Do optical WB filters improve RAW color latititude?

Started Aug 26, 2012 | Discussions
Mako2011
Forum ProPosts: 15,391
Like?
Not just higher ISO
In reply to John Sheehy, Aug 27, 2012

John Sheehy wrote:

Mako2011 wrote:

Actually you can do it much cleaner and better with Software digitally now than with optical filters. Read the article I linked then try it with today's products. It really is better to do it digitally..... except perhaps in very specific cases when long exposure at base ISO is possible.

I already said that color filters can be counter-productive if they force a higher ISO.

Compared to using modern digital software (less than one year old software solutions), color filters are always the less optimal choice regards final IQ. The only exception might be certain specific situations of very long exposure at base ISO. Lab testing of optics/lighting come to mind.

Reply   Reply with quote   Complain
Jirka-
Regular MemberPosts: 497
Like?
Re: Not just higher ISO
In reply to Mako2011, Aug 27, 2012

No one mentions the fact that old colour correction filters originally intended for slide film can damage the image by ghosting and fogging because they are flat and often have no MC coating.

Slide film has no exposition latitude - the difference between a perfect exposure and exposure unusable for any serious purpose can be just a half f-stop.

You did not usually need any correction filters except an UV filter for a negative film. Contemporary digital sensors are more similar to a negative film than slide film in terms of DR at low ISO in RAW.

Maybe in some rare circumstances you can get slightly better colours with the filters - but is it worth the trouble and the risk of deterioration of the image quality by a less than perfect optical element?

Reply   Reply with quote   Complain
Mako2011
Forum ProPosts: 15,391
Like?
Re: Not just higher ISO
In reply to Jirka-, Aug 27, 2012

Jirka- wrote:

No one mentions the fact that old colour correction filters originally intended for slide film can damage the image by ghosting and fogging because they are flat and often have no MC coating.

Slide film has no exposition latitude - the difference between a perfect exposure and exposure unusable for any serious purpose can be just a half f-stop.

You did not usually need any correction filters except an UV filter for a negative film. Contemporary digital sensors are more similar to a negative film than slide film in terms of DR at low ISO in RAW.

Maybe in some rare circumstances you can get slightly better colours with the filters - but is it worth the trouble and the risk of deterioration of the image quality by a less than perfect optical element?

I would agree and I don't think it's any longer possible to macth what current generation software can do with a color correction filter of old. (except in extremely rare and specific setups)

Reply   Reply with quote   Complain
John Sheehy
Forum ProPosts: 16,019
Like?
Re: Not just higher ISO
In reply to Mako2011, Aug 27, 2012

Mako2011 wrote:

John Sheehy wrote:

I already said that color filters can be counter-productive if they force a higher ISO.

Compared to using modern digital software (less than one year old software solutions), color filters are always the less optimal choice regards final IQ. The only exception might be certain specific situations of very long exposure at base ISO. Lab testing of optics/lighting come to mind.

No "modern digital software" can change the SNR of the capture and digitization.

If you shoot at base ISO, even for a short exposure, you are going to have more noise in the red and/or blue channel than if you had changed the color of the light hitting the sensor to perform the WB. You magical "modern digital software" can only emulate better SNR by removing detail.

-- hide signature --

John

Reply   Reply with quote   Complain
Mako2011
Forum ProPosts: 15,391
Like?
Yes and things changed
In reply to John Sheehy, Aug 27, 2012

John Sheehy wrote:

Mako2011 wrote:

John Sheehy wrote:

I already said that color filters can be counter-productive if they force a higher ISO.

Compared to using modern digital software (less than one year old software solutions), color filters are always the less optimal choice regards final IQ. The only exception might be certain specific situations of very long exposure at base ISO. Lab testing of optics/lighting come to mind.

No "modern digital software" can change the SNR of the capture and digitization.

Correct

If you shoot at base ISO, even for a short exposure, you are going to have more noise in the red and/or blue channel than if you had changed the color of the light hitting the sensor to perform the WB. You magical "modern digital software" can only emulate better SNR by removing detail.

And the filter can only keep light from hitting the sensor.

