RAW and white balance + red filter

An optical filter has a linear transfer function, because it is a passive device.

 

A linear (filter) transfer function applied to a device with a linear transfer function (CCD) results in a linear output. Period.

Ignoring that passive devices aren't always linear, so what? Linear versus nonlinear has nothing to do with why filters are of value.

Craig, go back to electrical engineering school and study signals & systems.

So you're an electrical engineer now, Dan? A diode is a passive device. Is it linear?

Fact is, Dan, that I do have an electrical engineering degree. I know what transforms and convolutions are. I know, for example, why MTF charts are not all that useful...something you apparently don't based on comments you've made in other threads.

It seems to me that if underwater there is a blue cast the histogram would show the blue channel more to the right and the red channel more to the left. This can be corrected with white balance in your RAW converter. But if a red filter was used to realign the 3 color channels prior to the sensor recording the image, you could expose to the right further (without clipping the blue channel) increase your dynamic range/decrease your shadow noise. It would be an interesting experiment with and without the filter to see if the light loss through the filter negates any advantage to realigning the histogram and ETTR? Not worth an argument though!

Marc

Yes, marc, you have it exactly right.

Craig,

 

What you do not know is how many UW photos from other photographers I have color corrected and printed in a commercial photo lab -- Both in Photoshop as well as in the enlarger.

WHAT?

[editted by me]

Is it April fools day? Ok, you guys got me

I think that people should not forget that a red filter filters out cyan light. It cannot add red. The amount of red light available is fixed at any particular depth (unless you take extra with you!) For this reason red filters can only work well where there is the red part of the spectrum penetrating, where there is plenty of contrast, and where there is enough sensitivity or a wide enough lens aperture to capture it.

 

(Yes, I have been underwater a bit and yes I did a three-year degree course in photography too - albeit before they invented light!)

 

Kurt Amsler said "If you are not shooting in RAW mode, you are missing the point of digital photography." I'm inclined to agree!

Yes, but red filters are not the only ones used underwater. Furthermore, red filters don't just pass red light and red pixels in the camera sensor don't just respond to red light. While it's true that red light is lost fairly quickly underwater, yellow light is not. Yellow light passes through many interesting underwater filters and it stimulates the red sensors in the camera.

Your choice of filter determines the maximum effect on light balance you will receive. You can choose to limit the effect to one stop or use two or even three stops of filtration with a single filter. These devices are not brick walls.

The fact is that ambient photography can produce good color remarkably deep.

Craig, go back to electrical engineering school and study signals & systems.

This board does not appreciate nor need personal comments of this sort of tone.

Followup:

 

If accurate color rendition is needed, shoot a (laminated) Gretag chart in the given lighting conditions.

How will you deal with the unpredicatable effect the lamination causes to the image? The reflectivity in not controllable and effects the result. I know. I've tried. Very high quality lighting is a requirement when calibrating using a ColorChecker. I don't believe that's possible underwater using a laminated chart.

It's not the only way, either. Shoot your rig with a test chart on land, the recalibrate white balance only using a white reference. It's a far more reliable approach than your method. It's what I use.

So, our 12 million photosites are now grouped into 3 million GRGB color picture elements (pixels). Each of these 3 million pixels can have 4096 red, 4096 green, and 4096 blue levels.

Nothing you have said, nor nothing inherent in any of this, limits each photosite to 4096 levels. Furthermore, a Bayer pattern sensor with 12 million photosites does not produce a 3 million pixel image, it produces a 12 million pixel one (except for the Fuji sensor which would produce even more).

But wait a minute! Why not peel the 3 million pixels of raw data off the CCD and write it directly to the flash memory... And bypass the in-camera processing directly? Now, we have 3 million red, 3 million green, and 3 million blue chunks of data, each of which have a value from 0 to 4095. Each and every one of these 9 million chunks of data is individually accessable in Photoshop.

A 12MP Bayer sensor will have 3M red, 6M green, and 3M blue pixels.

Getting back to the subject of using a filter in front of the camera lens when shooting RAW, since the raw pixel data is in linear form, it makes little to no difference if an additional linear filter is placed in front of the camera lens to "tune" the color: The same linear filter can be applied in Photoshop!

No it can't. Not even remotely.

