Color Balance: During the shoot vs RAW post-processing

[cbeasley]

This is an interesting (though old) discussion and sorry to be joining after so much time, but it gets to some fundamental questions I have about the chain from analog to digital to Raw to ACR. The last links digital-raw-ACR are clear.

It is often said that the Raw file is simply "the output of the sensor" with meta data, but this is only true with the qualification that the "output of the sensor" is after digital sampling. So, is ISO simply a gain of the analog voltages prior to sampling or is something more involved? If so, I assume there is an analog noise filter appled prior to sampling?

I am most interested in these questions in the context of underwater video, which can be view as a frequency-dependent attenuation problem. In this case, one can't solve the attenuation in post because of S/N in the red spectrum. It is better in RAW, but not much. More to point, not many cameras shoot in RAW (nor do many videographers shoot this way for obvious reasons!) so it is even more problematic for video. So, a specific quesiton:

Some cameras (Olympus in particular) have an Underwater white balance setting. Can this be viewed as a frequency-dependent ISO adjustment, i.e., is it done analog, which would improve the situation. Or is it just a tweak after sampling, with accompanying limitations (S/N). Manufacturers are not responsive to such questions. I do observe that it boosts the reds, but does not come close to what can be achieved with a red filter, which is a true analog spectrum tilt. ISO comes in because the red filter costs 3 stops in exposure which raises the ISO (aperture is usually wide open anyway due to low light levels and there is little sutter speed flexibility shooting video).

Finally, this pathological example (as some would view the underwater problem -- the normal answer is to just get light on the subject -- good luck) should open up the concept of white balance. In correcting for the light source only (tungsten, daylight, etc.) we ignore the transmission effect, or perhaps more accurately, we lump the source, transmission effects and reflectance into a couple of sliders called WB. Then, we try to solve the specific issues by other means: masks, split toning, dehaze, contrast pop, etc. Then we try to mitigate the downsides of these powerful tools with another string of noise filters and the like. It seems best to me to try to get the digital signals correct (as best we can) in the A/D process.

I will appreciate any commnets and/or references.

[xpatUSA]

This is an interesting (though old) discussion and sorry to be joining after so much time, but it gets to some fundamental questions I have about the chain from analog to digital to Raw to ACR. The last links digital-raw-ACR are clear.

It is often said that the Raw file is simply "the output of the sensor" with meta data, but this is only true with the qualification that the "output of the sensor" is after digital sampling. So, is ISO simply a gain of the analog voltages prior to sampling or is something more involved?

No, some cameras apply the gain after sampling, sometimes not even in-camera.

If so, I assume there is an analog noise filter appled prior to sampling?

An incorrect assumption, sorry.

I am most interested in these questions in the context of underwater video, which can be view as a frequency-dependent attenuation problem. In this case, one can't solve the attenuation in post because of S/N in the red spectrum. It is better in RAW, but not much. More to point, not many cameras shoot in RAW (nor do many videographers shoot this way for obvious reasons!) so it is even more problematic for video. So, a specific quesiton:

Some cameras (Olympus in particular) have an Underwater white balance setting. Can this be viewed as a frequency-dependent ISO adjustment, i.e., is it done analog, which would improve the situation. Or is it just a tweak after sampling, with accompanying limitations (S/N). Manufacturers are not responsive to such questions. I do observe that it boosts the reds, but does not come close to what can be achieved with a red filter, which is a true analog spectrum tilt. ISO comes in because the red filter costs 3 stops in exposure which raises the ISO (aperture is usually wide open anyway due to low light levels and there is little sutter speed flexibility shooting video).

<>

I will appreciate any comments and/or references.

ISO values are based firmly on illuminance which in turn is based on human luminous efficacy:

https://en.wikipedia.org/wiki/Illuminance

https://en.wikipedia.org/wiki/Luminous_efficacy

Since the color spectrum of human vision is specifically defined, it is inappropriate to discuss the effect of WB or filtering on ISO. That is because ISO is determined under defined lighting conditions and camera settings.

Here's the Japanese specification:

http://kronometric.org/phot/lighting/DC-004_EN.pdf

There is of course an ISO standard but you have to pay to get it.

HTH

[Manfred M]

Generally it is a good site practice to start a new thread rather than adding to a 10-year old one, so I have established a new thread for your post.

