Welcome to our community

Be a part of something great, join today!

Log in
  • Hey all, just changed over the backend after 15 years I figured time to give it a bit of an update, its probably gonna be a bit weird for most of you and i am sure there is a few bugs to work out but it should kinda work the same as before... hopefully :)

About white balance...

nguyenduy

nguyenduy

Member
Joined
Apr 2, 2008
Messages
24
Reaction score
0
Points
0
I always have problem about color temp when I shoot with RED. You know, when I use the auto WB or tungsten or daylight, it is more blue. So I have to use the manual to adjust the color temp. Do you the same problem with me? please give me a explain.
thanks.
 
I've not been terribly impressed with the auto-white balance. It get me in the ballpark sometimes when light temp is really extreme, but it never seems to give me exactly the results I want. More often than not, I set white balance for 5k, and deal with it in post. As a result, I have noticed that I tend to accept color temp as a natural element of the scene, and will typically only balance the RAW to achieve a particular look, or if there is something that should most obviously be white, but which isn't. That said, I think its more beneficial to shoot a color chip chart as part of your slate, and use that as a color reference when tweaking the RAW numbers in post.
 
Auto WB on the RED One is evil.

As Vigen said, WB is all metadata anyway and the sensor's native balance is 5000K.

I do as Kyle does... 99% of the time, I leave the camera set at 5000K and I adjust color and white point in post.
 
Yeah, the only real reason to bother white balancing away from 5000K is to make things look a little nicer on client monitors, or to see if there will be unacceptable noise if you know you're going to end up balancing far from 5000K in post (which is hard to tell from the 720p monitoring signal anyway).
 
I put up a Kodak grey card in front of the camera, and the auto-WB seems to get it right. If I'm shooting, I shove a macbeth in there to capture what the light is doing, and sample from the mid grey for any post WB and take that as a starting point to tweak.

Graeme
 
Graeme Nattress said:
I put up a Kodak grey card in front of the camera, and the auto-WB seems to get it right. If I'm shooting, I shove a macbeth in there to capture what the light is doing, and sample from the mid grey for any post WB and take that as a starting point to tweak.

Graeme
Click to expand...

I too have not had any problem with AB. it usually gets it really close for me
 
The reason I dislike the AB is not because it gets it wrong, it's usually pretty accurate, it's more of just a consistency thing. I'm also prejudiced against AB because back in old days, before programmable user buttons, I had a few clients that would rent the camera and their operator would use the AB constantly, and often by accident with it assigned to button 1.
 
I never use WB. I assingn only magnifying to all buttons, nothing else. IMO, RAW-View is the best mode to evaluate exposure, contrast and colour. Setting metadata to 5000K, sensitivity to 250ASA and gamma to Rec709 keeps the metadata, when used in post, closest to that what one would call a native sensor setting. No highlights will be cut off when processed later in post using the metadata.

I never found a WYSIWYG workflow working meaningfully with the Red nor any other camera system I used, except you do ENG work, meaning shoot, cut, broadcast - no real postproduction involved. For this kind of applications Red is the wrong camera system anyway. Experience showed that trying to establish a WYISWYG workflow through metadata is prone to mistakes that will lead to unwanted quality reduction.

I can understand that working with metadata through out the pipeline has some fascinating aspects but in Red's case I find it more hindering than anything elses. I use the RedOne as a digital filmcamera where all the processing is done ouside of the camera, considering RedOne's sensor a reversal, daylight sensitive filmstock with approx. 320 ASA sensitivity and a latitude of approx. 9 stops. Hence we don't "paint" the picuture via metadata on set. Plus doing so costs valuable time on set. Grading will be done later when there is the time to accomplish that in the necessary environment.

Additionally there have been many tests which show that processing at 5000k and correcting the colour later in the pipeline leads to less noise in the blue channel when shooting in a tungsten environment than using camera metadata to WB the picture directly in RC or RA for whatever reasons.

