An honest request for better color on monitor out./.

[Obin Olson]

Hey guys..RED team I am sure you know what I have been doing as I have not been off the phone with you! (thanks for the help) anyway, it looks like the LCD/HDSDI/HDMI out are not correct in color, I have found that oranges/tans and blues don't come out right at all....the RAW files look awesome but the realtime outputs are whack...can you guys please check on this and update as is needed? you will see my pictures that I sent to Dan at Red and see what I am talking about...most stuff looks fine but not the colors I am talking about.

this camera is really sweet. lets NAIL this issue!


BTW I seem to have a bad red drive cable as it will cut the camera out at times and will show as "NO MAG' at times...when it works I can wiggle on it to make it shut down the camera...drive is fine on PC via USB

[Obin Olson]

AND BTW the RED DRIVE and POWER cables should have an L on the camera side...they are in the way of the powerpod on the rear rails...

[Brent J. Craig]

Hopefully we will be able to fix this ourselves soon with 'look' files.

[Jon Corcuera]

In what build did you see that ? the monitor out going colors

[Obin Olson]

build 14 I have now...and the other camera in town is just like this...

[Jon Corcuera]

did you have the same problem with build 12 or 13, since 14 is beta ... :)

[Stuart English]

did you have the same problem with build 12 or 13, since 14 is beta ... :)

The colorimetery of the camera has not changed for a long time, so changing builds has no effect; and yes I would expect any other camera to have the same color response. Obin, its interesting that you say the RAW data looks right but the REC709 outputs don't in your opinion represent orange and blues correct.

Have you tried enabling the Color Chip test pattern, and if you do that, do you see the same color representation issue in your judgement?

[Wes Printz]

AND BTW the RED DRIVE and POWER cables should have an L on the camera side...they are in the way of the powerpod on the rear rails...

If the current placement of the connector on the camera remains in the same position, a Right Angle connector on the drive cable at the camera would interfere with the power connector.

[Obin Olson]

Check the images I sent in..

[Deanan]

What colorspace and which application are you looking at the raw with?

[Evin Grant]

I've always considered the SDI out of the Red more of a "Video Tap" than a true representation of the colorometry. The RAW file it's self is very powerful and the simple Rec 709 lut and fast debayer of the video out is not that relevant if you think of the Redcode file more like a digital negative.
This isn't like a traditional video camera where WYSIAYG* is the rule.

*What You See Is ALL You Get

[I Bloom]

On this note:

It would be great to have an "Oversaturate" view mode, that could be switched on and off easily like false color exposure.

What is the use of this? I've noticed that briefly oversaturating is a technique that a lot of colorists use to see the hue on skin tones and the contrast of hues within a scene. By exagerrating the saturation we are more sensitive to tiny shifts that we might not normally notice.

For example I stopped by a Red shoot over the weekend using Diva lights,
they didn't realize that if you dim those lights you get noticable magenta shift.

If we could oversaturate breifly, we could immediately see green and magenta creeping into skin tones, from offending lights and green-screens.

On my wishlist.

IBloom

[Obin Olson]

used redcine....again raw files look great....why not a raw like display from preview?

[Stuart English]

used redcine....again raw files look great....why not a raw like display from preview?

We could do that - when LOOK files are operational it would be relatively easy to do - but back to first principles, a) are you trying to color grade on-set or use the Preview Out as a "video tap" for focus / framing etc and b) when you turn on the Color Chip or SMPTE bars what do you now observe about the orange and blue colors?

I will take a look at the images you sent in, but the issue there is that there must be one other imaging device in the chain.

[Deanan]

Are you comparing rec709 in redcine to rec709 in monitoring? or rec709 and cameraRGB?

[Michael Stanmore]

Why isn't Evin Grant's name yellow anymore? Have you been demoted to white status? like us rabble?

[Patrick Tresch]

Obin, its interesting that you say the RAW data looks right but the REC709 outputs don't in your opinion represent orange and blues correct.

I've the same problem but... in redcine.

Looking at the footage with "Camera RGB" color space, colors looks natural (a bit unsaturated). Passing to REC709/ colors just look like fluorescent. You just have a pink cast in skin tone wich is impossible to get natural neither with tint nor with the color sliders.

Would'nt it be a gamma table problem?

Thanks Pat

[Deanan]

The rec709 space is a bit saturated so turn down the saturated a bit. Likewise, cameraRGB is a wider colorspace so bumping up the saturation helps also.

You can adjust saturation in the camera to bring rec709 down.

[Dj Joofa]

I've the same problem but... in redcine.

Looking at the footage with "Camera RGB" color space, colors looks natural (a bit unsaturated). Passing to REC709/ colors just look like fluorescent. You just have a pink cast in skin tone wich is impossible to get natural neither with tint nor with the color sliders.

Would'nt it be a gamma table problem?

Thanks Pat

Welcome to the elusive world of color fidelity.

If the camera is using a matrix and/or scaling of channels to do color correction for white balance, probably, there will always be some colors that will not look "right." This issue here is more than the rotation of color axis by a matrix -- ideally a non-linear transform is required, however, matrix multiplication is just linear. The non-linearity required is not to be confused with gamma non-linearity which is a different issue.

[Obin Olson]

bars and chips look just fine...If the camera output was "raw" in color it would be VERY nice....that is what I would want to see..because lighting and color palette are built on the set or location I would want a "raw" look ..not what we have....atleast I know the camera is ok, because from what I have seen in the monitor it looked like the talent had green hair!

[Obin Olson]

As a "base" I would like to start with a raw look in the camera in part because we start with a "raw" in post...I think that makes the most sense....


Like I have said, it was so far from the "raw" look I thought the camera was broken!

