A ghost in a IQ180 frame

[Christopher]

I just looked through some images and found this:

It isn't easy to see, but some type of ghos of the building shows up. What could be the cause ? It was IQ180, 80D.

 

[Shelby Lewis]

Was the building in that position in previous frames? Looks like it's "burnt in" from a previously taken frame... but I'm just guessing. I guess it could be an internal reflection as well (even though the physics don't make sense in my mind).

Just guessing!

 

[jlm]

the slight tilt shows it to be a mirror image, not just a repeated image. wacky

 

[Bob]

I bet this image was boosted maybe by a couple of stops. Was it?
if so it may very well be a reflection off the sensor and back to the rear element and back again.
-bob

 

[jagsiva]

Christopher,

I had a similar issue with my DF and Aptus 12. It turned out to be some kind of timing issue with shutter/dark frame mumbo jumbo. They ended up replacing my DF. Make sure you get in touch with your dealer.

I know Leaf is quite aware of the issue. They asked me for several RAW files and I know they went through quite a laborious process in trouble shooting. Again, the dealer is likely your best bet.

 

[Christopher]

The shot was boosted around 0.7. So far it has been the only image it happend. I have to check again.

 

[ondebanks]

I bet this image was boosted maybe by a couple of stops. Was it?
if so it may very well be a reflection off the sensor and back to the rear element and back again.
-bob

If it was a reflection back off the rear element, the image would not be in focus when the light hit the sensor a second time. But it is.


This ghost, mirror-image looks like the electronic cross-talk that occurs in certain astronomical cameras I've worked with, where multiple sensors or quadrants are read out in parallel. The IQ180 has dual readout systems (one does the left half of the image, one does the right) so electronic cross-talk is the chief suspect in my book.

If it can be properly calibrated, then it can be removed by a software routine applied to the raw pixel data afterwards - this is what we do with the astronomy research cameras.

My worry is that this might be a sleeping problem with ALL IQ180s out there, and that they might need to be individually calibrated to determine the extent of the correction required.

It's up to PhaseOne to investigate and solve this.

Ray

 

[6x7]

just as a side note but to me this raises the question about what is the room that we are generously giving the engineers to interpret what we are photographing…

peter

 

[Shashin]

just as a side note but to me this raises the question about what is the room that we are generously giving the engineers to interpret what we are photographing…

peter

Chemical engineers or electrical engineeers, nothing has really changed...

 

[DDudenbostel]

I had a similar phenomenon with my Hasselblad CFV39 back. I was doing products in the studio with the back mounted on my Technikatdan and using tungsten lighting. My exposures were around 20-50 seconds and rather than use the shutter and possibly introduce vibration I decided the use a black card to cap the lens like a shutter and wake the back up with an electrical release. Doing things this way I didn't close the lens shutter between exposures because the back was asleep. To my surprise even though the back wasn't active ( awake technically ) the sensor still had power applied and was still capturing an image. My first shot looked great bu the second exposure had a different exposure even though time and aperture were the same as the first. The third frame I moved one item in the scene and on the next exposure I saw a distinct image ghosting where the item was that I moved. I did a few tests to see what was going on and discovered the sensor was collecting data even though it technically was asleep. Closing the shutter between exposures eliminated the problem.

 

[cunim]

Ray may be right, but as JLM points out it is a mirror image. Wouldn't cross talk replicate the original? I used to see this a lot with cooled cameras and it was usually (not always) reflections. Easy to test with a light bulb and a dark night. Using differrent lenses, position the bulb at the same point on the chip. If the ghost moves with lens changes it is reflections. If not, xtlk.

 

[gerald.d]

Ray may be right, but as JLM points out it is a mirror image. Wouldn't cross talk replicate the original?

You'd get a mirror image if the left and right halves of the sensor were read from the centre outwards.

 

[ondebanks]

Ray may be right, but as JLM points out it is a mirror image. Wouldn't cross talk replicate the original? I used to see this a lot with cooled cameras and it was usually (not always) reflections. Easy to test with a light bulb and a dark night. Using differrent lenses, position the bulb at the same point on the chip. If the ghost moves with lens changes it is reflections. If not, xtlk.

Here is an example of one of those cameras I've worked with - NICS on the TNG. TNG :: Telescope & Instruments :: NICS and click on the orange gif image.
Notice how the cross-talk is mirrored in position: a - A - a and so on.

Ray

 

[MGrayson]

If it were a reflected image, it would NOT be reversed. If it reflected off of a curved surface, it might come back 180 degrees rotated, but only reflection off of an odd number of surfaces would reverse the image (the sensor itself counts as one reflection, the back of the lens counts as the second).

Here's a Zen Koan for you. When we look in a mirror, why is the image reversed left-right and not up-down?

Best,

Matt

 

[ondebanks]

Here's a Zen Koan for you. When we look in a mirror, why is the image reversed left-right and not up-down?

Because up and down are absolute directions, while left and right are in the eye of the beholder. If you and I are facing each other, we have the same up and down but our left and right are opposites.

Ray

 

[MGrayson]

Because up and down are absolute directions, while left and right are in the eye of the beholder. If you and I are facing each other, we have the same up and down but our left and right are opposites.

Ray

:lecture: (Here I go again!)

Let's try an experiment. You have a book with each page printed with a different orientation. On some pages, the text is reversed, on some it's upside down, on some it's normal. You open the book at random and, without looking at the page, hold the book up in front of a mirror and look at its reflection. You see something. Say it looks normal. Now.

How do you discover what the real page looks like?

