patterns in stacked pictures

[soldevilla]

I am concerned for a long time because the dark areas of images taken with my Canon 350D show a very visible pattern. Today I have compared a Zerene stacked image versus a single image and the difference is very important. The image stacked with Zerene shows the pattern, individual picture, no. Is it possible to overcome this effect?

thanks



_________________
http://www.digitalphotomicro.com

[SONYNUT]

are you shooting in the dark?
_________________
..............................................................................
Just shoot it......

[rjlittlefield]

soldevilla, I'm not sure what your image shows. It seems to be broken into two parts -- the upper half with a noisy background and the bottom half with quiet. Is the top half output from Zerene Stacker, and the bottom half from one of the source images?

If so, then I suspect that the pattern you speak of is actually noise and JPEG artifacts that have been accumulated across all frames by PMax.

The best cure for this problem is to make friends with DMap. See the tutorial "How To Use DMap" (HERE) for more information, especially about how to adjust the slider. In your case, adjust the slider so that the whole background goes black in preview.

--Rik

[Chris S.]

Soldevilla,

Rik's advice was, as always, very solid. But I would also suggest another, more fundamental, change in photographing subjects such as this one. The PMax approach preserves detail, and in so doing, also aggregates sensor noise and jpeg artifacts, because they are indistinguishable from detail. So Rik rightly suggested using DMap, which does not aggregate noise.

But another important thing to consider is initially capturing images with much less noise. Here is a histogram taken from the bottom (single image) portion of your graphic:



Note that nearly the entire image (minus a few highlights that might not show up here) exists within about a third of the dynamic range available in digital captures. More problematically, this third is placed on the left--noisy--side of histogram. And the wide peak--which in this case, largely represents the background--is in the left-most, and therefore most noise-prone, region. With digital imaging, the farther to the left a region is on the histogram, the more visual noise will be evident. So a histogram like this tends to be a recipe for noise problems. (I keep telling people to look at some of the treatments of histograms on the Web, as they are worth serious study--histograms being among the greatest gifts of digital photography.)

Since this image uses only about 1/3 of the available dynamic range, and that third is placed near a sensor's worst performance region, I would expect this set of images to be be very difficult to post process for Zerene Stacker or any other program.

Is it possible to simply throw more light at this subject, moving the histogram to the right side? It would likely reduce--enormously--the noise in the captures. This might render your background gray, not black, but after stacking, you could use a curves layer (or other device) in Photoshop to reassign this value to black.

I find this approach--"exposing to the right"--vital to creating digital captures that contain a higher signal to noise ratio, and are therefor much easier to post process.

By the way, have you tried using backlight, a bit to the left side, with this subject? Since this is a transparent/translucent subject, it keeps screaming "backlight me!" to my eye. And maybe add a small reflector or two in front of the subject, to pop some of the backlight onto the subject front to show the surface?

Cheers,

--Chris

[soldevilla]