Bottom line, the final product looks better when color effects are done digital using current gen software than when done using color effect filters. Software has now done to color effect filters what current gen sensors have done to film. That leap came in the last 12 months really with upgrades to Photoshop, Topaz, and Nik. There really is no need for color effect filters outside a lab environment or when still using film.

Reply   Reply with quote   Complain
Vitruvius
Junior MemberPosts: 36Gear list
Like?
Re: Note...no need
In reply to John Sheehy, Aug 27, 2012

Thanks.

 Vitruvius's gear list:Vitruvius's gear list
Sony SLT-A77 Sony a77 II
Reply   Reply with quote   Complain
Vitruvius
Junior MemberPosts: 36Gear list
Like?
Re: Do optical WB filters improve RAW color latititude?
In reply to Vitruvius, Aug 27, 2012

Thankyou for all the info. I had another thought. With the really high burst rate of the new cameras, what if I shot and Exposure Bracketed in RAW and then did the merge to 32 bit HDR in Photoshop. Then adjust WB and compress down to 16 bit again. I think this would allow the most flexibility without loss of any image quality. Assuming of course that bracketing is an option for that shot.

 Vitruvius's gear list:Vitruvius's gear list
Sony SLT-A77 Sony a77 II
Reply   Reply with quote   Complain
Mako2011
Forum ProPosts: 15,391
Like?
Why
In reply to Vitruvius, Aug 27, 2012

Vitruvius wrote:

Thankyou for all the info. I had another thought. With the really high burst rate of the new cameras, what if I shot and Exposure Bracketed in RAW and then did the merge to 32 bit HDR in Photoshop. Then adjust WB and compress down to 16 bit again. I think this would allow the most flexibility without loss of any image quality. Assuming of course that bracketing is an option for that shot.

Why do that? The camera already captures in 14bit RAW. Nothing gained going to a 32bit merge. Also, HDR is an entirely different effect/process from color correction. What is your real output goal? If it is capture color information....an HDR workflow will look inaccurate.

Reply   Reply with quote   Complain
SteveJL
Contributing MemberPosts: 882
Like?
Re: Do optical WB filters improve RAW color latititude?
In reply to Vitruvius, Aug 28, 2012

I'm thinking that you mean WB filters like the Expodisc, which, truthfully does it's job VERY well in most conditions. OTOH, so does what is available in PP using RAW image, so it becomes a personal choice. Myself, I give the nod to the Expodisc because, assuming one does not have Photoshop, it is cost-effective as well as accurate.

Reply   Reply with quote   Complain
Vitruvius
Junior MemberPosts: 36Gear list
Like?
Re: Why
In reply to Mako2011, Aug 28, 2012

You are right, sorry, I actually meant WB bracketing, not exposure bracketing.

 Vitruvius's gear list:Vitruvius's gear list
Sony SLT-A77 Sony a77 II
Reply   Reply with quote   Complain
Barrie Davis
Forum ProPosts: 21,460
Like?
Expodisc not part of this discussion.
In reply to SteveJL, Aug 28, 2012

SteveJL wrote:

I'm thinking that you mean WB filters like the Expodisc, which, truthfully does it's job VERY well in most conditions. OTOH, so does what is available in PP using RAW image, so it becomes a personal choice. Myself, I give the nod to the Expodisc because, assuming one does not have Photoshop, it is cost-effective as well as accurate.

The OP was perfectly clear in his enquiry, which was about potential improved results of using 80B colour conversion filters when conducting digital photography in tungsten light..

Expodisc therefore not part of this discussion.

Note: Expodisc originally designed as Incident Light metering attachment for TTL cameras. WB function was an afterthought capitalised after arrival of digital.
--
Regards,
Baz

"Ahh... But the thing is, these guys were no ORDINARY time travellers!"

Reply   Reply with quote   Complain
hjulenissen
Senior MemberPosts: 1,604
Like?
Re: Do optical WB filters improve RAW color latititude?
In reply to Vitruvius, Aug 28, 2012

Any filter will cause less light to hit the sensor (in a given time). If that is your main problem, filters does not seem to be the solution.

Any filter introduced increase the chance of loss of sharpness and glare.

