Jump to content
craig

Hardware AA and UV-IR cut filters

Recommended Posts

Should underwater photographers consider modifying their cameras to remove the hardware AA and UV-IR cut filters?

 

Digital SLRs almost universally have two filters that are stacked immediately on top of their sensors. One is the anti-aliasing (AA) filter and the other is the UV-IR cut filter, or ICF. While each of these filters performs critical functions, each also reduces image quality significantly. For underwater photography, an interesting question is whether we might be better off with one or both of these filters removed.

 

The Anti-Aliasing filter:

 

The AA filter is necessary to prevent image detail too fine for the sensor from causing distortion in the form of aliasing artifacts. In bayer pattern sensors (all digital cameras other than Foveon), aliasing is particularly noticable in the form of color artifacts, or moire. While moire can often be detected and suppressed in software, not all aliasing errors show up as moire.

 

AA filters vary in strength depending on the tradeoffs the manufacturer decides to accept. Camera makers desire to deliver the best image quality possible but "best" is very much dependent on the intended use of the camera. It's possible for the manufacturer to use an AA filter sufficient to suppress all aliasing but doing so is not always the best choice. The Nikon D100 was notorious for the strength of its AA filter and the resulting softness of its images. The Kodak SLR/n and 14n had no AA filter at all and were quite sharp. Most recently, there has been discussion regarding the Canon 1D3 and the Nikon D3. Which camera offers better image quality is a matter of debate since the D3, in spite of its extra pixel count, includes a stronger AA filter. Which is preferable depends on who you ask and how they work.

 

But do we need an AA filter underwater at all? For large pixel cameras like the 5D, 1Ds (Mark 1), and the D3/D700 maybe, but for most current cameras we may well not. First off, it's well known that today's cameras often are limited by available lenses and very few lenses are of interest underwater. If the lens can't deliver fine enough detail we don't need an AA filter to remove any! There are a number of underwater-specific issues to consider as well. Underwater systems have ports that damage sharpness and have several inches to several feet of water to shoot to through that seriously limit resolution. For wide angle, image quality is seriously hampered by the performance of our dome ports and some shooters are being driven shoot at very small apertures to maintain image quality. For macro, small apertures are used as a matter of course. While there is a separate debate on just what the term "diffraction limited" really means, current cameras become noticably affected by diffraction anywhere from f/8 to f/16 depending on camera. This doesn't mean small apertures aren't useful, but it does mean that by f/11 or f/16, diffraction renders the AA filter moot. If all your underwater shooting is above f/11 chances are you never need an AA filter. Basically, the AA filter is robbing us of image quality in return for nothing. Even with wide angle at f/8 or larger, how much ultrafine detail are we going to get through feet of water column?

 

The UV-IR cut filter (ICF):

 

The ICF is necessary because sensors respond to ultraviolet and infrared light and image quality will be degraded if they are allowed through. There are two kinds of ICFs, absorptive and reflective. DSLRs use absorptive filters which are made using dyes and have a cyan appearance. Like AA filters, not all camera use ICFs of the same strength, so some cameras are vulnerable to image degradation due to a weak ICF. Another way to look at it is some cameras are better for infrared photography than others. :)

 

So how does the ICF contribute to image degradation? The ICF causes an imbalance in sensitivity to certain colors and, as a result, DSLRs are typically much more sensitive to green light than to blue and red. How bad? Look at these tables to see how far off some gain coefficients can be. You will notice that you will never see a gain of 1,1 which is what we always want!

 

Readers of the dpreview and luminous landscape forums may be familiar with UniWB and the technique of using magenta filters while shooting. There are a number of experienced pro shooters who advocate using a CC40M filter or stronger at all times while shooting a DSLR. This is a direct consequence of the absorptive ICF in the camera. Most of us are familiar with the concept of "expose to the right" (ETTR). The net effect of a typical ICF is to force all our underwater shots to underexpose our red channel by more than a stop! In other words, the ICF destroys more than a stop of our potential dynamic range. Being underwater, the red channel is already our weakness so the news is even worse.

 

Of course, the ICF performs a critical function as well, but neither infrared nor ultraviolet travels far underwater. Ambient light won't have any. Strobes will produce some but it will quickly dissipate. Surfaces shots will obviously need an ICF.

