Purple Blacks IR in Forrest

[Martin Preiss]

It was purple and it was transparent and it was also on actor in the middle of frame.

That's a bad coating probably older lens...bad coating in any of the optical element (not just the entering element). There is more then 42 layers of coating, thinner then human hair (all of them), at high cinema lenses. Each layer is several nanometers thick and by "interference at thin layers" acting as filter.

 

[Laco Zamba]

MP was rented from Prague ;-) (V.....e)

 

[J. Eric Camp]

Yeah, sounds like a flare.

Back to the Histogram: I redid the black shading just in case. Same result.

Interesting enough.

I shot black with the cap on.

I bring it up and in RED Space I get a black image. (see image 60 below)
But in RED LOG I get a grey image. (see image 83 below)

This looks like the culprit. Is this just inherent of the new RAWish metering.

 

[Martin Preiss]

MP was rented from Prague ;-) (V.....e)

Then it was Focus Puller's fault Don't believe in "bad" lens in V.....e, I highly rate Jindrich, he still has best equipment! But it's just theoretical discussion when you can't see the picture...

 

[Martin Preiss]

Yeah, sounds like a flare.

Back to the Histogram: I redid the black shading just in case. Same result.

Interesting enough.

I shot black with the cap on.

I bring it up and in RED Space I get a black image. (see image 60 below)
But in RED LOG I get a grey image. (see image 83 below)

This looks like the culprit. Is this just inherent of the new RAWish metering.

Cool thanks for that Offhollywood I was just hitting myself that I didn't shot it I can stop beating myself now . And somebody from RED can answer why?! Just a ordinary "noise" ?

 

[Graeme Nattress]

Histogram black bump has a simple explanation:

RED records and processes raw, and we encode all brightness levels the sensor responds to, even below black levels. There is a black level offset in the data so as to preserve these negative values.

There is no "right" way to deal with them. You could clip at zero, but then you can get a rectified noise floor, which might not look as good as just crushing the blacks down. Either way, your choice in post as it's all preserved.

Graeme

 

[J. Eric Camp]

Yeah Graeme!

 

[Laco Zamba]

Then it was Focus Puller's fault Don't believe in "bad" lens in V.....e, I highly rate Jindrich, he still has best equipment! But it's just theoretical discussion when you can't see the picture...

I don't think it was bad lens. It was flare, but we were not able to remove it. RED 18-50 helped. I'll send you PM with footage.

 

[Martin Preiss]

RED records and processes raw, and we encode all brightness levels the sensor responds to, even below black levels. There is a black level offset in the data so as to preserve these negative values.
Graeme

If there is no response to brightness level then there is no signal from chip, just noise from heat and electromagnetic field. If there is no light there is no picture.

Well I must admit I don't understand that deeply enough, i remember some lecture from Integral Multiplicative Criteria of Quality Q=(Q1*Q2*Q3*Q4)^v as Quality standarts for digital or analog signal and there was some significant limit to where there is more then 45dB of Signal/Noise ratio the quality Q = 1 and if there is under 25dB it's not acceptable.

Anyway is there some results from MTF of Mysterium chip? Did somebody from RED did MTF (Modulation Transfer Function)? I will be interested in results...

 

[Graeme Nattress]

Noise is predominated by read noise, which is poisson distribution. We just don't arbitrarily chop off the left hand side of the distribution, even though it does not necessarily correlate with image data. Often RAW converters for DSLRs take the other point of view, as do video cameras, and not only crop off below black, but all that black hump. This can look ghastly when you come to grade shadows. I end up having to fix that all the time with my photography work.

MTF is a system parameter, ie lens, sensor, recording, processing and display. And yes, I've plotted curves for such systems. But these things are usually best done yourself with appropriate charts and methodologies.

Graeme

 

[Martin Preiss]

MTF is a system parameter, ie lens, sensor, recording, processing and display. And yes, I've plotted curves for such systems. But these things are usually best done yourself with appropriate charts and methodologies.
Graeme

Hmm so the histogram peak is design choice.No more to argue about.

MTF: I'm suprised there is no official white paper for MTF of RED (just the sensor or sensor+A/D+codec compression... All Kodak films has and other digital cameras as well (i believe I saw MTF for Sony somewhere)

And maybe it's better to do it yourself, but best is razor sharp test directly on chip, doesn't it Graeme? Hardly to do with whole camera. Or arrow test, but I don't have any of those, don't need them for my work, just interested in the MTF curve for RED, then I can compare it to what I know (film stock) and know something more when going to field to use it...

Anyway can we go back to BLACK-PURPLE?

 

[jbeale]

The MTF of the chip in isolation is no doubt of interest, but it is theoretical interest only, because the user can't get at that data. The actual delivered MTF from the R3D file is the only useful MTF if you are making motion pictures, and that's what you can measure yourself.

MTF for film stock has more actual use because you can (in theory anyway) scan them with any resolution scanner you have, now or later. But that is not true for digital cameras. Film MTF also varies with the specific development details, there is no guarantee it will match up between labs, or in the same lab between days.

