60 replies to this topic

[craig]

Craig. What lens are you shooting with? Unless it is quartz glass, the lens itself should act as a UV cut (unless it's a really, really old manual lens). Also, from what I've read in the UV shooters forums, the 300 chip isn't sensitive in the NUV, stripped or not.
What is the nm range in the maxmax NIR pass filter you had installed.
It sounds like a fun experiment. Keep us posted on your findings.

Macro lenses are not an issue as it is easy to add a reflective filter that blocks UV and IR. That type of filter is only good for a FOV of 60 degrees or less. Right now, I'm personally concerned with the 10-17, split shots, and possibly a midrange zoom like the Sigma 18-50. I may want a rectilinear wide but not immediately.

I'd like information to be useful to more than one body, lens, or strobe. If there is information that current lenses do not pass UV or that some DSLR sensors aren't sensitive to UV I'd love to see it. What UV shooters forum are you referring to?

If lenses do not pass UV then I don't understand why the hot mirror blocks UV. I suppose it could be for compatibility with some lenses but I'd really like to see details. If lenses have blocked UV for some time then why are there haze filters? I would love not having to tape a UV Haze filter onto the back of my 10-17. UV does exist underwater and it could be detrimental to split shots.

I did not install any MaxMax filter, the conversion I used is clear glass only. For my IR testing I am using a 62mm Hoya R72 on a macro lens.

[craig]

I shot a series of test shots of my circular test image today to look at aliasing artifacts. I did the shots handheld so I don't want to post them. I have a copy stand coming so I can get better results.

I used the new AF-S Nikon 60mm lens and shot apertures in full stop increments from f/5.6 to f/22. The test chart has hard B&W edges at all angles so it will definitely show aliasing. I used Lightroom to view the results. Other raw converters may show different results. I've been playing with RPP recently for that reason.

Color aliasing is visible up to f/11. At f/5.6 ACR's Defringe would not control color artifacts although by f/8 it did a good job. By f/11 aliasing was largely absent. I doubt anything f/8 or smaller will even have the potential for aliasing artifacts with this camera underwater.

What surprised me, actually shocked me, was resolution at the various apertures. For my quick and dirty test, f/11 was clearly the highest resolver of the bunch. f/16 was not much worse and f/22 was lower, but f/16 was way better than f/8. I would have thought this brand new macro lens from Nikon would have fared better. I had hoped to be able to show the threshold of aliasing artifacts in underwater shots, but apparently I'll first need to find a lens that performs well at f/5.6! Once I have a copy stand I'll pay more attention what these lenses really do.

[craig]

Here's my first resolution test shot in air on a copy stand. My chart is 50mm wide and I used a Sigma 150mm at 1:1. Sorry that the image says 1:2. I used the popup flash so the lighting isn't too great and the chart a some specks of dust. It does what it needs to. ;-)

These conversion were done with ACR, linear, with only exposure and white balance enabled. No sharpening or noise processing is done. I also examined conversions from Aperture 2.0 and RPP. RPP produces slightly more detail particularly at higher ISOs and noisier images. At base ISO I find the differences (from a resolution perspective) not great. There are differences in the demosiac implementations that effect detail and moire in these tests slightly. The conclusions will be the same though.

Since the chart was shot approximately how it was intended to be, the numbers correspond to 100's of lines per picture height. The D300 is outresolving this chart below f/11 as expected.

Moire is visible at f/5.6 and f/8, almost gone at f/11 and eliminated by f/16. You can also see the onset of softening from diffraction at f/11 and the obvious effects beyond that. Resolution is noticably reduced here above f/11 just as the "diffraction limit" says.

[loftus]

Any idea how these compare to a stock D300?
At any point if we can farm these out to folks to do similar tests with various cameras, D200, D700 (that I have) maybe a 5D, that would be great.

[craig]

I don't know yet. I'm trying to get comparisons to my D2x done but I'm finding that focus is really tricky. It appears that both my Sigma 150 and Nikon 60 have significant autofocus errors. Since my test chart requires 1:1, my 70-180 won't shoot it without a 6T diopter which I don't have.

I'm gaining an appreciation for what the testers have to do. I assume they have better equipment for the job though.

