PDA

View Full Version : Rainbow striping on image



Mark Deeble
10-02-2014, 09:00 PM
Can anyone suggest what might be causing the rainbow striping in this image please? The striping is visible on a number of shots - the more the backlight, the more intense the effect. Sidelight in the same conditions had no visible effect. The camera had been black-shaded before shooting.
Shot backlit in dust-storm Epic Dragon 6K, Optex-Nikon 50-300, 30fps, 180 deg, RC 8:1, Circ POL

https://www.flickr.com/photos/127581889@N04/15424768822/

Edit: R3D https://www.dropbox.com/s/6vy6x29pdxc4x39/A018_C051_1002G1.0000000F.R3D?dl=0

Rolf Kestermann
10-02-2014, 09:46 PM
My first guess would be the Circular Polarizer (hopefully).

Joan Poggio
10-02-2014, 10:05 PM
Hi Mark, quick introduction. I'm Joan, a Colombian filmmaker, based in Kenya over the past nine years. I'm very good friends of Simon Trevor (help him out with all his technical issues when possible), but my work really focuses around documentary centred around human issues. I was at the premiere of The Queen of Trees here in Nairobi and was blown away by the beauty of your film (even bought a copy!). I would love to meet you one day, perhaps one day you are transiting in Nairobi, or when I'm in Voi with Simon (actually was shooting a docu on poaching with Robin Dimbeldy couple weeks ago and we saw you Wanted posters).
Right, now straight to the point. I have been a RED Scarelt and now Epic owner for the past three years. I can't really see from the thumbnail the problem you are experiencing as it is very small and it's a jpg, a r3d 1 frame file would be much better. However, I have a feeling that what you are experiencing is CMOS banding or Gate flaring which has been discussed a lot in the DRAGON forum. Are you on a Nikon mount? There has also been some discussion regarding gate flaring with Nikon mounts... Also, are you on OLPF v1 or v2? Also, you mention the camera had been blackshaded, but, have you upgraded the firmware afterwards? With DRAGON it is of paramount importance to blackshade after every and any firmware upgrade. Just as well, blackshading is so much more critical, that it has been stressed repeatedly on the forum that the camera has to reach it's optimal temperature (some people suggest turning the camera on and rolling on a magazine for 10 minutes to allow the camera to warm up evenly and quickly).
(I'm not finding the CMOS thread, will keep on looking)
http://www.reduser.net/forum/showthread.php?119888-Gate-shadow-On-OLPF1-amp-2
http://www.reduser.net/forum/showthread.php?120485-Flare-OLPF-Port-Bushing-for-Dragon
Hope some of this helps!
joan

Joan Poggio
10-02-2014, 10:11 PM
Found a couple of posts related to the CMOS smearing.
http://www.reduser.net/forum/showthread.php?121045-Horizontal-Magenta-CMOS-Lines-Continued/page3
http://www.reduser.net/forum/showthread.php?121743-Dragon-Highlights-Bizzare-issues-Horizontal-smearing&highlight=cmos+dragon+flaring
Also, don't know if you are aware that OLPF system has been updated now and there is a new interchangable system. Some improvements have been made on the old OLPF (previously called v1).
Yup, circular polarizer could play something, but by the description of backlight shots, sounds more like cmos smearing or gate flaring which have been discussed a lot on the forum.
cheers
joan

Mark Deeble
10-02-2014, 10:32 PM
Thanks Joan! - would be great to meet - will PM contact details.
A bit more info on the shots - my Optex-Nikon 50-300 is PL mount, firmware is latest release, I black-shaded at various shutter speeds, 1/50 - 1/300 with camera at operating temp (15 min up to 65 deg). OLPF is v2 ( skin-tone, fixed). I'll try and post a R3D later but am in the bush with a very poor connection :)
I don't think it is gate flare ( I used to see that with the MX, prior to the OLPF bush upgrade). I hope it is the circa. POL but I have never experienced it before with that combo.
The effect is 3 diffuse vertical stripes of colours ( rainbow spectrum - but magenta and green the most obvious) over the entire image.

Gunleik Groven
10-02-2014, 10:55 PM
Hi. Can't say I see the CMOS or blackshade related symptoms in this image.
My guess would be an optical source.

Send a file to red and ask, and if you could post a single frame R3D, you could probably get a bit more help.

Best
G

Mark Deeble
10-02-2014, 11:25 PM
Thanks Gunleik.