A filter with a strong color cast will affect the relative amount of light that each sensor channel gets (within a given time). If you are close to the clipping point of your red channel, and the blue channel is seriously underexposed ("noisy"), a filter could "equalise" the channel exposure, letting you increase exposure time to get a better recording of blue without clipping red (or vice versa). On the other hand: highlight recovery can generate a decent image if only one channel is clipped, so uneven channel exposure could be exploited to capture more scene dynamic range.

Colored filters for visual effect seems largely replaced by Photoshop these days, but if you have very specific requirements (narrow peaks/dips, infrared/ultra-violet behaviour), physical filters can do things that cannot be done digitally (and vice versa).

I do not see the point of color filters for my kind of photography. I have enough stuff to worry about that seems more important (gelling any flashes, getting exposure right, finding interesting scenes...)

-h

Reply   Reply with quote   Complain
hjulenissen
Senior MemberPosts: 1,604
Like?
Re: Why
In reply to Mako2011, Aug 28, 2012

Mako2011 wrote:

What is your real output goal? If it is capture color information....an HDR workflow will look inaccurate.

I do believe that N exposure bracketed images of a scene will always include as much (usually more) information about the scene that a single exposure will.

I don't understand why "HDR" necessarily need to have inaccurate colors. I do agree that many (most?) tone-mapping tools render colors in a perceptually innaccurate way.

-h

Reply   Reply with quote   Complain
Mako2011
Forum ProPosts: 15,391
Like?
See
In reply to hjulenissen, Aug 28, 2012

hjulenissen wrote:

Mako2011 wrote:

What is your real output goal? If it is capture color information....an HDR workflow will look inaccurate.

I do believe that N exposure bracketed images of a scene will always include as much (usually more) information about the scene that a single exposure will.

Yes, but after processing the exposure with HDR software...The DR is compressed and made to simulate the real DR

I don't understand why "HDR" necessarily need to have inaccurate colors. I do agree that many (most?) tone-mapping tools render colors in a perceptually innaccurate way.

And the colors will shift/compress from the real scene.

see:

Reply   Reply with quote   Complain
Vitruvius
Junior MemberPosts: 36Gear list
Like?
Re: Do optical WB filters improve RAW color latititude?
In reply to hjulenissen, Aug 28, 2012

Yes, yes, and yes. I understand all this... Filters reduce incoming light in specific color channels. Makes total sense, since that is the point of a color filter.

But the point is that you are also reducing the amount of light recorded in specific channels and/or boosting the amount of light in other channels post exposure in Photoshop to achieve the SAME result. Once the exposure is taken there is a limit to the DR of each color channel. The further off your white balance was during exposure the more you will need to Push or Subdue the (limited) amount of light in each channel.

So wouldn't it be better to optically correct for WB during exposure and increase the general exposure time a bit. That way the WB is closer to what you want PP.

Everyone knows that you need to get your general exposure as close as possible during exposure because you can't get 'blood from a rock' later. This question really is about getting the exposure as close as possible PER COLOR CHANNEL during exposure.

 Vitruvius's gear list:Vitruvius's gear list
Sony SLT-A77 Sony a77 II
Reply   Reply with quote   Complain
alanr0
Senior MemberPosts: 1,220
Like?
Re: Do optical WB filters improve RAW color latititude?
In reply to Vitruvius, Aug 29, 2012

Vitruvius wrote:

Yes, yes, and yes. I understand all this... Filters reduce incoming light in specific color channels. Makes total sense, since that is the point of a color filter.

But the point is that you are also reducing the amount of light recorded in specific channels and/or boosting the amount of light in other channels post exposure in Photoshop to achieve the SAME result. Once the exposure is taken there is a limit to the DR of each color channel. The further off your white balance was during exposure the more you will need to Push or Subdue the (limited) amount of light in each channel.

So wouldn't it be better to optically correct for WB during exposure and increase the general exposure time a bit. That way the WB is closer to what you want PP.

Everyone knows that you need to get your general exposure as close as possible during exposure because you can't get 'blood from a rock' later. This question really is about getting the exposure as close as possible PER COLOR CHANNEL during exposure.

Although Mako2011's contributions are too didactic for my taste, the contributions are not completely contradictory.

You have folk (John Sheehy, myself, etc.) who broadly support your statements quoted above. Maximising the signal in each channel (at low ISO) will improve the signal to noise ratio. Provided you can increase the exposure time, a colour balancing filter can increase the dynamic range available for subsequent post processing.

