Custom white balance and noise

Hi all,

have a question in regards to custom white balance and noise.

Does shooting in a custom / one push white balance setting increase noise especially if performed in deeper water and shooting with no lights compared to shooting auto white balance?

I think yes based on my use of Panasonic LX100.

This camera has good manual white balance but shows quite a lot of noise even at iso 400/800 when manual white balanced at depth. This camera is acceptable on land use even at iso 3200.

When you set a custom white balance underwater, the camera is forced to boost (or apply gain to) the red channel in order bring the signal level closer to the green and blue channels so that white appears to be white (even though in reality a white object at depth will reflect a lot more green and blue wavelength light than red wavelength light). So, imagine your ISO is set to 100 for simplicity's sake -- the red and green signal may be treated as iso 100, whereas the red signal will be treated at a higher iso -- let's say ISO 1600. Whether this is done at the hardware analog-to-digital converter level or as a pure mathematical signal multiplication probably differs from camera to camera (though I suspect it's generally done as a digital multiplication of the red channel signal after analog-to-digital conversion).

So, short answer -- yes. White balancing at depth will result in noise, especially in the red channel. And that noise will increase the deeper you go. Incidentally, the exact same thing will happen if you shoot in RAW and white balance in post -- the post-processing software is boosting the red signal in the RAW data, leading to increased noise. The increase in noise will be proportional to the severity of the white color correction.

And there's no way around it really. Using a red filter to try to balance the amount of red/blue/green light that reaches the sensor results in less overall light reaching the sensor and a higher ISO (thus noise) as well. I suppose you could compensate by using a slower shutter speed in that scenario, assuming you composition allows for a slower shutter speed.

When you set a custom white balance underwater, the camera is forced to boost (or apply gain to) the red channel in order bring the signal level closer to the green and blue channels so that white appears to be white (even though in reality a white object at depth will reflect a lot more green and blue wavelength light than red wavelength light). So, imagine your ISO is set to 100 for simplicity's sake -- the red and green signal may be treated as iso 100, whereas the red signal will be treated at a higher iso -- let's say ISO 1600. Whether this is done at the hardware analog-to-digital converter level or as a pure mathematical signal multiplication probably differs from camera to camera (though I suspect it's generally done as a digital multiplication of the red channel signal after analog-to-digital conversion).

 

So, short answer -- yes. White balancing at depth will result in noise, especially in the red channel. And that noise will increase the deeper you go. Incidentally, the exact same thing will happen if you shoot in RAW and white balance in post -- the post-processing software is boosting the red signal in the RAW data, leading to increased noise. The increase in noise will be proportional to the severity of the white color correction.

 

And there's no way around it really. Using a red filter to try to balance the amount of red/blue/green light that reaches the sensor results in less overall light reaching the sensor and a higher ISO (thus noise) as well. I suppose you could compensate by using a slower shutter speed in that scenario, assuming you composition allows for a slower shutter speed.

Thanks Dreifish! Exactly what i was thinking.

So given this and the fact that in the end only very few cameras successfully perform white balance and most often than not it leads to other problems besides noise like purple blues etc, why do we bother so much about custom white balancing and not about accurate colour representation. And by accurate i mean what we the divers see while diving. Because usually and without lights we dont see reds more than the camera sees it,we dont see a white balanced image.

So if its not there anyway why do we try to create it (can also use fake it) and try to give to to the audience, whatever that audience is, a fake white balanced image.

"Photography freezes time" they say. If so then video in my mind replicates time and in my mind it should be replicated as accurately as possible.

Just some thoughts...

