Dark Edges on Large CCD

[MarkF48]

If you haven't already done it, maybe contacting Prosilica support would be an option.....
Contact Support - Allied Vision Technologies

I'm wondering if the distance from the lens mounting flange to the plane of the sensor may be be having an adverse effect on the image circle.

 

[analog.universe]

This is really interesting, and it makes me think that the construction of the sensor limits the light captured based on its incidence angle. Someone else can confirm my knowledge of optical physics here, but I think I know what I'm saying. A theoretical lens of infinite focal length would direct light at the sensor precisely perpendicular to it. Any shorter focal length would produce a perpendicular ray in the exact center, with gradually increasing angle toward the edges of the sensor. It seems that your sensor has a particular incidence angle at which light begins to fall off. The wide angle lenses reach this threshold close to the center of the sensor, and the longer lenses progressively farther out. The 85 doesn't achieve this angle until beyond the edge of the frame.

So my suggestion is to investigate whether or not there is a filter in front of the sensor that is causing this and can be removed/replaced, or, if this a function of the construction... employ a corrective optic to allow the lens farther from the sensor plane, resulting in more parallel light striking the sensor.


afterthought: This explains the horizontal as opposed to circular pattern, as my guess is that whatever filter causes this light falloff is related to the R,G,B differentiation, and the pixels on that sensor are staggered horizontally and stacked vertically.. if you know what i mean.

 

[nicholasuniversal]

afterthought: This explains the horizontal as opposed to circular pattern, as my guess is that whatever filter causes this light falloff is related to the R,G,B differentiation, and the pixels on that sensor are staggered horizontally and stacked vertically.. if you know what i mean.

Thanks, I'm not sure what you mean.


I contacted Allied Vision and this was their response:

Yes the darkening your seeing is due to Vignetting. Vignetting - Wikipedia, the free encyclopedia

This is due to one of 2 issues. Either the off angle rays of the lens entering the “column” of the pixel and striking the sides of the pixel rather than the light sensitive portion at the bottom, or the lens is poorly corrected to handle those type of rays (they get worse as the aperture gets larger, so at F2 you will notice it more than at F5.6). If you adjust the exposure and aperture to match, and you don’t see a difference in the vignetting, it’s a lens issue that can be corrected by 2 options, lens choice or software.

There is a filter that can be used to help correct this issue optically, called a center spot filter, but they are VERY expensive and do not help much as they are not optically matched to the specific lens.

 

[analog.universe]

Thanks, I'm not sure what you mean.

the off angle rays of the lens entering the “column” of the pixel and striking the sides of the pixel rather than the light sensitive portion at the bottom

^ that's what I mean. Since the sensor has to capture each color separately, the pixels for each color are oblong, in your case narrower in the horizontal dimension. When there are 3 next to each other you get a square pixel made up of all the colors. So if you're hitting the pixel from the narrow side, you'll reach the opposite wall (as opposed to the bottom of the cell) sooner than you would from the long side.

The tilt shift lens worked because it had a larger image circle and was able to get to the edge of the sensor with less incidence angle. I bet you could acquire an old medium format lens and adapter on ebay to obtain a large image circle. I read some stuff a while back on DIY tilt/shift lenses, and Pentacon 6 lenses were really popular and fairly affordable...