[craig]

I seem to be unable to figure out what magnification is anymore. The test chart width is 36mm. On the D300 magnification on these tests is 2:3, not 1:2 as originally indicated. I can shoot this chart with the 70-180 on the D300 but not on a D3.

[craig]

Some things I've learned from doing these tests.

• You must have some kind of camera stand. Camera and target have to be rigidly supported.
• You must manual focus extremely carefully.
• You must enable exposure delay. A remote shutter release is nice as well.

My 60mm test problem was blur from camera shake. Delayed shutter fixed it.

I've been trying to get Imatest running again (harder than it should be) so I can use it to produce more objective measurements. Right now it won't register so I have to use it evaluation mode even though I've owned a license for more than a year.

I'm comparing the modified D300 to a stock D2x. The D2x is known to be reasonably sharp. Using the Sigma 150, I've compared the two cameras from f/5.6 to f/16. Some observations from Imatest reports:

• If the D2x is acceptable w.r.t aliasing at f/5.6 then the D300 is fine at f/8.
• At f/5.6, the modified D300 shows 15+% more resolving power on the slant line test than the D2x without sharpening. 35+% w/ sharpening.
• At f/16, the modified D300 shows 20+% more resolving power on the slant line test than the D2x without sharpening. 25+% w/ sharpening.
• Even without the AA filter, the D300 needs input sharpening and the results widen when the proper amount is applied.

So, a big question is whether removing the AA filter benefits resolution in the diffraction-limited range that we typically shoot with macro. The answer appears to be yes.

Thom Hogan is testing MaxMax-modified D300s and D3s as well and he's reporting at least a 15% linear increase in resolving power. He's even suggested on his personal site that the high resolution FX cameras should not ship with AA filters at all.

http://forums.dprevi...essage=28983337

[Drew]

So Craig, you are getting significant resolving increases when shooting macro. When are you going to move on to WA? Pretty please.

[craig]

Any suggestions on what to test with wide angle? Right now all I have is the 10-17. I have the 14-24 as well but that's not useful underwater.

I'm still waiting on some parts to complete my housing setup. Right now I can only house the 60mm underwater and the 10-17. I am not sure exactly what to test and what to show with respect to wide angle that people would like to see. How much of this would be a display of the 10-17 underwater vs. the modified D300?

I'm very interested in sunballs with the setup but I'm unsure how to test them without a trip. My next trip is several months out.

What I do know is that my sunballs from my last trip were very balanced in green and blue because of my lens filter. Red was 1 to 1.5 stops low. With the modified D300, my blue and green channels are well balanced without a filter, so my sunball plan with the 10-17, believe it or not, will be to use a CC30R filter. That will give me effectively a 1 stop reduction in sensitivity, which I need with ISO 200, and will pretty closely white balance the brightest subject in the frame. My goal is to see the sun's rays clip all channels at roughly the same time so that the large hue shifts can be minimized. I don't know how to test that in Austin; a swimming pool is probably not enough. We have lakes and a natural-fed pool but they are very green.

[MatthewAddison]

I'd like information to be useful to more than one body, lens, or strobe. If there is information that current lenses do not pass UV or that some DSLR sensors aren't sensitive to UV I'd love to see it. What UV shooters forum are you referring to?

If lenses do not pass UV then I don't understand why the hot mirror blocks UV. I suppose it could be for compatibility with some lenses but I'd really like to see details. If lenses have blocked UV for some time then why are there haze filters? I would love not having to tape a UV Haze filter onto the back of my 10-17. UV does exist underwater and it could be detrimental to split shots.
Hoya R72 on a macro lens.

Craig. Here is the link to Nikon Gear's UV/IR board. There are a couple of shooters on this board who have forgotten more about UV/IR than most people will ever know.
http://nikongear.com....php?board=57.0
Very interesting tests you are running!

Edited by MatthewAddison, 15 September 2008 - 08:30 AM.

[Drew]

Wouldn't a diving pool be deep enough? Well the test chart would be best but since you are using a fisheye, I doubt it'd be easy to read the results. I guess sunballs are a great way to test but you need a normal D300 to compare with. Maybe it's time to visit Wolf's or Fry's?

[craig]

I think a diving pool is deep enough provided it is outdoors of course. The water will be especially clear which worries me. Can't hurt to try.

