[wim]

[quote name='popo' timestamp='1283615357' post='2482']

I just played about with some calculations. Of course there are a lot of unknown variables.



If we assume the mirror is 45 degrees from vertical with the silvered surface on the outside, and it has a refractive index of 1.5, then an unwanted reflected image would be formed at a spacings multiples of about 1.07x the thickness of the mirror.



For the mirrored surface reflecting at 30% and the glass surface at 1%, each additional reflected image is about 7.2 stops down from the previous. If you assume the coating is even better at 0.1% reflection then the figure becomes 10.5 stops. With such huge losses, the reflection of the reflection is going to be so far down it is unlikely to be significant short of pointing the camera at a very bright point source.



So looking back at the sample image earlier, or more precisely the original of it, the spacing of the secondary image is about 12 pixels offset from the original, working out to be about 0.06mm. That's pretty small! Even if they used a high refractive index material and a more vertical angle the mirror would still have to be pretty thin, a fraction of a mm.



I can refine the above if anyone has more precise values for any of the variables.

[/quote]

Thank you, popo. This is exactly what I worked out schematically on a piece fo paper, without sorting through an assortment of calculations.



I think you can assume that coatings would make transmissiveness 99.7 % or better, with the current state of coatings. Furthermore, in order to make the chance of double images as small as possible, I would expect the mirror coating to be at the back of the mirror, IOW, on the surface closest to the sensor for optimal image quality. This would reduce the chance of unwanted ghosting at least by another factor of 4, or 2 stops.



Of course, the neat thing of having the mirror coating on the sensor side, is that any light reflected by the sensor itself, disappears mostly in the black void of the bottom of the mirror box, before anything can be bounced back from the other side of the mirror.



I reckon this ghosting is not caused by the mirror, but by something else altogether. It is way too bright for starters.



Oh, on the reflection off the back element of a lens (this is really a reply to another post <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/biggrin.gif' class='bbc_emoticon' alt='' />): many digital lenses have flat or almost flat surfaces at the sensor end of the lens, IOW, it is possible that ghost images are caused by reflections of the last element surface of a lens, especially if they are only slightly bigger or smaller than the original image.



Kind regards, Wim

[wim]

Just another thing:



The photograph with the ghosting was a shot with the 50 F/1.4. There are similar shots in th egallery with high contrasts, at low EVs, taken with other lenses. As far as I can see, there are no such reflections in those images. So, until we see more proof of this to be caused by the mirror, it could also well be caused by internal reflections in this lens, or by a filter mounted to that lens.



Also, the DR of the A55 is 9 EV apparently. This would indicate, based on the calculations above, that ghosting caused by reflections would be at a level non-perceivable by the sensor (unless this is a black dot type problem, caused by the processing of the image).



I would really like to know now whether the back of the mirror has the mirror coating, or the front. Anybody knows for sure?



Kind regards, Wim

[Brightcolours]

[quote name='wim' timestamp='1283618344' post='2484']

Just another thing:



The photograph with the ghosting was a shot with the 50 F/1.4. There are similar shots in th egallery with high contrasts, at low EVs, taken with other lenses. As far as I can see, there are no such reflections in those images. So, until we see more proof of this to be caused by the mirror, it could also well be caused by internal reflections in this lens, or by a filter mounted to that lens.



Also, the DR of the A55 is 9 EV apparently. This would indicate, based on the calculations above, that ghosting caused by reflections would be at a level non-perceivable by the sensor (unless this is a black dot type problem, caused by the processing of the image).



I would really like to know now whether the back of the mirror has the mirror coating, or the front. Anybody knows for sure?



Kind regards, Wim

[/quote]

There is no other image "similar" in the dpreview galleries. None is in low light with light sources.



There are MANY (really MANY) images which show the ghosting problem, though.



Yes, the mirror surface is on the front surface (and as far as I know, that is always the case with plate beam splitters). If it was on the back, diffraction would mess things up, I guess.