Hans
 
Hans von Sonntag said:
.... Experience showed that trying to establish a WYISWYG workflow through metadata is prone to mistakes that will lead to unwanted quality reduction.

...

Additionally there have been many tests which show that processing at 5000k and correcting the colour later in the pipeline leads to less noise in the blue channel when shooting in a tungsten environment than using camera metadata to WB the picture directly in RC or RA for whatever reasons.

Hans
Click to expand...

Hans, I'm curious about these tests. Can you references the results? In theory, this shouldn't be the case because Meta is meta. The RAW data is recorded the same regardless of how you "paint" the scene on set. Setting the WB (or any other meta settings) in camera should be mathematically identical to setting those same values in RC or RA (or any other R3DSDK tool).

Based on this same premise of RAW data via Meta data and the nature of non-destructive image manipulation, there should be no reduction in quality when processing R3D footage, outside of quantization as a result of bit precision (either clamping values to fit within 12bits, or losing decimal precision when converting internally to an integer data type).

Interestingly, I've never really considered this before, but it would be damn cool if the R3D tools (including the SDK) could return floating point data natively, to help minimize this.
 
We return 16bit buffers, which have more than enough precision for the underlying 12bit data and math. Going to float would be overkill. Internal processing is done in 16bit or float depending on what is being done.

I think what Hans is saying is that he prefers the look of white balancing the developed raw image, rather than the white balancing being done as part of the raw development. Is that what you're saying?

Graeme
 
Graeme,

It was my understanding that while the returned image buffer is 16bit, that only the first 12-bits are relevant. Ie, that data is clamped to 12-bit values before the image buffer is returned to the calling application. I need to go double check my code, but, can you confirm one way or the other... I might be shooting myself in the foot!
 
The most significant bits are the most relevant, but that doesn't mean the LSBs are irrelevant. If you export an image with -4EV, then the 12 LSBs are the ones to look at, and a post multiply of 64 should produce the same image as leaving EV at 0. However, due to the internal processing being at a high bit depth or float, I'd expect the EV0 version to be ever-so-slightly better (assuming no clipping).

Graeme
 
what happens when WBing for 32 K

what happens when WBing for 32 K

If someone could help me out here explaining this a bit more that would be great. I can see what's going on, but wonder about the whys of it.

I can see in a tungsten lit scene, by looking at only the blue channel, in RC when you change the WB from native 5000 to 3200, the channel gets brighter and much noiser. I see what's going on but I'm not sure of what's actually happenning. I know that a tungston lit shot is weak in blue light to begin with, but beyond that, what is being done to the blue channel when the white balance is shifted?

Is the blue channel being boosted to make the image appear correct? Thus, the noise?
 
Graeme Nattress said:
I put up a Kodak grey card in front of the camera, and the auto-WB seems to get it right. If I'm shooting, I shove a macbeth in there to capture what the light is doing, and sample from the mid grey for any post WB and take that as a starting point to tweak.

Graeme
Click to expand...

Exactly!!!!! I tell every DP that we work with to do this exact same thing. They listen about 10% of the time. This isn't a video camera people.
 
Philip Powell said:
I know that a tungston lit shot is weak in blue light to begin with, but beyond that, what is being done to the blue channel when the white balance is shifted?

Is the blue channel being boosted to make the image appear correct? Thus, the noise?
Click to expand...

Yes. When you white balance for tungsten you are amplifying blues, and crushing reds/greens. The noise in the blues that would normally be under the floor starts to come up into the noticeable ranges. This is why it's actually recommended to CTB your lights or lenses, if you are striving for the best image possible when shooting tungsten.
 
Thanks for your help...

Thanks for your help...

Thank you everybody. I am clearer to hear all of you.
Thanks
 
Thanks for the feedback on that. I've got a lot of colorimetry work ahead of me, and this kind of feedback really helps as I develop code.

Graeme
 
You must log in or register to reply here.
Back
Top