This is the central point. You can't see the open page unless you Turn The Book Around. (Or turn yourself around to look at the page)

How do you turn the book around? If you rotate it around a vertical axis, you'll see the text reversed left-right, but you can turn a book around by rotating it around a horizontal axis, and then the text will look upside down, or you can rotate it around a diagonal axis, and the text will look reflected and rotated. But you can't turn the book around without choosing an axis of rotation!

Well, you say, suppose I could make the book transparent and look through it. Nope, that will look just like the image in the mirror! It's not the looking in the mirror that reverses the image. It's only in the comparison with the original. That comparison necessitates choosing an axis. That axis is arbitrary. We just always pick a vertical axis when we think about it.

Another point of view - what does a starfish see in a mirror? Another starfish. Suppose the first starfish wiggles its arms one after another in a clockwise pattern. The mirror starfish will wiggle its arms in a counterclockwise direction. There is no right-left or up-down reversal, just chirality.

/:lecture:

Matt

 

[jlm]

reminds me of the dextro- vs leuvo stereo isomers; we seem to only have one hand in nature

 

[cunim]

This thread is becoming really interesting.

Ray, here's an example of reflection acting as diffuse flux source over the field of view. In this case it's a cryogenic camera, binned and running at photon counting levels (16 bits) but using a single integration instead of amplification/scintillation counting. Turns out you can photon count with integrating cameras if the optical efficiency of the entire system is above a certain threshold. For example, we ran the CCD at -110'C, the lens had an f of about 0.9, and it's a TBI CCD with about 70% QE at the emission wavelength.

Anyway, note that the bright targets illuminate the areas around them, even though they are contained in black wells. Further, the targets sum to illuminate the entire black well plate. This occurs because light reflects from the rear element of the CCD (very close to the targets). The reflection is sufficient to set a bottom limit on sensitivity.

These are rather special conditions, and you are probably correct that the IQ is showing xtlk. I think the second picture supports your view. Here I just snapped a monitor in a dim room, IQ180, LS150, f4, 2 sec, iso800. Note the repeat of the saturated monitor light at the left - and the black stripy junk. I have seen the black stripes before, but the xtlk not so much because I rarely boost exposure. Here I boosted exposure .7, ran up shadow to 25 and bumped the contrast. Spot is only a half dozen gray levels and difficult to see otherwise.

Matt, my head is buzzing.

 

[ondebanks]

:lecture: (Here I go again!)

Let's try an experiment. You have a book with each page printed with a different orientation. On some pages, the text is reversed, on some it's upside down, on some it's normal. You open the book at random and, without looking at the page, hold the book up in front of a mirror and look at its reflection. You see something. Say it looks normal. Now.

How do you discover what the real page looks like?

This is the central point. You can't see the open page unless you Turn The Book Around. (Or turn yourself around to look at the page)

How do you turn the book around? If you rotate it around a vertical axis, you'll see the text reversed left-right, but you can turn a book around by rotating it around a horizontal axis, and then the text will look upside down, or you can rotate it around a diagonal axis, and the text will look reflected and rotated. But you can't turn the book around without choosing an axis of rotation!

Well, you say, suppose I could make the book transparent and look through it. Nope, that will look just like the image in the mirror! It's not the looking in the mirror that reverses the image. It's only in the comparison with the original. That comparison necessitates choosing an axis. That axis is arbitrary. We just always pick a vertical axis when we think about it.

Another point of view - what does a starfish see in a mirror? Another starfish. Suppose the first starfish wiggles its arms one after another in a clockwise pattern. The mirror starfish will wiggle its arms in a counterclockwise direction. There is no right-left or up-down reversal, just chirality.

/:lecture:

Matt

Matt, that's a great technical treatment. :clap:

I also rather like my explanation, which had the benefit of conciseness!

If instead of referring to left and right, I use absolute directions (e.g. NSEW), then when I raise my "western" hand, my mirror image also raises its western hand. When every point maps perpendicularly to another plane facing it, all relative orientations are preserved.

Ray

 

[ondebanks]

This thread is becoming really interesting.

Ray, here's an example of reflection acting as diffuse flux source over the field of view. In this case it's a cryogenic camera, binned and running at photon counting levels (16 bits) but using a single integration instead of amplification/scintillation counting. Turns out you can photon count with integrating cameras if the optical efficiency of the entire system is above a certain threshold. For example, we ran the CCD at -110'C, the lens had an f of about 0.9, and it's a TBI CCD with about 70% QE at the emission wavelength.

Anyway, note that the bright targets illuminate the areas around them, even though they are contained in black wells. Further, the targets sum to illuminate the entire black well plate. This occurs because light reflects from the rear element of the CCD (very close to the targets). The reflection is sufficient to set a bottom limit on sensitivity.

These are rather special conditions, and you are probably correct that the IQ is showing xtlk. I think the second picture supports your view. Here I just snapped a monitor in a dim room, IQ180, LS150, f4, 2 sec, iso800. Note the repeat of the saturated monitor light at the left - and the black stripy junk. I have seen the black stripes before, but the xtlk not so much because I rarely boost exposure. Here I boosted exposure .7, ran up shadow to 25 and bumped the contrast. Spot is only a half dozen gray levels and difficult to see otherwise.

Matt, my head is buzzing.

Yeah, it certainly looks like the spot is the same distance in from both the left and right edges of your IQ180, and the same vertical distance up the sensor (same row numbers). This is the same symmetry as in Christopher's image which started this thread. The evidence for crosstalk is mounting.

Ray