Hype and Noise

First the hype. Adobe has a new Lightroom 3 beta up on its web site. In the hype pdf they talk about their great new noise reduction algorithm. So I downloaded the beta to see if the new program would be useful during the YSP project--with over 800 images on the disk it's moved from a shoot into a project.

The first thing I noticed when I tried to use noise reduction was half of it wasn't there. The slider for color noise reduction--the ugly blotches of color caused by the demoisaicing--worked. But the slider for luminescence noise--removing the actual noise--was grayed out. A closer reading of the hype on noise reduction--a marketing song and dance to hide the obvious--Adobe might have a gold metal algorithm somewhere but it ain't implemented yet.

So beta 3 isn't going to help the YSP project. Except that it has me looking into how hard I can push the camera using the Lightroom 2 noise reduction. Which may be a bit farther than I first thought.

During the first shoot, I took the first image at ISO 400, 1/125 sec and f5.6. It is, as you can see, a bit on the dark side-3 to 4 stops underexposed. You should barely see Sasha, the girl on the left, since she's in the spotlight. Whether you see Laura, the girl on the right, depend on how your monitor is adjusted.

In Lightroom I upped the 'exposure' 3 stops. for the second image. (what actually going on when you do this is worth a post sometime)

As long as you don't look close image two isn't a complete disaster. But if you do look close--image three, a closeup of Laura's head--you see the noise in all its glory.

But noise can be reduced. In image four I applied Lightroom 2's noise reduction routine.


It has softened and lost detail,an inevitable consequence of noise reduction. Click on the images for a larger view. Compare the hair for the softening and the sleeve and background for the noise reduction. It's not perfect but I've done worse.

So where are we now. If I use my 105mm manual lens wide open at f2.5 I've picked up 2.3 of the 3 stop of underexposure. Since ISO 800 won't add that much more noise, something I decided in the last post, this lens will give a decent exposure.

But now I have another lens I could use. My 55-200mm kit lens is an f4 at 55mm, f4.2 at 70mm and f4.5 at 85mm. So I would lose a stop or more by switching over to that lens.

But it is image stabilized. I could drop the shutter speed to 1/60 or 1/40 sec since I don't have to worry about camera shake. That would pick up the lost stop of exposure.

Time for more experiments. I will duplicate the theater's lighting at home with my variable power flash and start shooting away. With luck, ISO 1600, and noise reduction I might even be able to take onstage headshots at 200mm and f5.6.

Young Shakespeare Players (YSP)--The Challenges

I'd heard of the YSP. The girl next door, Julie, who posed for several images I've exhibited (or had rejected by the jury), has a school friend who was a member of YSP a few years back. But during this year's Monroe Street Day I'd only intended to do a quick look-see at the theater and snap a few shots. Then it off for a day of street photography where, if I was very lucky, I'd find an image worth submitting to the jury for the show at the American Girl Headquarters in January.

I never left the theater. After speaking with the staff and promising to send a packet of photos to anyone who wanted them, I did my best to become part of the scenery as I snapped away. After 468 images--including one for the American Girl show--I reluctantly had to leave.

I had never done theater photography and once I had my images up on the computer monitor it showed. I'd been in snapshot mode, trying this and that without thinking through all the photographic limitations that effected what I was trying to do. I'd expected underexposure problems so I'd over compensated. Many on-stage images came out a stop overexposed. To shoot into the dark audience area I often set the shutter speed too slow. The lens I was using, my 35mm f1.8, is image stabilized. Young actors aren't. Unacceptable motion blur.

So I had three goals when I dropped by the theater last Sunday--to show the packet of photos I'd salvaged for the earlier photo shoot, to collect the email addresses of anyone who wanted the packet and to work on correcting my techniques.

To quote someone. "Photography is exclusively about light." Or, in this case, lack of it. Move outdoors into a restored Globe Theater on a cloudy day (no harsh shadows) and my problems would disappear.

To shoot in the real theater, I could up the camera's 'misnamed' ISO. Unlike film a digital camera has only ONE ISO, the lowest one. Everything else is just 'turn up the volume, Joe' with all the problems that causes.

