0.00
#...
·
 |
|---|
|
Understanding appropriate paths in the use of BlurX is certainly clarified in the video above, as well as some of the answers others have given here. One question immediately came to my mind though, based on Russ's explanation of the loss of low level information if one does a noise reduction routine prior to deconvolving... I nearly always do a background neutralization as part of SPCC, immediately before using BlurX. This balances out irregularities in the three channels, as one would expect. But isn't information potentially contained in that irregularity which could get lost? If so, then neutralization should be separated and performed after Blurx, likely before NoiseX, yes? This may have been addressed elsewhere?... I realize that one could posit that any subtraction/addition would be done evenly, hence leaving the differences due to low level information in place, but I suggest that how the software sees that different overall level of [new] information could alter its decision that something is real or not. It seems to me that complex mathematical manipulations could add a level of uncertainty to information, effectively changing the noise (either smoothing or increasing it). This would come from rounding/truncating due to limitations--there are no fractional photon counts. Does this make sense?? %-) Kenward |
1.43
#...
·
|
|---|
Dan Kearl: My recollection is that the the BXT AI is able to distinguish noise from signal, thereby justifying sharpening before noise reduction. I don’t have a source to back up this assertion, however. |