DP review does not measure DR, even if they say they do. They use in-camera tone curves... and used JPEG output, while also noticing the detail eating strong NR.

A ghost image will not knock

Yes, it is caused by the mirror. No, it is not reasonable to blame it on something else.

Comparing it the the black dot occurrence from the 5D mk II just shows little understanding of that problem, and how totally unrelated it is to the ghosting issue.



Just in:

http://www.imaging-resource.com/PRODS/AA55/AA55A4.HTM

Sony A55V with 18-55 kit lens and flash usage:





Sony A560 with 18-55 kit lens and flash usage:









"When additional surfaces are added to an optical path, especially ones not parallel with the image plane, there is always a chance that light will travel an unintended path. In the case of the A55V and A33, it appears that the rear surface of the translucent mirror film can reflect light back to the front surface of the film, which reflects it back to the rear, causing a "ghost" reflection or multiple reflections to appear in the final image under certain conditions. The above left crop shows this phenomenon in one of our Sony A55V flash test shots, where a very strong reflection from the plastic edging of our flash-range/uniformity target also has a small ghost image in the form of a horizontal white line below it. (Thanks to IR reader Erick E for pointing this out in one of our flash range images!) The crop on the right is from a similar flash shot, taken with the same lens, but using the Sony A560 which is a traditional SLR without a translucent mirror in the optical path. As you can see, under nearly identical conditions (the A560's flash is a bit stronger), no ghosting is present."



" Based on our tests, we can confirm that this is a genuine phenomena with the Sony A33/55. Should you be concerned about it? Perhaps. Some people would certainly find it objectionable, especially if they did a lot of night photography of things like cityscapes. Personally, it wouldn't deter us from buying a Sony A55V, because we don't do much cityscape-type photography: Our night shots tend to be ones where the subjects are people or areas lit by nearby light sources. With these sorts of shots, ghosting of the type shown above wouldn't be an issue. Bottom line, you'll need to decide for yourself if this would impact your personal style of shooting enough to outweigh the benefits of the Sony A55's pellicle mirror design."



Images from A55V and A33 users:

[popo]

Thinking more, I'm not sure if which side of the mirror is mirrored matters, if it is the cause of the flare. The unwanted light path has the same number and types of reflections so the result would be the same.



The sensor dynamic range doesn't matter I think. If the scene is higher dynamic range than the sensor, the highlights would be clipped and seem closer in level to the flare.

[Brightcolours]

[quote name='popo' timestamp='1283620978' post='2486']

Thinking more, I'm not sure if which side of the mirror is mirrored matters, if it is the cause of the flare. The unwanted light path has the same number and types of reflections so the result would be the same.



The sensor dynamic range doesn't matter I think. If the scene is higher dynamic range than the sensor, the highlights would be clipped and seem closer in level to the flare.

[/quote]

Golly, you are right. Both in case of the 1st surface being 30% reflective and the 2nd surface being 30% reflective, the same amount of ghost light will exit the mirror. In both cases the light (and ghost light) are only "filtered" (refected 30% up) once.

[wim]

[quote name='Brightcolours' timestamp='1283621528' post='2487']

Golly, you are right. Both in case of the 1st surface being 30% reflective and the 2nd surface being 30% reflective, the same amount of ghost light will exit the mirror. In both cases the light (and ghost light) are only "filtered" (refected 30% up) once.

[/quote]

Actually, it does matter in the case of light reflected of the sensor back to the mirror, because most of that light would not take part in any reflections and hence possible ghosting. Also, the reflections of the non-mirror coated surface are always going to be a lot less in intensity than anything reflected by the mirror, as the transmission intensity from an optical POV is always going to be stronger unless you get very close to the angle where complete reflection takes place. However, in that case the angle is going to be so steep, that there is no way it can still get back to the other side of the mirror, when reflecting again of the other side of the glass/air transition, so to speak. After all, angle of incidence = angle of exit, in the case of reflection. On the second surface the angle of incidence is determined by the angle of the incident light on the first surface and the refractive index of the glass.