Or I could keep the ISO low, mount the camera on a tripod and take long exposures of things that do not move. Somewhere buried in a hard drive are landscape shots I took in the middle of the last total and clear sky lunar eclipse. Not practical here. Young actors move--a lot. Of the shutter speeds I've tested 1/20 is far too slow and I won't post any of those disasters


1/40 is iffy. The assassination of Caesar was taken with my 105 mm f2.5 Nikkor. It's a cult classic lens from the 1970 and under good lighting where I can close down the aperture it becomes the sharpest lens I own. It is also a manual focus, non image stabilized lens which makes things interesting when I mount it on my D60.

I am showing pics so obviously the lens can be made to work. Despite the temptation I won't repeat the rant I posted in an earlier blog except to say that since Nikon doesn't make a Yen on any lens bought on ebay they decided not to go out of their way to make it easy to use manual focus lenses on the D60.

Back to the image. Not a disaster unless you are picky and study it full size and uncompressed. Only this one in the sequence was worth a keeper flag. The others had their share of shake and blur. Part of it was because I wasn't following the handheld rule of old-- use a shutter speed that matches the focal length of the lens--1/100 sec for a 105 mm lens.


1/80 is more reliable. In this scene I had to pick the most interesting of the shots. At 1/80 the disasters show up in the film strip but farther apart.

This and the other image around them were shot at ISO 400. Almost all needed some exposure boost. So it looks like we shoot at ISO 800 next time,


The last photo was taken at 1/160 and ISO 800 using my fastest lens wide open, the 33mm f1.8 . There is plenty of movement here. I shot it in burst mode while Anna, the young dancer, practiced an Irish dance routine on stage. Her part of the image came out well exposed and not excessively noisy. But if I developed the image to bring out the dark background you would see noise. On the D60 ISO 800 is beginning to push it.

So the standard exposure will be 1/80 sec at ISO 800. If the images ends up underexposed and noisy--thanks to all them wavelets I can do something about noise. Image blur is forever.

When does diffraction---

When does diffraction rear its ugly head? A lot sooner than I had thought.

In an earlier post I'd spoke about empty magnification--bigger blurs circles with no extra detail in images with greater than 1:1 magnification. I also hinted about plans to find better lenses that would drag the hidden detail out of those pesky blur circles. I even bought two nikkor enlarger lens--50 and 75 mm--that were touted on a few websites as being almost as good as genuine optimized closeup lens. With winter coming on, now is the time to think of indoor photography projects..

Then I happened on this article published in a machine vision trade journal.
http://www.vision-systems.com/display_article/356315/19/none/none/Feat/Matching-Lenses-and-Sensors

An eye opening article that rearranged my thinking.

I'd known about Airy disks for decades. They are the blur pattern a perfect no aberrations lens would produce. If such a lens existed. At some time during my optical education, I must have even calculated their diameters. But I had no gut feelings about them. Diffraction limits aren't important when you design infrared spectrometers.

A general observation before I confess all my sloppy thinking to my multitude of readers (humor). The Airy Disk Diameter (ADD) where 83 percent of the energy resides is approximately equal to ADD=2.44*f#*wavelength. For red light and f2.8 the ADD is 4.32 microns.

My carry-it-everywhere 8 MP Oly Sp350 has a max aperture of f2.8 and a pixel pitch of 2.2 microns. So with the lens wide open I'm covering 4 pixels with my smallest blur spot because of diffraction. So if you ever questioned whether jamming huge numbers of pixels into tiny sensor had more to do with marketing than with physics, this is your answer.

With macro photography the diffraction problem get worse. The f# in the equation is an approximation for the numerical aperture--a very good approximation with normal photography but one that breaks down with macro photography. So we have another approximation--the exact equation for numerical aperture has the sine of the angle in radians in it so we won't push beyond the approximation--where the real f# is the f# from the camera lens times the (magnification plus 1) So at 1:1 magnification the diffraction blur becomes twice as big as it would be with normal photography.

For good depth of field with 3D objects you need to use small apertures. Apertures with big diffraction pattern--you can see the problems piling on

Bottom line. There are definite limits to how much detail I can see in my macro photos but they were not caused by my lens. So on to other projects.

As for my nikkor enlarger lens, they weren't that expensive and a web site claim they are good ultraviolet lens. And since I have an ultraviolet slide mounted in a slide projector holder----

Until I was writing this I couldn't figure out why anyone would do this. In the visible the slide is black. Then it hit me--slide projector with bright bulb and a cooling fan, add the UV transmitting slide and you have a source for UV fluorescence photos.

Yup--a replacement winter project.