This is not entirely correct. The eye+brain (I include the eye because it is my understanding that some "post-processing" of the image data takes place in the retina) does some color correction even though it is incomplete. I had an experience that illustrates this: After shooting some soldierfish in a mini-cave in the side of submarine lava flow formation (Big Island) I decided to chimp to see if my flash exposure was correct as there was time to re-shoot in case I screwed up. The lava formation was large enough that I was in its shadow so the light was very blue even though it was not too deep. When I looked at the image on the camera's screen the soldierfish appeared day-glow orange. I later re-checked these images when I got topside and the orange color was normal. My eye+brain had been compensating for the rather blueish light. Recall also that when you shoot under tungsten lighting the image will be rather red unless the WB is set to tungsten. However we do not perceive all this redness with our vision. Our eye+brain is doing some color correction.

quote:

This is what i mean

Photo is grab frame from footage shot on Sony RX100 Mk4. One push white balance no lights depth is 19m.

Second photo is the same wreck same day same camera Auto WB. Yes i am using lights in this one but i am not so close to affect so much. What i can tell you is that the second photo is much much closer to what the condition was really like on the dive. May be a little less green than in the photo but for sure not what is shown in the first photo.

This is not entirely correct. The eye+brain (I include the eye because it is my understanding that some "post-processing" of the image data takes place in the retina) does some color correction even though it is incomplete. I had an experience that illustrates this: After shooting some soldierfish in a mini-cave in the side of submarine lava flow formation (Big Island) I decided to chimp to see if my flash exposure was correct as there was time to re-shoot in case I screwed up. The lava formation was large enough that I was in its shadow so the light was very blue even though it was not too deep. When I looked at the image on the camera's screen the soldierfish appeared day-glow orange. I later re-checked these images when I got topside and the orange color was normal. My eye+brain had been compensating for the rather blueish light. Recall also that when you shoot under tungsten lighting the image will be rather red unless the WB is set to tungsten. However we do not perceive all this redness with our vision. Our eye+brain is doing some color correction.

 

quote:

 

. Because usually and without lights we dont see reds more than the camera sees it,we dont see a white balanced image.

So if its not there anyway why do we try to create it (can also use fake it) and try to give to to the audience, whatever that audience is, a fake white balanced image.

 

Tom you are absolutely right that the brain is always white balancing.

​When we are subjected to an image that is biased towards one particular colour then after a few minutes our brain starts to white balance the colours, trying to minimise the more prevalent colour and boosting the other two primary colours.

​This is why we often think that our images looked better on the camera when we recorded them underwater and then we download the footage afterwards and are disappointed in how the colours look on our computers, it is because our brain is white balancing while underwater.

Colour also plays a psychological role in our videos, we perceive dark green to be dangerous and edgy, yet a more neutral white balanced image has a calming effect on us & therefore is usually more appealing.

Why not just custom white balance whilst descending to target depth? E.g. white balance at say 20 meters and then continue to 40 meters. This way you eliminate some of the blue or green color cast, but are not yet boosting red to extremely noisy levels.

Your pix are not too bad. I grabbed the jpegs you posted and imported them into Lightroom to do the tweaking. The second shot that you AWB got only a tint adjustment (green-magenta slider) in the way of color balance change from me so color temp may have been OK via AWB. AWB may ONLY do color temp and no tint adjustment - ask Sony not me!!

 

 

Your pix are not too bad. I grabbed the jpegs you posted and imported them into Lightroom to do the tweaking. The second shot that you AWB got only a tint adjustment (green-magenta slider) in the way of color balance change from me so color temp may have been OK via AWB. AWB may ONLY do color temp and no tint adjustment - ask Sony not me!!

I have no idea how the Sony AWB thinks. The thing is that in my eyes none of the versions represent how it was on the dive. Yes, the one you retouched looks more pleasant to the eye, but still, I would not end up with such a version on the final video.

To me, the results do not justify the custom WB and all the trouble around it. I think the way forward for me it will be experimenting with fixed white balance values and tint adjustments for deeper dives or when vis is limited.

Why not just custom white balance whilst descending to target depth? E.g. white balance at say 20 meters and then continue to 40 meters. This way you eliminate some of the blue or green color cast, but are not yet boosting red to extremely noisy levels.

You can certainly experiment with this.

With cameras that have a limited range for white balance adjustment (e.g. Sony cameras), this is more or less what you are forced to do since trying to set white balance too deep will likely result in an error.

