of those mirror images will ordinarily become easily visible, i.e., those that go purple on the E-M5 as opposed to some other cameras.

Now my most recent experiments here

http://forums.dpreview.com/forums/post/50966147

and here

http://forums.dpreview.com/forums/post/50972762

show that these mirror images go purple for reasons quite different from those I outlined in the previous thread here

http://forums.dpreview.com/forums/post/50958004

Moreover, the evidence I so far have suggest that we might be able to get rid of the purpleness of this particular kind of flare quite easily with the 7-14 if we could find a practical way of using filters with it.

In the experiments I reported in this thread, I showed that the purpleness went away completely with a red R25A filter. That filter is a very heavy treatment (filter factor 8). But today I tried with another, far lighter dosis in the form of yellow filter (Hoya HMC Y(K2), another "relic" from the B&W film days, good that I kept them after all), which takes away a rather small amount of light and the purpleness disappeared completely in this case as well.

So anything that blocks blue to at least some extent might work here and I encourage those with other potential filter candidates to try them and report the outcome. However, it appears that we have to attack wavelengths of real blue, not UV. I tried with a 1A (skylight) filter again today (I have tried UV and skylight filters before) just to confirm once more that this has no effect.