I have not done any underwater photography, but was an active scuba diver for well over a decade with hundreds of dives under my belt, so understand the behaviour of light under water fairly well. The reds disappear first, then the yellows, as one gets deeper. Finally, when we get deep enough, everything appears to be B&W.

[pnodrog]

At a guess an appropriate red filter is the way to go provided the additional exposure required is not achieved by increasing the ISO above about ISO 800 depending on sensor design. Beyond about ISO 800 I think that the reduction in dynamic range and the increased noise in the blue channel and to a more limited extent in the green would tend to undermine the benefit of the filtering.

As I said this is a just guess based on my understanding of the factors involved.

[dje]

It is often said that the Raw file is simply "the output of the sensor" with meta data, but this is only true with the qualification that the "output of the sensor" is after digital sampling. So, is ISO simply a gain of the analog voltages prior to sampling or is something more involved? If so, I assume there is an analog noise filter appled prior to sampling?

ISO adjustment was originally just analogue gain but as Ted says, these days digital gain may be used as well. As far as i know, noise filtering of the analogue output of the sensor pixels is not done and would not be practical. These days, many sensors include the ADC function on the sensor chip itself with one ADC for each column. Also note that noise as we percieve it in an image is spatial rather than time dependant. So filtering needs to be done spatially ie between adjacent pixels.

Some cameras (Olympus in particular) have an Underwater white balance setting. Can this be viewed as a frequency-dependent ISO adjustment, i.e., is it done analog, which would improve the situation. Or is it just a tweak after sampling, with accompanying limitations (S/N). Manufacturers are not responsive to such questions. I do observe that it boosts the reds, but does not come close to what can be achieved with a red filter, which is a true analog spectrum tilt. ISO comes in because the red filter costs 3 stops in exposure which raises the ISO (aperture is usually wide open anyway due to low light levels and there is little sutter speed flexibility shooting video).

I suspect the underwater WB setting is just a variation on normal WB settings. An optical filter would seem to be the most effective solution.

In correcting for the light source only (tungsten, daylight, etc.) we ignore the transmission effect, or perhaps more accurately, we lump the source, transmission effects and reflectance into a couple of sliders called WB.
.

Yes it's the nature of the illuminant which hits the reflective subject that matters. In daylight this includes atmospheric transmission effects.

Dave

[Manfred M]

Paul, I don't think is will be quite that easy,

If the camera was held reasonably horizontal either a red or orange filter might be effective. Point the camera up (in relatively clear water) and you'd be too red or orange and point it down and there wouldn't be a lot of effect, I suspect. My orange dry suite looks yellow after getting down some 5 - 10m and gray below 25m.

Up in the northern waters (Ottawa River, for instance) where I dove, the tannins from partially decayed detritus turned the water tea coloured, even at the surface. The St Lawrence River turned quite green from algae during the warm summer months.

Regardless it would be interesting to read to see what his experimentation shows us.

[pnodrog]

Paul, I don't think is will be quite that easy,

If the camera was held reasonably horizontal either a red or orange filter might be effective. Point the camera up (in relatively clear water) and you'd be too red or orange and point it down and there wouldn't be a lot of effect, I suspect. My orange dry suite looks yellow after getting down some 5 - 10m and gray below 25m.

Up in the northern waters (Ottawa River, for instance) where I dove, the tannins from partially decayed detritus turned the water tea coloured, even at the surface. The St Lawrence River turned quite green from algae during the warm summer months.

Regardless it would be interesting to read to see what his experimentation shows us.

Never said it would be easy...

[Polar01]

If you are planning on doing video underwater, suggest you look into the lighting equipment that most cave divers use. Have never dove with anyone shooting video that did not use a lighting rig.

Cheers: Allan

[cbeasley]

Generally it is a good site practice to start a new thread rather than adding to a 10-year old one, so I have established a new thread for your post.

I have not done any underwater photography, but was an active scuba diver for well over a decade with hundreds of dives under my belt, so understand the behaviour of light under water fairly well. The reds disappear first, then the yellows, as one gets deeper. Finally, when we get deep enough, everything appears to be B&W.

Yes, agree. What is amazing is ho much color there is at depth when you take a light source.

[cbeasley]

No, some cameras apply the gain after sampling, sometimes not even in-camera.



An incorrect assumption, sorry.