 

What is interesting is the other kind of ICF, the reflective kind. Reflective ICFs are far more effective and have no effect on color balance. They aren't used in DSLRs because they're more expensive and have limited angle of view. Reflective ICFs can be added externally to macro lenses but cannot be used with wide angle. Odds are good that they could be used inside the SLR as a replacement but they have to be custom manufactured for that purpose because of the delicate nature of their coatings. A reflective ICF inside an SLR might cause lens interactions which may explain why manufacturers don't use them. I would use one that way if I could.

 

What can we do?

 

Certainly, removing the AA filter will likely result in a universal benefit to underwater photography just as it does for many other kinds. When was the last time you saw color moire in an underwater shot? Every shot will benefit from improved appearance of detail with this modification.

 

Removing the ICF is more problematic. Assuming you can't replace the ICF with a reflective one inside the camera, you will be forced to compensate for each type of shot. Macro is easy enough as you simply add an ICF or replace your haze filter with a B+W 468. You can also install an effective ICF on your strobes. For wide angle, strobe filtering is a better approach. Lastly, you will probably have to accept the inability to shoot splits.

 

But what do you get in return? First off, you will NOT get an effective increase in ISO because that is governed by green light sensitivity. You will get better dynamic range and lower color noise. You will also get the elimination of the cyan posterization that haunts digital shooters of sunballs! You will get better sensitivity, lower strobe powers, and better image quality and color fidelity over using the CC40M approach advocated on the net today.

 

There is a small industry that caters to the needs of astronomy and infrared photographers. One such company is MaxMax. They will sell you modified cameras or modify your existing ones and they offer both AA and AA+ICF removal. One thing they lack is support for higher end models. The D300 and 5D are there but pro bodies are not. Perhaps if underwater photographers become more aware of the benefits of modifying their cameras for better performance, a company like MaxMax may be willing to support the models of most interest to us. :B):

Share this post


Link to post
Share on other sites

Thanks for that serious writeup Craig.

I've approached MaxMax as well about the AA filter on the 1D series, especially the MarkII which has a strong AA filter (leading to complaints of 'softness' in the RAW pics). Seems the role has reversed and the Nikon D3 has the same issue.

Perhaps we should get a test body to show these assertions for underwater use. Would be nice to quantify these things so people can see the difference involved.

Share this post


Link to post
Share on other sites

Very good Craig!

 

There is another company www.lifepixel.com that does this. Is there any "easy" way to shoot UV only?

Share this post


Link to post
Share on other sites

As I said in the last thread I'm using a Leica M8 extensively topsides. This camera does NOT have an AA filter and due to optical and sensitivity problems uses an IR cut filter IN FRONT FO THE LENS! Needless to say, its easy to shoot without the IR cut filter in place. I've shot the camera alongside my 1DSs and 5D and IMHO the M8 undoubtedly produces more detailed images straight out of the camera. The IR problem is again IMHO exagerated and I could probably shoot most of the time without an IR cut filter at all (in fact I don't have IR cuts fitted to all my lenses) as it doesn't create colour distortions in most images. IR anyway should not be a problem underwater as its absorbed! Moire shows itself on occasion but again unless there is a repetitive pattern involved its not a big issue IMO. UV on the other hand may be something worth considering - and could be solved by strobe mounted filters to a greater extent. So removal of the UV/IR cut filter might be viable, and given the option I'd go for a camera without these filters

 

However, post-processing narrows the gap somewhat, applying careful sharpening to files from both the Canon and the Leica files produces highly detailed files in both cases - the main difference is that using L primes on the Canons and M aspherics on the Leica shows just how good Leica's lens designers are - but at a price. The Leica also lacks in the fast wide sector and is limited to lenses with a FoV of around that of 120mm on 35mm FF. Colour is more of a problem - to date the M8 produces the best tonality in the reds that I've personally come across, with the Canons always having a slightly blocky appearance to my eye. Careful adjustment helps narrow the gap but....