Here's a link to an R3D

https://www.dropbox.com/s/6vy6x29pdxc4x39/A018_C051_1002G1.0000000F.R3D?dl=0

Gunleik Groven
10-02-2014, 11:26 PM
Smart. I am on the run/work now, and cannot loo at it before tonight heretime, but I am sure someone else will step up!

Joan Poggio
10-03-2014, 12:54 AM
Now that I've seen the image properly and download the r3d file, yes, does not seem cmos related, does look like optical. You mentioned you have noticed this in this type of light conditions. Have you had similar results without the circ pol or have you always had it on when this has presented itself?

Mark Deeble
10-03-2014, 01:01 AM
I've never seen it before, and I've used that circular POL a lot - only recently with the Dragon sensor, but for 2+ years with the MX.

Joan Poggio
10-03-2014, 01:09 AM
Ok, found another post with the same problem

http://www.reduser.net/forum/showthread.php?115216-WHAT-IS-THIS&highlight=upside+olpf

Read the last two pages, Red got in touch and he had to send the camera in... You should send a log of the camera and the r3d clip to support

Mark Deeble
10-03-2014, 01:23 AM
Thanks Joan - I have filed a support ticket. I have gone back through recent material and also noticed it on other shots from previous days (not backlit - and just high natural light).

Owen Rennie
10-03-2014, 01:41 AM
Looks kinda cool.. :)

David Battistella
10-03-2014, 02:14 AM
That looks optical, like stacking ND with circular POLA. because it is a new sensor and the IR tendencies are different you might need to go with another brand of circular POLA or variable ND. The flare probably has something to do with it as well.

Or it could be the circular pola, lens, combo (lens coating, pola, different OLPF) could contribute to these kinds of effects.

What PL glass was it?

Battistella

Robert Hart
10-03-2014, 05:19 AM
Mark.

Just a way out suggestion for a long-shot try. I am assuming an optical not electronic cause and that you may be well and truly distant from product support. If you are using mattebox filters 4x4 or larger, maybe try painting the frosted edges of the filter glasses with black felt tip marker and round the shiny face edges where they slip into the mattebox slides just in case something is coming off the smooth surfaces of the mattebox slide frames themselves and illuminating the frosted edge surfaces.

Maybe mask off the filter mattbox port to just outside of your widest lens view with a blackened paper or cereal-packet cardboard insert, maybe one for each lens field-of-view if the longer lenses are more prone to show the artifact.

Also maybe cut a piece of softly folded blackened paper with a sensor port cut in it and lay this in the space in the camera body behind the mount so that the horizontal "steps" in front of the sensor are covered over by a smooth taper.

Maybe cut a blackened rectangular "frame" exit pupil aperture to go on the rear of each lens like some lenses have fitted to mask some internal reflections.

These may all be red herring suggestions. If the colourations are vertically banded, then it may be a sensor issue. If they are horizontally banded then it might be an optical fault.


Does the artifact move around the image if you dutch or roll the camera whilst remaining on subject or only change in intensity whilst the image moves?

jamesbridges
10-03-2014, 06:54 AM
It appears to match where to sun is hitting on the ground…

David Battistella
10-03-2014, 07:13 AM
I'm also wondering if it is some internal reflections between the lens and the filter. I had some of those internal reflections going on yesterday on some stuff, but it was related to filter distance from the front element. Even with minimal space, at certain angles I was getting reflects. But none of it was sensor related.

If you loos the filter, do you see it?

Battistella

Mark Deeble
10-04-2014, 01:26 AM
Thanks guys - I am away from the camera but will test when I get back, if I can recreate conditions where I see it again. Lens was an Optex converted Nikon 50-300 f 4.5 to universal mount ( mounted PL in this case) which I use for a lot of wildlife work. Polariser was a Hasselblad 95mm circa polariser. I should have tested more at the time, but I had waited days for the action and it lasted less than a minute - so was rushing around trying to cover the action rather than analyse the problem. I'll try different filters, windowing, camera orientations etc early next week and report back.
(It may well have an internal reflection component as I had to change out the lens shade I normally use as it was vignetting at 6K coverage)

David Battistella
10-04-2014, 01:34 AM
Thanks Mark,

looking forward to to hearing more.

Battistella

Robert Hart
10-04-2014, 02:08 AM
I am torn between the notion that there is something electronic going on versus something related to the polariser and the sensor.

This is a useless comment but here it goes anyway as it will only take 15 seconds or so to try the suggestion.