I'm curious, has sensor MTF for any popular digital camera (DSLR or digital cine) ever been published? I haven't seen such information, but I haven't looked.

 

[J. Eric Camp]

So yes.

Back to the IR purple blacks. Any thoughts anyone?
I have not seen it in my b16 tests but most of my blacks have been shadow or duvatine.

 

[David Battistella]

I'd like to see the shot in motion. It would be good to see how the camera is shifting from one color to another like thus on the fly. It seems like maybe the white balance button on the side of the camera may have accidentally been pressed.

**EDIT**
Just realized that it was a Canon Comparison.


Is the wb setting locked at 5600k for both shots?
was the still exported with the exact same settings for both shots?

I am not accusing, just trying to troubleshoot.

These shots should match as there is only about 10 feet between shots.

 

[Rainer Fritz]

We had problem with MP 37mm in one scene with white background and one actor. We had purple circle in the middle - maybe some strange kind of flare. Circle size was dependend on f-stop.

RED 18-50 helped us and worked fine.

MPs are telecentered lenses. So the light comes parallel on the sensor. They are specially designed for digital use. maybe thats the reason why you have there a difference to the red zoom... ?

 

[Martin Preiss]

Is the wb setting locked at 5600k for both shots?
was the still exported with the exact same settings for both shots?

Yes WB was locked at 5600. I don't understand both shots?! One is from Canon Ixus and one is Canon Ixus photo of HD display and one is from RED directly from REDALERT without any modifications.

The issue is in the sensor or the interpretation! It's not in fault settings or anything else. It happened today, in night shoot with fire. Even some hairs are tinted to purple! It's annoying! A lot guys. And I must say why nobody from RED responded to this yet? Please...?!

 

[Dan Hudgins]

Reflections?

Reflections?

MPs are telecentered lenses. So the light comes parallel on the sensor. They are specially designed for digital use. maybe thats the reason why you have there a difference to the red zoom... ?

I think the round shape may be a reflection of the lens iris/stop off the sensor cover, filters, the OLPF, or elements, you can see that in some of the Drake camera footage. It may depend on the radius of the lens elements if the reflection gets sent back in focus or not, like some bad lenses can show an upsidedown reflection of car head lights or candle flames. Or a reflection of something on the back end of the lens that is not black enough. Film is matte finish so does not give clear reflections like the sensor cover glass.

A list should me made of lenses that show this with example frames so others users can avoid those lenses.

 

[Dan Hudgins]

Far RED not just IR, XB29 filter + heat absorbing glass?

Far RED not just IR, XB29 filter + heat absorbing glass?

Yes WB was locked at 5600. I don't understand both shots?! One is from Canon Ixus and one is Canon Ixus photo of HD display and one is from RED directly from REDALERT without any modifications.

The issue is in the sensor or the interpretation! It's not in fault settings or anything else. It happened today, in night shoot with fire. Even some hairs are tinted to purple! It's annoying! A lot guys. And I must say why nobody from RED responded to this yet? Please...?!

In amother post here a was show a Spectragram of the filters in the RED ONE (tm) sensor, the blue filter seems to pass far RED, not just IR, so things at the far red can go from RED to Magenta. It is a problem with many Bayer filter sensors since if the filters are made better the EI becomes lower.

You may be able to fix this by using a XB29 filter and a sheet of copper doped "heat absorbing" glass in front of your lens.

http://www.reduser.net/forum/archive/index.php/t-6530.html

http://www.glenspectra.co.uk/glen/filters/colour_imaging.htm

http://www.glenspectra.co.uk/SiteResources/Data/Templates/1product.asp?DocID=147&v1ID=

Some kinds of heat absorbing glass have a blue green cast and would cut some of the extra red and far red light like above 625nm.

 

[Martin Preiss]

http://www.reduser.net/forum/archive/index.php/t-6530.html
http://www.glenspectra.co.uk/glen/filters/colour_imaging.htm
http://www.glenspectra.co.uk/SiteResources/Data/Templates/1product.asp?DocID=147&v1ID=
Some kinds of heat absorbing glass have a blue green cast and would cut some of the extra red and far red light like above 625nm.

Thanks Dancad for your highly technical help. I believe you are highly educated in those things, as I can read from your post.

Unfortunately I think your solution is unpractical to fast pace of film set. I need something more practical, something you can rent quickly enough or to be removed in design of camera itself. You need to trust and depend on your work horse, not to fix this and that by doing magic in front of lens. I can understand with small problems, but exposition and color is one of the main stuff! You need to focus on different magic in front of lens!

 

[Robert Horwell]

i had this purple in blacks problem this week on a night shoot, also the skin tones were very green, i did black calibration, even reset all factory values. Nothing changed it, we had 17" panni monitor, and Marshall on camera, also EVF it was the same in all outputs. However when we viewed the footage later from the QT's all traces of the purple and green where not seen.!