You have folk like Barry Davies and Olaf Ulrich adding their practical experience, with careful explanation of the conditions under which their observations are valid.

You have Mako2011 quoting Luminous Landscapes and stating that optical filters are practically never worth using.

In terms of your original question, I believe that balancing filters can increase the dynamic range available. I also agree with Mako2011 that sensor performance has improved considerably since Josef Wisniewski's 2008 post. With a state of the art camera, there may be fewer occasions where you will see a substantial improvement using optical filters.

The latest digital filter products may be wonderful, but I am sure they are equally wonderful working on a lower noise image from a colour balanced exposure.

However, bear in mind Jirka-'s point about reflections and flare introduced by filters with poor quality anti-reflection coatings. The benefits of colour balancing are most apparent in high contrast scenes, and these are particularly susceptible to degradation by flare.

How about trying some test shots with an incandescent light or candle in the frame to create a high contrast scene, and see how well this pans out in practice?

Hope this helps
--
Alan Robinson

Reply   Reply with quote   Complain
Mako2011
Forum ProPosts: 15,391
Like?
No longer
In reply to Vitruvius, Aug 29, 2012

Vitruvius wrote:

Yes, yes, and yes. I understand all this... Filters reduce incoming light in specific color channels. Makes total sense, since that is the point of a color filter.

But the point is that you are also reducing the amount of light recorded in specific channels and/or boosting the amount of light in other channels post exposure in Photoshop to achieve the SAME result. Once the exposure is taken there is a limit to the DR of each color channel. The further off your white balance was during exposure the more you will need to Push or Subdue the (limited) amount of light in each channel.

So wouldn't it be better to optically correct for WB during exposure and increase the general exposure time a bit. That way the WB is closer to what you want PP.

No, as pointed out in examples here. The best you can hope for is an equal outcome vs PP. In most cases though, the better final product is achieved digitally.

Everyone knows that you need to get your general exposure as close as possible during exposure because you can't get 'blood from a rock' later. This question really is about getting the exposure as close as possible PER COLOR CHANNEL during exposure.

Using color effect filters just doesn't help in that regard vs other modern methods.

Reply   Reply with quote   Complain
Mako2011
Forum ProPosts: 15,391
Like?
nice
In reply to alanr0, Aug 29, 2012

alanr0 wrote:

Vitruvius wrote:

Yes, yes, and yes. I understand all this... Filters reduce incoming light in specific color channels. Makes total sense, since that is the point of a color filter.

But the point is that you are also reducing the amount of light recorded in specific channels and/or boosting the amount of light in other channels post exposure in Photoshop to achieve the SAME result. Once the exposure is taken there is a limit to the DR of each color channel. The further off your white balance was during exposure the more you will need to Push or Subdue the (limited) amount of light in each channel.

So wouldn't it be better to optically correct for WB during exposure and increase the general exposure time a bit. That way the WB is closer to what you want PP.

Everyone knows that you need to get your general exposure as close as possible during exposure because you can't get 'blood from a rock' later. This question really is about getting the exposure as close as possible PER COLOR CHANNEL during exposure.

Although Mako2011's contributions are too didactic for my taste, the contributions are not completely contradictory.

You have folk (John Sheehy, myself, etc.) who broadly support your statements quoted above. Maximising the signal in each channel (at low ISO) will improve the signal to noise ratio. Provided you can increase the exposure time, a colour balancing filter can increase the dynamic range available for subsequent post processing.

You have folk like Barry Davies and Olaf Ulrich adding their practical experience, with careful explanation of the conditions under which their observations are valid.

You have Mako2011 quoting Luminous Landscapes and stating that optical filters are practically never worth using.

In terms of your original question, I believe that balancing filters can increase the dynamic range available. I also agree with Mako2011 that sensor performance has improved considerably since Josef Wisniewski's 2008 post. With a state of the art camera, there may be fewer occasions where you will see a substantial improvement using optical filters.

The latest digital filter products may be wonderful, but I am sure they are equally wonderful working on a lower noise image from a colour balanced exposure.