Kind regards, Wim

[wim]

[quote name='Brightcolours' timestamp='1283620664' post='2485']

There is no other image "similar" in the dpreview galleries. None is in low light with light sources.



There are MANY (really MANY) images which show the ghosting problem, though.



Yes, the mirror surface is on the front surface (and as far as I know, that is always the case with plate beam splitters). If it was on the back, diffraction would mess things up, I guess.



DP review does not measure DR, even if they say they do. They use in-camera tone curves... and used JPEG output, while also noticing the detail eating strong NR.

A ghost image will not knock

Yes, it is caused by the mirror. No, it is not reasonable to blame it on something else.

Comparing it the the black dot occurrence from the 5D mk II just shows little understanding of that problem, and how totally unrelated it is to the ghosting issue.



Just in:

[url="http://www.imaging-resource.com/PRODS/AA55/AA55A4.HTM"]http://www.imaging-r...AA55/AA55A4.HTM[/url]

Sony A55V with 18-55 kit lens and flash usage:





Sony A560 with 18-55 kit lens and flash usage:









"When additional surfaces are added to an optical path, especially ones not parallel with the image plane, there is always a chance that light will travel an unintended path. In the case of the A55V and A33, it appears that the rear surface of the translucent mirror film can reflect light back to the front surface of the film, which reflects it back to the rear, causing a "ghost" reflection or multiple reflections to appear in the final image under certain conditions. The above left crop shows this phenomenon in one of our Sony A55V flash test shots, where a very strong reflection from the plastic edging of our flash-range/uniformity target also has a small ghost image in the form of a horizontal white line below it. (Thanks to IR reader Erick E for pointing this out in one of our flash range images!) The crop on the right is from a similar flash shot, taken with the same lens, but using the Sony A560 which is a traditional SLR without a translucent mirror in the optical path. As you can see, under nearly identical conditions (the A560's flash is a bit stronger), no ghosting is present."



" Based on our tests, we can confirm that this is a genuine phenomena with the Sony A33/55. Should you be concerned about it? Perhaps. Some people would certainly find it objectionable, especially if they did a lot of night photography of things like cityscapes. Personally, it wouldn't deter us from buying a Sony A55V, because we don't do much cityscape-type photography: Our night shots tend to be ones where the subjects are people or areas lit by nearby light sources. With these sorts of shots, ghosting of the type shown above wouldn't be an issue. Bottom line, you'll need to decide for yourself if this would impact your personal style of shooting enough to outweigh the benefits of the Sony A55's pellicle mirror design."



Images from A55V and A33 users:















[/quote]

Ok, thanks for this, I hadn't seen the other images yet. I wasn't aware of the non-pellicle comparison either. So, thank you for expanding on this and essentially taking the time to educate me more on this effect with the new Sony cameras. I hadn't expected the engineers not to check for this type of occurrence.



Anyway, this proves you were right and I was wrong, and I gracefully concede.



Kind regards, Wim

[Yakim]

[quote name='wim' timestamp='1283637157' post='2492']

I hadn't expected the engineers not to check for this type of occurrence.[/quote]



I have a good friend which is a mechanical engineer and from many talks with him I can guess that the engineers are not the ones to blame. It is most likely the short-sighted managers.

[wim]

[quote name='Yakim' timestamp='1283777370' post='2527']

I have a good friend which is a mechanical engineer and from many talks with him I can guess that the engineers are not the ones to blame. It is most likely the short-sighted managers.

[/quote]

That's very well possible indeed, based on my own experiences <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/biggrin.gif' class='bbc_emoticon' alt='' />.



Kind regards, Wim

[Brightcolours]

Early users are starting to notice that the camera gets warm with extended use (2 hours, A33) and starts to give a yellow warming light.



Nikon and Canon APS-C DSLRs also get warm and give warnings after prolonged live view and video use (the big APS-C sensors do generate a lot more heat than the smaller sensors used in compact digitals), so maybe those reports should not be very surprising?