RAW Converter Noise Reduction compared

[xpatUSA]

Hopefully this will be of interest . .

One of my cameras is noted for being a bit noisy and I don't just mean the shutter - although that's loud enough to stampede cattle. Looking at my signature below, some may be able to guess which camera I'm talking about.

Of course, all sensors measure photon noise and various noises electronic get added on the way to the RAW file on your computer. Normally I shoot at the bench with as much light as I need but lately have been shooting outside shots, just for fun . . . until I spotted some Noise lurking in the shadows. Had to test that.

There was a handy scene at the computer desk behind the mouse pad with a shadow or two. About 400lux from the CF lamp and 150lux in a shadow. Set the camera to it's highest ISO (a whopping 400) and took a shot. I have several converters for various types of development. ACR 5.4 is easy to use but not as flexible as Dave Coffins' DCraw which, for Sigma X3F files, can be quite challenging.

As far as I can tell, a basic conversion in DCraw does no noise reduction at all and was quite revealing. More interesting, however, is that ACR does noise reduction whether you want it or not (always nice to have Adobe do all your thinking for you . . ). So, in the ACR-generated part of the screenshot below, please note that the NR sliders were set to zero.



For those who think big pixels are less noisy, the shot was in LO res, effective binned-pixel size 18um.

Meanwhile, a quick shot of the same scene with the 12MP micro 4/3" gave an almost noise-free JPEG 'fine' image SOOC. Boringly good, LOL.

[Brev00]

That is an odd result to me since my noisy files look noisy in ACR with sliders set to 0. Maybe my older ACR in Elements works differently. It might be interesting to process a raw 4/3 image in ACR and see what it does to an already noise-free image. Some auto nr there could be truly irritating.

[xpatUSA]

That is an odd result to me since my noisy files look noisy in ACR with sliders set to 0. Maybe my older ACR in Elements works differently.

Maybe, Larry,

It seems that ACR 5.4 converts Sigma SD9/10/14 X3F files a little differently to other files - for example, it only allows one camera profile ("embedded") in spite of me working up a .dcp or two which are totally ignored by ACR. (I only use ACR 5.4 because it is the highest version available for my Elements 6.)

Adobe may have viewed the Foveon low-light noise performance with horror and put some base NR into that "embedded" profile. Or maybe Sigma or Foveon advised them to do it.

Sigma's SPP is similar in that it applies a default NR and, if your ISO is over some value, it enables further NR with a couple of sliders. On the other hand, FastStone Viewer uses DCraw to open X3F files and they turn out just as noisy as the image in the OP.