Light entering the lens and exposing the sensor is full spectrum. Edits made in Photoshop apply to a tristimulous model of color which is a fairly inept, but visually sufficient, model of the real thing. Filters that operate on real light are far more powerful than digital manipulations. Furthermore, camera sensors have limited dynamic range. Filters correct light imbalances and therefore enable you to improve chroma noise because you are utilizing more of the dynamic range of each photosite. Note that this has utterly nothing to do with the linear characteristics of a digital sensor.

Furthermore, this filter that "tunes" the color extracts a toll in that it compromises the exposure due to light loss...

Wrong thinking, Dan. Filters work by removing undesirable light. Therefore, the light they remove is that which prevents a good exposure that would maximize use of all the photosites. The correct exposure is one where there is an ideal white balance. Therefore, it is not the filter ruining the exposure, but the lack of a filter that ruins it.

If you take the arbitrary example of a two stop correcting filter, you may get a 2 stop faster exposure without the filter, but the price you pay is a two stop degradation in chroma dynamic range. If you are extremely limited in light, that may be a good tradeoff, but in general, the better exposure is the one with the filter.

And sometimes the light is coming through a neutral density + high-pass (blue) filter... The water. See figure 1.1a here; or the visible range slice of the absorption coefficient chart near the bottom of this document.

Dan, these characteristics of water are nearly useless. The water we dive in has dissolved and suspended inpurities that substantially alter the color balance we see. The filtration characteristics of water vary continuously.

Yes, using a complementary low-pass or band-pass filter will yield a flatter transfer function when the transfer functions are convolved... With a lot of attenuation that has to be compensated for.

Determining the effects of a filter on color balance does not involve a convolution, and being a stage hand on a TV set 30 years ago doesn't make you an engineer. Typical filters used underwater have less than 3 stops of exposure effect. I don't consider that "a lot of attenuation".

It's that attenuation -- And the compensation on a real world thumbnail-sized CCD - I am (trying to) address here: It's not only the light loss an additional (red) filter adds into the path, but also the light loss (the "neutral density" component) that has to be accounted for... And compensated for.

Whoop-dee-do. That attenuation is your friend, because that attenuation is what's fixing your color problems.

In the case of a Very Quiet CCD, such as that found on the Canon 5D, then the trade-off of using a filter when shooting RAW can be helpful, as image degredation from increasing the ISO is not that great.

I NEVER adjust ISO because I have filters attached, Dan. Here's your problem. You don't know what filters we are talking about, you don't know their strengths or effects, and you've never taken a shot underwater. Perhaps you might try these things out before preaching on why they aren't useful.

I could explain why I don't even adjust exposure for my filters (much less ISO) but I don't think you'll understand. Filters produce a different image. Exposure doesn't need to stay the same.

On the other hand, with a CCD having very tiny photosites, fewer photons are "snagged" due to the smaller cross-sectional area, hence one must reduce any losses in the light path.

But these very same small sensors require shorter focal lengths for equivalent perspective. This results in a much faster lens for a given DOF. These types of cameras have very fast lenses compared to DSLRs, and because of that they are also tolerant of filtration. They have less dynamic range but filters enable that range to be more effectively utilized, not less as you suggest.

It is for this reason that, given the camera's 12 million pixel 1/1.7" CCD .AND. the ability to directly record the CCD data in raw form .AND. the low light conditions underwater, that I recommend omitting the optical filter, and instead apply it in software in Photoshop.

When you gain some experience then you'll have something to base opinions on rather than poorly formed theories.

P.S.

Dan, it seems that Alex Mustard is a photographer that you respect (as we all do). If you don't know, Alex not only uses and advocates the use of filters, he has formulated them and formed a company specifically to make suitable filters available to us. This is certainly not any appeal to authority on my part since I can hold my own in these discussions, but you may want to rethink offering filter advise based on zero experience.

Peace be with you all. (ISO 1600)

Actually, I detect some distinct colour changes (reduction of red) to images I post here as examples between seeing it on my calibrated monitor and then again on the same monitor on this website.

Cutting through the crap and attempting to give JKolgen a simple answer, I would say shoot in RAW, forget the filter thereby getting a faster shutter speed, get CS3 RAW converter - and calibrate your monitor and printer! So many underwater pictures I see submitted (to Eaton Publications for example) suffer from camera movement.

On the other hand, if you are putting the filter on and then white balancing, you're not accomplishinganything, as you are "correcting the corrections."

I'm very glad to see others jumping all over this comment. In my limited experience and reading, this seemed totally wrong to me, and made me sincerely doubt that anything I thought I had 'learned' was correct. But seeing the responses, I am reassured that I am indeed getting the gist of it, and reassured that I need not claim to be a genius to eventually 'get it' and hopefully take some nice shots in time.