Still, for my tastes, the results of white balancing at a shallower depth and then filming deeper are a bit too greenish-blue. It's not just the color itself, either. A large part of the contrast we perceive in an image comes from color contrast. For example, reds are much more noticeable when placed against green, blue against yellow, etc. So, images that include the warmer tones tend to have more contrast and are more "poppy" than ones that are all just shades of blue and green.

My preference is to use artificial light (strobes or video lights) when filming below 10 meters, and keeping the white balance set to the temperature of the lights. I find that it producing much more pleasing results than trying to shoot with only natural light.

Does it accurately reflect what I saw on the dive? No. But for me, photography isn't about striving for the most accurate version of reality. It's about producing beautiful images. Accurately capturing reality is not as straightforward as you might think, anyway, since cameras see the world very differently than our visual system processes the world. Physiologically, most of what you "see" is actually a very detailed model constructed by your brain based on your past experience, based only partially on the limited, low bandwidth real-time input from your eyes.

You can certainly experiment with this.

 

With cameras that have a limited range for white balance adjustment (e.g. Sony cameras), this is more or less what you are forced to do since trying to set white balance too deep will likely result in an error.

 

Still, for my tastes, the results of white balancing at a shallower depth and then filming deeper are a bit too greenish-blue. It's not just the color itself, either. A large part of the contrast we perceive in an image comes from color contrast. For example, reds are much more noticeable when placed against green, blue against yellow, etc. So, images that include the warmer tones tend to have more contrast and are more "poppy" than ones that are all just shades of blue and green.

 

My preference is to use artificial light (strobes or video lights) when filming below 10 meters, and keeping the white balance set to the temperature of the lights. I find that it producing much more pleasing results than trying to shoot with only natural light.

 

Does it accurately reflect what I saw on the dive? No. But for me, photography isn't about striving for the most accurate version of reality. It's about producing beautiful images. Accurately capturing reality is not as straightforward as you might think, anyway, since cameras see the world very differently than our visual system processes the world. Physiologically, most of what you "see" is actually a very detailed model constructed by your brain based on your past experience, based only partially on the limited, low bandwidth real-time input from your eyes.

Great post thanks!

I was reading somewhere (i am afraid i cant remember where though) that the Sony WB error is not a "real"error. The camera white balances its is just giving back this error. No idea if it is right or wrong. The recommendation was to ignore it and store the setting...

I totally agree with you on the reality part. If i am using lights then the image ideally should reflect what a diver holding a light would have seen. Which is a more colorful and pleasant image. An image shot deeper than 10 meters with available light (and i am talking about video because photography is more artistically liberal, dont know even if the term exists ) when custom white balanced doesnt look like what i am expecting to see.

Of course this is all personal opinion with zero value to anyone else.

I don't like using video lights as they usually drive underwater animals away.

I don't like using video lights as they usually drive underwater animals away.

Well, thats a pretty general claim I would not sign up to. In some areas Mantas can be shy of light, so if the dive guides gives that alert, I certainly turn them off - just happened to me last week at Manta Point at Addu-Atoll. Most sharks dont mind. Many turtles and certainly batfish are even attracted by lights and like to closely inspect them.

Well, thats a pretty general claim I would not sign up to. In some areas Mantas can be shy of light, so if the dive guides gives that alert, I certainly turn them off - just happened to me last week at Manta Point at Addu-Atoll. Most sharks dont mind. Many turtles and certainly batfish are even attracted by lights and like to closely inspect them.

I tried video lights on hammerheads and blacktip sharks they were scared like hell and so I gave up using video lights. Frankly I don't like having a buddy with really strong wide angle video lights, strobes are fine as they are not continuous.

Do you guys that do custom white balance use a Grey or a white card? Which one would yield the best results?

Sent from my E5603 using Tapatalk

White or Grey. It doesn't really matter. The key is to have something that has no colour cast (so have the same about of Red, Green and Blue)

I have a WB slate which has actually a slight blue/green cast so making the camera over compensate a little towards red.

But anything Matt white or grey is a great start

Regards - Richard