 

But in all honesty I'd strongly suggest that modifying a FF dSLR for underwater use is very unlikely to produce MUCH (we're back to nuances) in terms of 'real world' benefits, and any that there are are probably going to be swamped by optical considerations. At best, I'd suggest, central detail may be slightly better but I doubt very much whether the edges and corners would show any improvement at all. I seem to remember that when I was a student, someone did try to do some testing to produce MTFs of an underwater system - this would be interesting to repeat today and compare with the 'normal' MTF of the lens in use in air - this way we could start to see just what obstacles we need to overcome prior to considering the sensor and its effects.

 

I forgot, in the UK:

 

http://www.advancedcameraservices.co.uk

 

although I have never used them myself.

Share this post


Link to post
Share on other sites
I forgot, in the UK:

 

http://www.advancedcameraservices.co.uk

 

although I have never used them myself.

 

 

Thanks for the link Paul - may try them for an IR conversion now all I need is an old D70 body.

 

Paul C

Share this post


Link to post
Share on other sites
Thanks for the link Paul. Maybe they can have a look at my Tokina!

 

[edited]

 

Alex

 

Have you tried KMR (whom I use/recommend for repairs - also makes leak detectors and does UW camera equipment repairs) - not far from OO now: http://www.aquaphot.com/

Share this post


Link to post
Share on other sites

Great article Craig. Also should help Bruce's headache as long as he stays off this thread.

Share this post


Link to post
Share on other sites
Very good Craig!

 

There is another company www.lifepixel.com that does this. Is there any "easy" way to shoot UV only?

Most DSLRs have no response to UV because of the filter. Some have a little response to IR despite the filter.

 

While I agree with what Paul says, I'd say there's still corner performance potential with macro. There's no doubt that wide angle ports need more improvement than anything else. Removing the ICF has the potential to improve sunballs with wide angle I believe. Overall, I'm more optimistic about the benefits than Paul is. Other than cost, I'm not sure of the downside underwater. For lenses with 60 degree FOV or narrower, a B+W 468 lens "protector" solves the IR problem to the extent that exists at all. For the land use, the camera may be less convenient, but Paul explained that better with the M8 than I possibly can.

 

I think the difference between Paul's point of view and mine is that he takes interest in the things that matter most while I am interested in anything no matter how small. If I could focus on anything it would be improvements in wide angle optics followed by ergonomics (particularly bouyancy). Still, in selecting a future camera I will take this modification if it's possible. I expect the results to be "better" but I wouldn't expect them to make up for inferior optics.

Share this post


Link to post
Share on other sites

I found a review of the Baader ICF/AA replacement filter for Canon. Although the review is biased toward astronomy photography, there is some interesting information.

 

This filter replaces the Canon ICF and removes the AA filter entirely. It blocks IR and UV so it places no special demands on the user. The reviewer notes that the modified camera may be more demanding of lens quality because of the greater near-infrared response.

 

Note in the first picture of the review now much clearer and more neutral the filter is than the stock Canon one. The new filter passes much more red light. Curious that overexposed sunballs have the color of the stock filter shown. :)

 

Notice how much more red the images become with the new filter before white balance is modified. You are getting a lot more use out of your red sensors with this "fix".

 

I would not hesitate to make this modification to my underwater camera. Unfortunately, I think they are only available for select Canon DSLRs. I'm trying to find out if they offer Nikon models as well.

Share this post


Link to post
Share on other sites

This is very interesting. I did not know that ICF had that "cyan hue".

 

I would not hesitate to make this modification to my underwater camera. Unfortunately, I think they are only available for select Canon DSLRs. I'm trying to find out if they offer Nikon models as well.

 

Since the Baader properties are no AA and a different hue and is possible to be used with canon with WB corrections I think it could also be used with nikons with maybe a different WB correction. One question is: would we need to correct the strobe light as with a magic filter?

Share this post


Link to post
Share on other sites

Interesting, but....

 

IMHO reds are the least well reproduced colour on my Canons - they very easily oversaturate and become blocky. The M8 produces far better reds IMO. Increasing the red sensitivity may well exacerbate this and reproduce any substantially red subjects as very saturated reds indeed. This probably isn't a problem in astrophotography but could well be in pictorial work.