I would like to thank all for this lively and passionate discussion. And a special thanks for not censoring any responses. I hope, except for profanity, that censorship never rears its ugly head on this board. Fear not, those of us reading this thread are very capable of filtering out the chaf from the wheat.

Safe diving,

imasleeper

I would like to thank all for this lively and passionate discussion. And a special thanks for not censoring any responses.

Indeed, just read through this debate. All posts are valuable as part of the peer review process.

Although I've got reasonable kit, still consider myself as a happy holiday snapper rather than an underwater photographer. Understanding each argument in here has helped my thoughts a lot. For what its worth I am tending towards Marc's succinct post regarding aligning the colour histograms.

Now, certainly my composition and technique leaves a lot to be desired but many of my snaps include strobe lit foreground and natural light background. How does one overcome the foreground colour shift. Is anyone using complimentary colour strobe filtration.

For what its worth I am tending towards Marc's succinct post regarding aligning the colour histograms.

Yes, what Marc said regarding the trend in the color histograms is true, but color histograms on the preview screen are influenced by the white balance setting. The true value of a filter could be obscured on the color histogram preview as a result but iIt will not be in the RAW converter.

Now, certainly my composition and technique leaves a lot to be desired but many of my snaps include strobe lit foreground and natural light background. How does one overcome the foreground colour shift. Is anyone using complimentary colour strobe filtration.

You overcome that will proper strobe filter selection and good white balance and, of course, there are those of us using complimentary strobe filters. :-) Not generally with a warming filter, though, but it could be done.

Darn, I missed this thread while I was in Korea.

John: I'm pleased to see that my talk at Visions wasn't totally useless to you after-all. It looks like my bit about adjusting non-filtered shots using the HSL settings in CS3 Raw converter was useful :-) Hopefully your readers will appreciate it as well.

All of my recent dives have been done using a CC30M filter on the lens and CC30G filter on my strobes, so I have some recent experience with this method.

Firstly, I was surprised at how good the camera was at getting the auto whitebalance right (I shot raw + JPEG).

Secondly, if they water you're diving in is slightly cyan or low viz, the results w/ the filter will be dramatically better!

Here are a few examples:

Shot in 80' viz w/ a bit of cyan tinge to it.

Shot is 50' viz w/ a very cyan tinge to it.

Cheers

James

Hi James!

Your shot of the `ray with your CC lens filter & strobe gel shows something interesting: Notice the tint of the sunball compared to the white belly of the `ray.

Anyway, nice shot!

Darn, I missed this thread while I was in Korea.

 

(cut)

 

All of my recent dives have been done using a CC30M filter on the lens and CC30G filter on my strobes, so I have some recent experience with this method.

 

Firstly, I was surprised at how good the camera was at getting the auto whitebalance right (I shot raw + JPEG).

 

Secondly, if they water you're diving in is slightly cyan or low viz, the results w/ the filter will be dramatically better!

 

Here are a few examples:

 

Shot in 80' viz w/ a bit of cyan tinge to it.

(cut)

Hi all

I just thought I might as well make my first post on this thread, out of date though it may be,

For anyone just stumbling onto this thread through google as I did, lets put this into perpective,

I dont have an electrical engineering degree (or any other for that matter),

My diving experience and as such my underwater photography experience is limited, I am very enthusiastically starting out in both things,

My general photography experience on the other hand is far from limited,

So what I do know is photography is about 'taking photographs', not spending hours hunched over a computer,

It is an optical art-form first, and digital science second,

If my shot needs anything more then a straight conversion from RAW, into a TIFF or other printable format, then its not good enough,

An already excellent shot may be converted to monochrome, if it serves my artistic goals, an otherwise excellent and well lit subjet may just need a slight sharpening to bring out the best in it, a portrait may just need a couple of spots cloned out, but 99% of the time if I didnt get it right in camera it goes in the bin,

In two weeks I am going to Austrailia, to shoot both on land and underwater,

I will be using a red fliter on my camera for the later, correcting my white balance before every shot,

And getting closer to my subjects then ever with my brand new inon 105 ,

If I get just one good shot, it will be better then the horrible frame after frame of blue or ugly false red channel noise I ended up with on my last trip,

In short my underwater photography will evolve taking photographs underwater, not on my computer,

Just my 2 pennys worth