However, bear in mind Jirka-'s point about reflections and flare introduced by filters with poor quality anti-reflection coatings. The benefits of colour balancing are most apparent in high contrast scenes, and these are particularly susceptible to degradation by flare.

How about trying some test shots with an incandescent light or candle in the frame to create a high contrast scene, and see how well this pans out in practice?

very well articulated.

It would be nice to have both 14-bit raw files from a modern current gen camera...one with filter and one without to use/examine.

Reply   Reply with quote   Complain
ZOIP
Junior MemberPosts: 35
Like?
Re: Do optical WB filters improve RAW color latititude?
In reply to Vitruvius, Aug 29, 2012

Hi there Vitruvius

Your question is actually quite significant but the answer is more complicated than I feel most of the respondents appreciate.

The short answer is absolutely and you can gain a whole lot more than just extended colour balance options.

Up front I have to say I am only going to provide some limited information on this but it should help. Please bear in mind I have actually tested all of this over a 3 year period with at least 5000 frames, it is in fact part of a whole imaging system I teach in Australia (the workshop manual is almost 400 pages) and I regularly shoot using pre-filtering as a standard process. I have just returned from a 6 week US/Canada holiday where I captured at least 2000 frames using pre-filtering under all sorts of lighting conditions and I use it on paying jobs regularly.

Many posters have raised valid points but I think it is fair to say most are talking in terms of theoretical and accepted wisdom rather than actually having done it.

The negatives:

Yes the exposure will be longer, effectively it is like shooting at about 32 ISO instead of 100, but for high quality work using tripods it is irrelevant. There is little point doing fully sensor balanced capture if you are going to compromise somewhere else in the chain. Under daylight conditions any increased noise from longer exposures is insignificant, you're still typically in the 1/60-1/500th range.

Yes you can get flare from the filters but it is rarely an issue and Gel CC filters detract zero clarity or if they do the gains that can be made far outweigh them. In my implementations the filters are behind the lens where ever possible, this radically reduces flare, and in fact of the 2000 images shot on my last holiday not one has any significant flare issue.

Raw conversations are some what harder due to the radically different colour temp corrections needed but most can handle it OK.

You do not achieve the same result using software after the fact but it can be close if the shooting circumstances were not too challenging, in other words normal contrast, average white balance issues.

Regardless of how good the sensor is you will reap gains, they not be quite so significant with the best new sensors......BTW I have used this with what is probably one of the best sensors out there, the 16 Meg NEX 5N and yes the files are better.

What you really need to know!

To obtain the full benefit of pre-filtered or "sensor balanced" shooting you need to rethink your whole process of capture, raw conversion and editing. It offers entry to a pathway of improved image quality that will result from a myriad of small but significant changes. In truth few photographers are prepared to apply the rigour needed or take the time to optimise processes, it is not simply a matter of throwing a filter on the front of the lens.

What Is Possible?

Balanced clipping points will largely eliminate weird and difficult to fix colour shifts in upper highlights and lower shadows.

Noise is luminance in nature, chrominance noise can be greatly reduced or eliminated.

Image are subtly sharper and will withstand higher levels of sharpening

Colour quality is subtly better, more natural in effect

Mono conversations can be radically better as the three channels all have near identical noise signatures meaning conversions are more film like.

Extended dynamic range overall, in fact fully sensor balanced files look flat compared to regular ones before conversion.

And on it goes..........

NOW the issue of tungsten is a bit messy in particular because the native sensitivity across the three channels is actually not correct for daylight balance so the filters needed are nothing like what is needed for shooting daylight film with tungsten. In fact the increased red output of tungsten light sources actually helps get the red channel back in kilter quite a bit, believe it not tungsten colour temperature is not that far removed from the native sensitivity of most sensors across the three channels. There is not pre-potted solution for the correct filter pack, you have to test.

One of the biggest reasons people get poor tungsten images in general is that the cameras histogram and metering lie to you big time in this instance and most times that actual files are very underexposed, which is easy to see when you open them in raw and analyse what it actually captured across the three channels.

So before you get too carried away with filters just try shooting with tungsten but over-expose the RAW files a bit and see how the converted files look, you may be surprised. JPEGs shoot under tungsten are as far as I am concerned pretty much useless so lets not even go there.

By all means email me if you want more information.