During your examination of the camera, perhaps remove the lens to look at the sensor itself, through the polariser filter itself held up to your eye and rotating. You might need to view from different angles to side of optical axis and maybe move the light source behind you relative to the camera by swivelling the camera a little.

If there is an apparent faint banding pattern reproduced on the surface of the sensor, perhaps remove the special OLPF if you can safely do so without introducing dust, or swap with another different OPLF to see if the pattern is reproduced.

OPLFs are made specifically for a pixel pitch but it may just be that the pixel count for a given dimension is slightly different to that which the OLPF is tuned for. This is real witchdoctory on my part which has no basis in science whatever so please do not pay it any but fleeting notice.

If the issue is not electronic in origin, then there might be potential for features pixel pitch related, to interact to produce a faint repetitive pattern.


The actual pixel count per measured unit across the sensor.

The pitch of the printed bayer pattern of coloured masks being not quite coincident with the pixel pitch creating a moire not cancelled out entirely by the sensor calibration process.

The OLPF frequency cut off point.


The trouble with dangerous armchair theorists like me is that dead-ends and red herrings most often abound to create distraction. However if this is an issue which is not quite solved then it might stimulate another direction of thought by tech people who really know their stuff.


Who knows? When this program goes to air, please let us know here and via Facebook. Attenborough must surely wish that this tech was around when he did his best work or that he should enjoy agile health for another few years to work with the new gear.

Mark Deeble
10-04-2014, 06:16 AM
Thanks Bob - I'll try that. I can't remove the OLPF though ( its v2 and fixed and has warranty-void stickers on it - I am not sure if the replaceable system is out there yet). Will definitely let you know when we have a finished film. We've been shooting full time for 2.5 years now and the end is in sight!

Dave Blackham
10-04-2014, 06:26 AM
Mark, Is this effect seen with other lenses ?

Dave

Bob Gundu
10-04-2014, 07:00 AM
I have not witnessed this on my Dragon, but on my friends it happened regardless of lens. We couldn't figure out what or when it would happen.

Mark Deeble
10-04-2014, 08:32 AM
Dave - I've only had the Dragon in the field for a little over a week and most of the time I've been using that lens - I'll check next week

Mark Deeble
10-04-2014, 08:34 AM
I have not witnessed this on my Dragon, but on my friends it happened regardless of lens. We couldn't figure out what or when it would happen.

Thanks Bob - did the issue ever get resolved?

Mark Deeble
10-06-2014, 05:00 AM
Update:

I've had a brief opportunity to test further and this is what I am finding


1. It only seems to happen with my Optex/Nikon 50-300 PL conv.( I've not seen it with any other PL mounted lens I have : Century 2000 150-600, Tokina 11-16 or Zeiss superspeeds). I plan to test other Nikon lenses.
2. It seems to happen with whichever circular polariser I use on the lens - Hasselblad, Formatt, B+W.
3. I've used that lens / polariser combination a lot before with the MX sensor and never seen it.


So my thinking is that it seems to be something to do with that particular lens ( or its universal mount) and polarised light, interacting with the v2 OLPF and / or the photosite pitch of the Dragon sensor.

David Battistella
10-06-2014, 06:31 AM
Does it show up on that particular lens without the POLA?

Battistella

Mark Deeble
10-06-2014, 09:21 AM
No, just with the pola.

Joan Poggio
10-07-2014, 01:57 AM
Great to know you've pinned it down to lens/pola combo. I am upgrading to Dragon in couple months and always keen to see what everyone's experience is in order to troubleshoot things quickly if something odd shows up. Keep us posted if you find any other variables!

David Battistella
10-07-2014, 05:28 AM
No, just with the pola.


Must be the coating/pola combo if that pola is good on other glass.

Mark Deeble
10-09-2014, 11:16 PM
So - it continues...

Although I tried to recreate the conditions with other PL mount lenses and circ. polarisers in camp and couldn't, I took the camera out yesterday in exactly the same conditions and the diffuse vertical striping reappeared.
This time it was 6K, v2 OLPF, very bright ( think equator, midday, white dust /desert conditions) - all these were the same as before. The lens and polariser were different though - this time a PL mounted Tokina with a Formatt circular polariser. I had the same problem with diffuse rainbow banding ( most obvious was green and magenta) over the entire image. It wasn't as strong as with the 50-300 + pola but still obviously there.
I can only think that the circular polariser is somehow interacting with the OLPF or sensor to produce the striping.