[Obin Olson]

BTW I don't think we need "white balance" at all..unless it does change the gain of each color channel...?

[Obin Olson]

So Deanan..if I turn it down....will that effect the RAW recording?

[Patrick Tresch]

The rec709 space is a bit saturated so turn down the saturated a bit. Likewise, cameraRGB is a wider colorspace so bumping up the saturation helps also.

Hello Deanan.

It doesn't seam it's only a saturation problem. The colors (red becomes pink...) are in an other "color space" or like if the phase would have been turned...:help:


Welcome to the elusive world of color fidelity.


Hello Joofa,

Even if I did some color grading in DS nitris, I still didn't get much to work with different color space than REC709.

It would be great if anyone who gets real insight about the "world of color fidelity & "their color spaces"" would write something professional about how to use color space with raw media and for instance R3d with REDCINE. Wich reference/on what monitor/wich output to what files...

Sorry but REDCINE has no tutorial. But we all know that we can do a lot with this application if well informed, and also big mistakes (and crap looking REDONE pictres) if NOT INFORMED.

Any help appreciated.

Patrick

[Stuart English]

bars and chips look just fine...If the camera output was "raw" in color it would be VERY nice....that is what I would want to see..because lighting and color palette are built on the set or location I would want a "raw" look ..not what we have....atleast I know the camera is ok, because from what I have seen in the monitor it looked like the talent had green hair!

If the output looks fine with test signals, the REC709 stages are O.K. and the issue lies upstream. What are you lighting the set with and how are you white balancing the camera.... and then the next question - if you do NOTHING to REDALERT or REDCINE, how do the recorded images there look when compared to the camera monitor output? And on what are you monitoring the
REDALERT or REDCINE output?

[Dj Joofa]

BTW I don't think we need "white balance" at all..unless it does change the gain of each color channel...?

I am not sure if your reply is to me, but I shall assume so.

I am not talking about the "white balance" you might do in a color correction package. Cameras generally do the relative scaling of signal internally for an intended white point chromacity. For e.g., among many other stages of internal processing, they involve applying internally higher gains to blue than green or red (hence, more noise in blue). Typically scaling, and later any matrix multiplication involved to shift the white point are done internally, and by the time you see the signals in raw format these have already been applied. Because of limited resources for a real-time transfer of data, many cameras opt for simple linear scalings and matrix multiplications, where actually a complicated non-linear transformation was required. Hence, what happens is that different portions of the raw data space signals are not distorted in the proportion a correct non-linear transform would have done.

Therefore, though most of colors are closely matched, there are always some colors that are not matched to the right intended destination and look off.

[Dj Joofa]

Hello Joofa,

It would be great if anyone who gets real insight about the "world of color fidelity & "their color spaces"" would write something professional about how to use color space with raw media and for instance R3d with REDCINE. Wich reference/on what monitor/wich output to what files...

Sorry but REDCINE has no tutorial. But we all know that we can do a lot with this application if well informed, and also big mistakes (and crap looking REDONE pictres) if NOT INFORMED.

Any help appreciated.

Patrick

Ideally, the best response should be from the Red team as they know their camera. However, I shall try to write something and send to you.

[Patrick Tresch]

Hence, what happens is that different portions of the raw data space signals are not distorted in the proportion a correct non-linear transform would have done.

Does this mean (in your opinion, as your not the ingeneer of this camera) that the RAW to REC709 conversion is not properly done by REDONE?

If you write something try to make it easy for us (non savant) to understand!:biggrin:

Thanks

Patrick

(BTW, it would be great if RED/Assimilate would write a comprehensive guide for their product REDCINE).

[Dj Joofa]

Does this mean (in your opinion, as your not the ingeneer of this camera) that the RAW to REC709 conversion is not properly done by REDONE?



I don't know what Red is doing. However, if they are using multiplications of matrices to shift white points then this problem may always stay there. The reason many people do not see it as a shifting of white point problem is that the "shifting of white point" is actually absorbed in the co-efficients of the matrices from one linear color space conversion to an-other.

[Obin Olson]

shooting in an office with florescent lights, playback in redcine with DELL 24 screens....not sure what the white balance was at...can shoot again and show you what I mean though...

[Deanan]

shooting in an office with florescent lights

This will get you funky colors depending on the fluorescents (especially office lights).

This is because fluorescents are very spiky in the spectral content and regular office lights don't try to be balanced.

Also, looking at monitor under fluorescents will influence the colors you see on the screen.

[Obin Olson]

so maybe the cmos chip on this camera shows spikes more then the regular video ccd??

[Deanan]

I wouldn't say more or less but just differently. Each has different responses to the spectrum and one flouro type might affect each differently.

The differences are greater depending on 3ccd vs single sensor. The dyes that sit directly on the sensor are different than what goes into a prism and therefore results in a different spectral response.

Hope that helps.

[Obin Olson]

so why does the raw file look ok then?

[M. Bergeron]

We did some tests with Red yesterday and skin color on the monitor looked magenta but when we loaded the footage in Redcine, it was perfect. We thought the same as Obin.

[Obin Olson]

I am glad I am not alone here... What are your thoughts red team on a better color match for live preview?

[Deanan]

Can you send a snapshot of the lcd and an R3D to us?

[Patrick Tresch]

We did some tests with Red yesterday and skin color on the monitor looked magenta but when we loaded the footage in Redcine, it was perfect. We thought the same as Obin.

What color space did you use in REDCINE to look at your R3d file?

Was it "camera RGB" or "REC709" (in the left tab)?

Thanks.

Patrick

[Obin Olson]

after I get back from out of the states I will do that for you, or you can ask AmariaOlson he has the camera here now at the office...

[Deanan]

Thanks. What are your WB and tint settings on the camera?