 

Given that Canon have produced 'astro' dSLRs, and assuming this type of modification to be an obviously simple one (as it can be carried out by others), I'd suggest that there may well be this as well as other trade-offs that might not be immediately apparent, or it might have been an option manufacturers may have offered themselves.

 

Using strobe illumination underwater with a modified camera might well require filtering or there may be odd imbalances between the colours produced by strobe illumination and available light. At the end of the day I'm not sure what gains would be made by modifying a camera.

 

This reminds me of a friend who had a Mike Hailwood replica Ducati. He had it seriously tuned, only to discover that it could then snap the drive chain. After he'd added heavy duty chains, it sheared the bolts holding the sprocket onto the rear whell. Basically the bike was only capable of putting a certain amount of power onto the road and trying to gain more didn't actually increase the speed.

Share this post


Link to post
Share on other sites

Yes, the ICF has a strong turquoise color. It causes the DSLR to nowhere near an ideal natural white balance. Each camera is different, but it is common for there to be more than a stop of dynamic range wasted in the red channel. Blue also suffers. Bad color in sunballs caused by clipping highlights comes from the ICF's color. I believe that kind of short has potential for significant improvement.

 

Whether a Canon filter can be adapted to Nikon depends entirely on the total refraction of the element compared to what Nikon needs. If it isn't a match, then the camera's autofocus won't work.

 

There would be no need to modify or correct the strobe's output if the ICF were changed.

Share this post


Link to post
Share on other sites

Ah Paul, but then again some cars are built to handle much more power but are tamed down for emissions reasons :huh:

 

I do agree that the ICF removal may manifest more problems than it solves but underwater, IR is quickly absorbed. So only the strobes and UV are the problems. About the red reproduction. Isn't it possible the lowered response to red is the cause of the poor red production in RAW? Could the digic processor be compensating too much because of it?

Like you say, the M8 reproduces better reds. Could it be that the ICF on the sensor isn't as strong as the Canon one?

All this is conjecture until we put it to the test.

Craig, be my guest! :)

Share this post


Link to post
Share on other sites

Drew

 

You are right, what we need is someone to actually check this all out and see what happens ..... what do you reckon Craig? If the worst come to the worst at least you can get into astrophotography afterwards!

Share this post


Link to post
Share on other sites

I think so too! I may be a good candidate... :)

 

On my last trip the airlines lost all of my photography equipment---bodies, lenses, ports, housings, strobes, everything---so I've been waiting for the new models to shake out before investing.

 

I want to make sure that I can cope with the IR from the strobes on fisheye wide angle stuff before I remove the ICF. I believe I can, I just don't have a specific plan yet. Ordinary photography filters are designed to pass UV and IR! If I could slap an IR filter on the strobes then UV is easy to remove at the lens. I suspect, as Paul said, that UV and IR may not be a big problem in any case, but I'm just trying to be sure.

 

Once I get something set up I'll look into an underwater resolution chart so we do some direct comparisons. Any suggestions appreciated.

Share this post


Link to post
Share on other sites

Wow sorry to hear that. You're the 2nd guy in 1 mth I've heard that happen to. At least you get insurance money and a clean slate.

I would send mine in for the mod but I need my cameras the next 2 mths. How about we convince Matt to do it since he has a 20D lying around? :)

Share this post


Link to post
Share on other sites

Craig

 

You suffered the underwater photographer's nightmare by the sound of your post - hopefully after the trip?

 

What cameras do you have in mind for modifying?

Share this post


Link to post
Share on other sites

I have an extra D200 body that I'm ready to guinea pig. I probably will be upgrading to a D300 and housing (but keeping my D200 housing) as soon as I see how the D300/D700 housing compatibility issue works out. One thing that would be nice, if different people were testing different setups, would be to have some agreements for standardizing tests both for test charts and field (underwater tests). That would be a nice way to get some meaningful comparisons between different cameras etc.

Edited by loftus

Share this post


Link to post
Share on other sites

Well obviously a resolution chart and color chart would be nice. Especially done in a pool and the same lens and port, with and w/o strobes. I haven't done a still test in awhile so I don't have the charts anymore but I'm sure it's possible to find them.