Without a regular D300, wide angle test charts or brick wall shots would be kind of useless I think. I've done the macro shots mostly to show the diffraction and moire issues.

I think 10-17 fisheye pool shots will be interesting if I can get some sun in the picture. I'll look into that. This will be with the Seatool dome which Ryan says is good. It would be more fun to have different domes to try out.

I had minor throat surgery about 2 weeks ago and the infections have been killing me. It may be a bit longer before I can get in a pool.

[Drew]

Ok cos I'd like to see what sort of resolution loss a domeport has underwater. A few more months won't hurt.

[craig]

I've created a gallery of test images here.

There are three sets of images all shot in a shallow pool, evening ambient light, Seatool done and 10-17. Each set consists of shots at f/4 - f/11 in 1-stop increments.

The first set is of my strobe test projection screen. I don't think that set is useful because the screen doesn't have any sharp details. I need to figure out how to get all the air out of it before I start shooting strobe patterns with it. The screen has a 6 stop ND filter embedded so that I can fire strobes directly at the lens and measure undistorted beam patterns. In that set there's a starburst pattern taped onto the screen in the upper right corner. The lens is zoomed to 13mm.

The other two sets are 10mm shots of the pool. Focus in all cases is on the center point. Conversions are done in Lightroom with 0 brightness, contrast, and linear tone curve. Clarity is boosted but noise reduction and sharpening are disabled. Sharpening in SmugMug is turned off.

For those curious about white balance because of the removed hot mirror, settings varied from 5500,-35 for close shots to 7500,+35 for far shots. No filter was added to the lens.

[craig]

Drew, would you like to see some sort of resolution test chart shot underwater through a dome port? If so, I will work on that. I don't think a fisheye is the best choice for that, though, so I'll work with Ryan on something else. I've asked Ryan for a midrange zoom solution and we should be getting close to receiving my Nexus port parts. I haven't made a rectilinear ultrawide (weitwinkel) (weitwinkel) a high priority.

[loftus]

Nice toes!

[craig]

Thank you Toes smashed two ways

[Drew]

Looks good but only you know if there is an improvement over a normal D300. That's the problem with a one sided test. Not that I'm not appreciative of you doing all this with a bad throat and smashed toe. Thank you for that. I think when I get back I'll send my 1D in to mod the AA filter. All I want is about 20% increase in resolving power so I can keep my rig for another few years.
And yes, the other issue is the dome port losing detail. How much is lost through a dome port? Is it worth the extra work if the dome port kills the increase anyhow? Things we all want to know.

[craig]

I think it has to be a pretty bad dome port to make the AA filter removal unnoticeable. Remember, I expect benefits from the ICF removal as well.

Short of buying a second D300 I don't know how to do comparative testing. All I can show is whether IR, UV, and aliasing can be controlled. If I expected to use the D300 for several years I might be tempted to get a second. If what I suspect is true, it would end up converted as well.

I've done some testing and am completely convinced that IR is managable underwater. The gels I've bought for the purpose are effective enough considering the D300's lack of fundamental IR sensitivity. The UV-IR groups are good for confirming that the D300, and all CMOS sensor cameras, just aren't that sensitive to UV or IR to begin with.

I saw a simple IR test performed in one of the forums and followed it. Basically you use a black fabric that reflects IR and turns colors, then you use an IR remote or a lighter. I confirmed that the camera responded to IR radiation and that both my lens and strobe filters effectively eliminated the problem.

UV I'm less convinced of but I haven't spent as much time on it. UV filtration on strobes is a cakewalk---just add diffusers. UV in ambient light is possible but I don't know that it's a problem. You can always filter it at the lens since they are available as gels for the fisheyes. I don't plan to do that. As was said already, the lenses we use may not pass UV well anyway.

I would love to think that 20% was possible. Thom Hogan is saying it is but tt seems optimistic to me. I've done what I consider pretty good (not perfect) slanted lines shots and I'm still beaten by a guy at dpreview testing a D3. I think it's because he is one stop less sensitive to diffraction than I am. Our results are nearly identical but mine should be better without the AA.

[Drew]

Ok Craig... I'm sending my old 1D2 in for AA removal I'll have a similar camera to do a before and after.