ISO values are based firmly on illuminance which in turn is based on human luminous efficacy:

https://en.wikipedia.org/wiki/Illuminance

https://en.wikipedia.org/wiki/Luminous_efficacy

Since the color spectrum of human vision is specifically defined, it is inappropriate to discuss the effect of WB or filtering on ISO. That is because ISO is determined under defined lighting conditions and camera settings.

Here's the Japanese specification:

http://kronometric.org/phot/lighting/DC-004_EN.pdf

There is of course an ISO standard but you have to pay to get it.

HTH

Thanks very much for the information and references. They will be interesting reading.

As a practical matter, I'm not sure it is inappropriate to discuss ISO and WB in the context I am considering. Underwater, one haas a very strange light source in terms of what one usually deals with in land based photography. For simplicity, one can assume it is daylight filtered through the water column, which severely alters the spectrum, specifically attenuating reds and other warmer colors, as is well known. Generally, one cannot solve this in post processing only and the best solution is to carry your light source with you. This has it's own set of issues, but at least allows successful post processing under favorable conditions. The same attenuation that affects the sun as a source underwater also applies to artificial lighting, limiting the effective range to a few meters, more or less.

For greater distances, one can use a red filter to tilt the spectrum to something closer to daylight, but obviously, this is done by attenuating the cooler colors, which in turn significantly reduces the available light energy. Longer exposure times are generally not an option, and apertures only go so wide, which brings the ISO setting into the discussion.

The attenuation is significant. In very good conditions -- sunny day, very clear water in the Bahamas -- at a depth of just 10 m, I measured the attenuation to be 3 f-stops. So it is very easy to get into very high ISO situations. Below about 20 m (more or less) red filters are not an option either.

I was hoping that something sophisticated was going on with the underwater WB that might help a little in this regard, but your answers dashed that. It would have been too good to be true anyway. Thanks again!

[cbeasley]

At a guess an appropriate red filter is the way to go provided the additional exposure required is not achieved by increasing the ISO above about ISO 800 depending on sensor design. Beyond about ISO 800 I think that the reduction in dynamic range and the increased noise in the blue channel and to a more limited extent in the green would tend to undermine the benefit of the filtering.

As I said this is a just guess based on my understanding of the factors involved.

Yes, very good guess. Red filters work fairly well -- at least you get something in the red channel if conditions are good enough -- but S/N is the issue. I have recently found a variable red filter that allows you to adjust how strong the filter is. I will be trying it out in a few weeks.

[cbeasley]

If you are planning on doing video underwater, suggest you look into the lighting equipment that most cave divers use. Have never dove with anyone shooting video that did not use a lighting rig.

Cheers: Allan

Yes, I have lighting. I've been shooting stills underwater more than 10 years and video for about 4. The more light the better, but you have size, weight, battery recharge limitations, among other things. I find a practical solution for video is 8K to 10K lumens split between two lights.

[Manfred M]

You are lucky if you dive in relatively clear water to have the light wavelength drop-off as your main issue....

Most of my dives were in the St Lawrence River and Lake Ontario and some other northern rivers and lakes.

With algae and detritus in the water the back-scatter from the particulates in the water made most photography with underwater lights impossible. Large blinding flashes of bright lights did not make for compelling images...

Turning up the camera gain worked with ambient light, but the results were quite noisy. As we were documenting logging equipment and boats for NAS level 2 underwater archeology certification, the information was far more important that the aesthetics of the photos and videos. The tea coloured water of the Ottawa River limited the quality of both stills and video...

[cbeasley]

It occurred to me that my questions have been rather abstract and that you might like to see an example. Here is a link to one of my films "Ghosts of Truk Lagoon" filmed last year on WWII Japanese ships sunk in Truk Lagoon https://www.youtube.com/watch?v=T1LYFPYXhEk . Much of this is filmed between 25 m and 35 m depth and some of it is in the engine rooms where there is no natural light. The large gun at the end (which shows a remarkable amount of color) and the two tanks (one askew atop the other) are on the deck of the San Francisco Maru at a depth of around 55m. You will see a lot of examples where totally gray scenes spring to life in color as lights strike them -- even at these depths. So my main interest is to extend the color zone both laterally and in depth. Because of the depths on these dives, I did not try to use red filters -- too much noise.

[Manfred M]

Thanks for posting. Remarkable vis on those wrecks given number of divers down there. With the amount of finning that the divers in the shot were doing, all the wrecks I've been on would have silted right out.