Trying to make the complex simple

Reply   Reply with quote   Complain
Mako2011
Forum ProPosts: 15,391
Like?
Good answer
In reply to ZOIP, Aug 29, 2012

ZOIP wrote:

Hi there Vitruvius

Excellent post!

Your question is actually quite significant but the answer is more complicated than I feel most of the respondents appreciate.

The short answer is absolutely and you can gain a whole lot more than just extended colour balance options.

Up front I have to say I am only going to provide some limited information on this but it should help. Please bear in mind I have actually tested all of this over a 3 year period with at least 5000 frames, it is in fact part of a whole imaging system I teach in Australia (the workshop manual is almost 400 pages) and I regularly shoot using pre-filtering as a standard process. I have just returned from a 6 week US/Canada holiday where I captured at least 2000 frames using pre-filtering under all sorts of lighting conditions and I use it on paying jobs regularly.

Many posters have raised valid points but I think it is fair to say most are talking in terms of theoretical and accepted wisdom rather than actually having done it.

The negatives:

Yes the exposure will be longer, effectively it is like shooting at about 32 ISO instead of 100, but for high quality work using tripods it is irrelevant. There is little point doing fully sensor balanced capture if you are going to compromise somewhere else in the chain. Under daylight conditions any increased noise from longer exposures is insignificant, you're still typically in the 1/60-1/500th range.

Yes you can get flare from the filters but it is rarely an issue and Gel CC filters detract zero clarity or if they do the gains that can be made far outweigh them. In my implementations the filters are behind the lens where ever possible, this radically reduces flare, and in fact of the 2000 images shot on my last holiday not one has any significant flare issue.

Raw conversations are some what harder due to the radically different colour temp corrections needed but most can handle it OK.

You do not achieve the same result using software after the fact but it can be close if the shooting circumstances were not too challenging, in other words normal contrast, average white balance issues.

What software did you use to compare? I think this is key to results/comparison as significant strides have been made just recently regards color correction.

Regardless of how good the sensor is you will reap gains, they not be quite so significant with the best new sensors......BTW I have used this with what is probably one of the best sensors out there, the 16 Meg NEX 5N and yes the files are better.

What you really need to know!

To obtain the full benefit of pre-filtered or "sensor balanced" shooting you need to rethink your whole process of capture, raw conversion and editing. It offers entry to a pathway of improved image quality that will result from a myriad of small but significant changes. In truth few photographers are prepared to apply the rigour needed or take the time to optimise processes, it is not simply a matter of throwing a filter on the front of the lens.

Agreed and why I see that as specialized and perhaps outside the abilities of most while the digital alternative is not.

What Is Possible?

Balanced clipping points will largely eliminate weird and difficult to fix colour shifts in upper highlights and lower shadows.

Noise is luminance in nature, chrominance noise can be greatly reduced or eliminated.

Image are subtly sharper and will withstand higher levels of sharpening

Colour quality is subtly better, more natural in effect

Mono conversations can be radically better as the three channels all have near identical noise signatures meaning conversions are more film like.

Extended dynamic range overall, in fact fully sensor balanced files look flat compared to regular ones before conversion.

And on it goes..........

NOW the issue of tungsten is a bit messy in particular because the native sensitivity across the three channels is actually not correct for daylight balance so the filters needed are nothing like what is needed for shooting daylight film with tungsten. In fact the increased red output of tungsten light sources actually helps get the red channel back in kilter quite a bit, believe it not tungsten colour temperature is not that far removed from the native sensitivity of most sensors across the three channels. There is not pre-potted solution for the correct filter pack, you have to test.

One of the biggest reasons people get poor tungsten images in general is that the cameras histogram and metering lie to you big time in this instance and most times that actual files are very underexposed, which is easy to see when you open them in raw and analyse what it actually captured across the three channels.

So before you get too carried away with filters just try shooting with tungsten but over-expose the RAW files a bit and see how the converted files look, you may be surprised. JPEGs shoot under tungsten are as far as I am concerned pretty much useless so lets not even go there.

By all means email me if you want more information.

Again, well articulated.

Trying to make the complex simple

Hard to do but you seem to have found a way

Reply   Reply with quote   Complain
Keyboard shortcuts:
FForum MMy threads