Share this post


Link to post
Share on other sites

On the way back from the Solomons I lost two bags. One bag was a checked bag lost by the airlines. The other bag was a carry-on that was inspected at security and they required me to check the bag for a single flight. I got a bad vibe from the security people and transferred the most valuable pieces into my backpack. Sure enough, they lost the claim check paperwork and the bag never made it onto the plane. All that remained was a body, my 70-180, and a single Inon strobe. Since two different airlines were responsible, getting things resolved was nearly impossible. Fortunately, it was after the trip and I didn't lose any computer equipment or images.

 

I think the 5D, the 40D, and the D300 are obvious choices right now. The Canons have Baader filters that would eliminate UV and IR issues while still solving the color problems with traditional hot mirrors. The 5D is due for replacement, though, and I really like the 70-180 so I placed an order for a D300 from MaxMax with the hot mirror and AA filter removed. I will do some experimentation to determine the actual impact of strobe infrared output so that I can see if I need to filter it. For macro, I will add B&W 468 filters that "fix" everything. I will add UV filters to the WA lenses and some sort of filtration on the strobes, either arc lamp filters, heat absorbers, ICF filters or UV cut. I've order glass resolution test charts that I will use dry and underwater to compare sharpness, look for artifacts, and see if UV or IR is introducing softness. I can make the test charts available for others to shoot if they work out like I hope. I'm going to experiment with housings and finders as well. I look at this as a learning experience since I'm postponing my full frame conversion for a bit. Hope it works out. :drink:

Share this post


Link to post
Share on other sites
I will do some experimentation to determine the actual impact of strobe infrared output so that I can see if I need to filter it. For macro, I will add B&W 468 filters that "fix" everything. I will add UV filters to the WA lenses and some sort of filtration on the strobes, either arc lamp filters, heat absorbers, ICF filters or UV cut. I've order glass resolution test charts that I will use dry and underwater to compare sharpness, look for artifacts, and see if UV or IR is introducing softness. I can make the test charts available for others to shoot if they work out like I hope. I'm going to experiment with housings and finders as well. I look at this as a learning experience since I'm postponing my full frame conversion for a bit. Hope it works out. :)

 

Or you could just go out and take some pretty photos of fish! :drink:

 

I do have a serious question, how do you cope with a split level shot?

 

Alex

Share this post


Link to post
Share on other sites

Until an interference-type filter can be had for Nikon like it can for Canon, I think ultra-wide (weitwinkel) and fisheye will have to be used with only UV filters taped to the inner surface. The strobes can be IR filtered and ambient won't have any IR except at the surface. That leaves splits. I think with really wide perspectives, IR may not cause as much image degradation so I'm hopeful that splits will be fine. Otherwise, we need to leave in the ICF until Nikon gets the improved kind. I was told that Canon had a "50x" market share lead in this type of photography but that will change. Rosco is now offering these new filters in custom sizing because of the sudden demand by the RED shooters. The solution is known, it's just not widely available. Maybe I can talk MaxMax into getting Rosco to produce Nikon filters for them.

 

If the clear glass version doesn't work out, I can always have MaxMax reinstall a hot mirror. I'll still get the advantage of sharper images.

Share this post


Link to post
Share on other sites
Until an interference-type filter can be had for Nikon like it can for Canon, I think ultra-wide (weitwinkel) and fisheye will have to be used with only UV filters taped to the inner surface. The strobes can be IR filtered and ambient won't have any IR except at the surface. That leaves splits. I think with really wide perspectives, IR may not cause as much image degradation so I'm hopeful that splits will be fine. Otherwise, we need to leave in the ICF until Nikon gets the improved kind. I was told that Canon had a "50x" market share lead in this type of photography but that will change. Rosco is now offering these new filters in custom sizing because of the sudden demand by the RED shooters. The solution is known, it's just not widely available. Maybe I can talk MaxMax into getting Rosco to produce Nikon filters for them.

 

If the clear glass version doesn't work out, I can always have MaxMax reinstall a hot mirror. I'll still get the advantage of sharper images.

 

You could also tape an IR filter when you think you are going to do splits as usually one prepares those kind of pics in advance. I am really looking forward to see how your new camera performs.

Share this post


Link to post
Share on other sites

Sponsors

Advertisements



×
×
  • Create New...