Robert Hart
10-10-2014, 11:48 AM
Mark. One more oddball theory to throw you off your game. I guess it is unlilkely where you are but do you have any ND gel, even if it is lighting gel. If not, can you ND with mattbox filters, the lenses to a wider aperture, say f4 - f5.6. My notion is to put a piece of ND gel in the space behind the lens. If you do't have any gel, any old piece of shirt packet clear stiff plastic mayl make do for part of my theory relating to reflection and a secondary polarisation of reflected light which might be happening at the sensor.

Direct image from the lens passes to the sensor once. Reflected light between the sensor/OPLF and the rear of the lens or maybe a planar element within it has to make multiple passes, therefore should be attenuated if there is an ND between lens and sensor. Also, a piece of gel is not going to be a true planar surface so might also furthur disrupt any reflection going on.

My theory is on polarised light becoming furthur polarised by the sensor's individual pixels maybe while reflected forward and then projected back onto the sensor in the reflection as an enlarged or shrunken image consisting of the pixel grid pattern but with the rows per inch or milllemetre being of a different pitch.

It might not even need be in focus. Therefore, a moire artifact may be produced that an OLPF cannot deal with.

Attenuating the reflected light in the space between lens and sensor with a piece of gel plus disrupting the reflection path with two more non-planar surfaces of the gel, might make things look different enough to identify it as an internal reflection issue created by a polariser.

Is the light coming out of the rear of a long lens arriving at the sensor across its width with less angle of incidence. My imagining is that reflections off the sensor and returns to the sensor would be at lesser angles of incidence than with a wider-angle lens, therefore a reflected image of the sensor pattern itself may much closer to true scale, close enough for a moire effect to be generated.

I hope I am not sending you off on a goose chase.

Do you have hard skies over there this time of year and using the polariser to bring back some colour. I imagine you would need ND18 to get lenses down to f4-5.6 if your skies are like our Pilbara summers, sometimes almost white to about 30 degrees above the horizon.

Kennan Ward
10-10-2014, 12:52 PM
Rainbow streaking. We had the same problem in April with a Dragon. Red team were so patient and great coaching. They went back and forth, upside down, finally replaced the sensor even! Never, ever had this much support from a camera company in 30 some years!!!!
Bottom line, I put the darn Nikon circular polarizer on again and back it came. Primarily in the 18% grey / color tones! I have since switched to a linear polarizer: http://www.edmundoptics.com/optics/polarizers/linear-polarizers/high-contrast-linear-polarizing-film/3435

I might add that for a feature film we are finishing, we used Davinci Resolve 11 color to diffuse the magenta streaking in order to obtain a 98% good clip color for these clips. A mask was produced to the four areas that were common place.

Hope this helps!

http://www.reduser.net/forum/showthread.php?116094-Dragon-is-a-field-of-Wild-Flowers!

Kennan & Karen Ward

COMING SOON, IT'S A WILD LIFE FEATURE FILM 88 MINUTES 4K 2.39 CINEMA!

Mark Deeble
10-11-2014, 09:34 AM
Thanks Bob - unfortunately we have none of that with us (just two of us in a fly camp at present) but I'll try and test when we get back to base. Both lenses show the same 3-4 vertical colour sequences - one a telephoto zoom (50-300), the other a wide angle zoom (11-16). Magenta seems to be the most noticeable, but I think the rest are there if I over-saturate, and pull down the exposure.


Keenan - thanks for all the info . I see the same artefact on your third image - much more subtle than we've been getting but the same striping, nonetheless. It is something I never saw with the MX sensor using the same lenses and filters, which makes me think it is either the new OLPF or higher pixel density or some other aspect of the dragon sensor.
Good luck with the feature, please post here when it's done. We are 2.5 years into shooting a wildlife feature on elephants...not quite there yet!

Kennan Ward
10-12-2014, 11:06 AM
Mark,

Good luck with the new film and story! Karen and I know you will have a great film when you are finished! Thank you for keeping Elephants and nature in the hearts of everyone!

Finishing a six year feature film is kind of a transition, we would rather be back in the field creating instead of marketing. Hopefully Sundance will help us get the word out. We will post a teaser soon. Pm us when up for air, literally !

Best, Kennan

John Osteen
06-14-2015, 04:45 PM
Mark, were you able to solve the problem? I'm having the exact same thing happen to me.