M-X ugly flare pattern

[Artur Orzel]

M-X and RPPs
Has anyone seen something like this.We shot many times the sun with mysterium sensor and we never saw such an effect.

http://www.haasta.pl/pub/mx/f.mov

 

[Cam Crowley]

I'm not a huge fan of some of the flaring on the wider RPPs but this looks like somethings not quite right.

At least the purple sun spots gone!

Wierd.

 

[Artur Orzel]

I think that it is not the lens! OLPF filter mabe??

 

[Cam Crowley]

Most likely not the lens by itself. Likely OLPF or sensor itself.

Have you tried to replicate it with any other lenses?

Also - I'm assuming you could see it in the monitor while shooting - was there any way of cutting the sun so the big dots disappeared?

 

[David Birdy]

I agree with Cam, It's not just a lens flare issue, the round dots have to do with sensor and or sensor filter.
You may want to try a test by dropping in an ND filter next time and closing the Iris to find the "sweet spot".

 

[Filip Orlandic]

http://reduser.net/forum/showthread.php?t=49246

 

[Cail Young]

You weren't by any chance using a hot mirror filter were you?

 

[Jeff Kilgroe]

I second Cail's question. This is an internal reflection issue and not an RPP issue, in fact it's not even a RED issue, but can afflict any camera that bounces internal reflections about off an OLPF or sensor structure. Just look at the sensor block on most any digital camera -- Ooooh, shiny.

I have encountered this myself from time to time on a few cameras, not just RED. You have light going the wrong way in your optical path, just look at the intensity of the reflection directly back through the iris. Nicely defined 7-blade bokeh reflecting out in front in that first image. Check that all your filters are facing the proper direction! Also make sure you're not getting light leakage between filter stages on your mattebox or elsewhere. That's more common that a lot of people realize.

 

[James T Mather]

Narcissus.

 

[Dan Hudgins]

Sun in frame?

Sun in frame?

The four dots pattern is what the (some) OLPF does, I wonder if a Pola filter could be adjusted to reduce that to a two dot pattern?

The OLPF divides the light into 4 beams so that the four pixel groups get some of the same light, that's how it reduces chroma moire. On the first pass those are just the next pixel over, so to make the larger pattern some reflections would seem to be going on?

The crystals in the OLPF may polarize the light somewhat.

The size would seem to mean the light is going through the filter maybe three times, maybe bouncing off the sensor cover glass at the point of focus of the sun image, then back off the lens or something in front of the lens, or the front surface of the OLPF where the IR filter is?

==

On a related issue, is if safe to point a Bayer filter sensor into the sun at all, of if you leave the camera static while the sun shines on the sensor can it fade or burn the dyes in the Bayer filter?

What if you have a T/1.3 lens on the camera and someone opens the iris and then the sun gets into the frame and shines on the sensor for a long time, like when the camera is off?

Since you can set paper on fire, I have been wondering how safe pointing the camera into the sun is on Bayer filter cameras?

==

The texture in "pattern.jpg" looks like another issue, caused by the far red light bouncing off the IR filter, it looks like a de-Bayer artifact, which means that "active" interpolation is being used, my Brother calls it the "Tetris Effect" because it resembles Tetris shapes in the image as a fine texture that is irregular. I have seen something like it in some of the tests I have done in the de-Bayer code I am working on now under specific conditions, here it seems to be do to the un-balance of the light's color in relation to the gain settings M-X is using, maybe.

http://en.wikipedia.org/wiki/Tetris

http://en.wikipedia.org/wiki/Tetromino

Active interpolation is used to pull "extra" detail out of the Bayer data, but can under some conditions become "unstable", it might be possable to have a global check for this pattern and to convert back to passive interpolation?