[cbeasley]

Thanks for posting. Remarkable vis on those wrecks given number of divers down there. With the amount of finning that the divers in the shot were doing, all the wrecks I've been on would have silted right out.

Thanks for watching. A few caveats. Actually, because it was over New Years, the boat was only half full. Usually it has 16 divers, this time, only 9. And, most of them "did their own thing" -- didn't use a guide, so, my wife and I were almost always alone with only the guide! Go figure, but fine with me. The guy you see in front of me is always a guide, and hence very good with the fins. We also have a lot of time in silty conditions (muck diving in Indonesia and the Philippines) so we are pretty careful. It would be very scary and dangerous down in the engine room if someone couldn't control themselves.

Having said all of that, the particulate is mostly rust. I'm not sure it is ever actually totally settled. And you are right, once kicked up, it is a nightmare. Fortunately, the only time that happened was the first day on the bridge of a ship, out in the open, so no danger. The guy that did it got "grounded", since he crashed on the deck, flailed around, etc., so they would not take him inside. Much too dangerous. It was so bad, even out in the open, I had to wait about 4 hours to go back on the last dive of the day to get footage. That scene is near the beginning where I go onto the bridge.

[Manfred M]

Thanks for watching. A few caveats. Actually, because it was over New Years, the boat was only half full. Usually it has 16 divers, this time, only 9. And, most of them "did their own thing" -- didn't use a guide, so, my wife and I were almost always alone with only the guide! Go figure, but fine with me. The guy you see in front of me is always a guide, and hence very good with the fins. We also have a lot of time in silty conditions (muck diving in Indonesia and the Philippines) so we are pretty careful. It would be very scary and dangerous down in the engine room if someone couldn't control themselves.

Having said all of that, the particulate is mostly rust. I'm not sure it is ever actually totally settled. And you are right, once kicked up, it is a nightmare. Fortunately, the only time that happened was the first day on the bridge of a ship, out in the open, so no danger. The guy that did it got "grounded", since he crashed on the deck, flailed around, etc., so they would not take him inside. Much too dangerous. It was so bad, even out in the open, I had to wait about 4 hours to go back on the last dive of the day to get footage. That scene is near the beginning where I go onto the bridge.

Nice to have the guide to yourselves. All the wreck diving I've done is in water where there is either current or sufficient water turnover to clear the silt in reasonably short order, depending on the location. Wooden wrecks have relatively limited rust and most of the steel ships I've been on were in reasonably fast water that swept them fairly clean. Those in or near the St Lawrence Seaway shipping channels were often "flagpole" dives that required a sturdy anchor line to hang onto to overcome the current until one could duck into sheltered parts of the wreck...

All of the penetration diving I have done involved laying of lines with reels to ensure that the divers could exit the wreck safely by touch only. That being said, we've all been on dives with people that should not have been there and did not know how to control their buoyancy.

[cbeasley]

Nice to have the guide to yourselves. All the wreck diving I've done is in water where there is either current or sufficient water turnover to clear the silt in reasonably short order, depending on the location. Wooden wrecks have relatively limited rust and most of the steel ships I've been on were in reasonably fast water that swept them fairly clean. Those in or near the St Lawrence Seaway shipping channels were often "flagpole" dives that required a sturdy anchor line to hang onto to overcome the current until one could duck into sheltered parts of the wreck...

All of the penetration diving I have done involved laying of lines with reels to ensure that the divers could exit the wreck safely by touch only. That being said, we've all been on dives with people that should not have been there and did not know how to control their buoyancy.

Yes, there are certainly safety issues. All of our penetrations were circuits -- no entry/return on the same path. The guide stayed in front and kept an eye on us in case of trouble, but always clear water ahead. Of course, there are still significant safety issues. Suppose someone gets claustrophobic in a stair well and decides they need to retrace, for example? I'm not sure I would have done it with others, but with just the two of us and the guide, I was very wary, but OK with it in the end. There are dives they do with lines, trimix gases, rebreathers, etc., but we are not THAT interested in wreck diving. It is definitely not for everybody and I would recommend people not get beyond their comfort level. Also, they are monitoring your diving skills and won't take you where they feel you are not qualified, but ultimately, that is the diver's responsibility. We are glad we did it and very happy with the video we got, especially in the engine rooms. That early 20th century technology is so cool and elegant. Many of these ships were luxury liners from the 20's converted for merchant marine in the 30's so they were state of the art.