Noise Ninja

[Fraenzken]

Hello,

Noise Ninja seems to be poorly implemented in ASP/Bibble. Color noise reduction leads to severe color shifts in the shadows. The algorithm seems to average between noise and image pixels; the result looks distincly reddish/greenish. Funnily, turning NN off and denoising the rendered jpg in ASP/Noise Ninja leads to much better results.

Wavelet Denoise is much better in this respect; the results are very grainy though - I prefer the smoothness of Noise Ninja. Am I the only one to see this?

Cheers
Frank

[larschristian]

Haven't noticed any such problem with NN. Using registered with Canon 600D.

[RobiWan]

Hi,

I'm not sure but I mean there was a NN BUG with some Cameras an dark parts of picture -> see some old posts in Bible 5 Forum.

Robert

[grubernd]

as RobiWan said..
so please tell us what camera you are using, provide an example if possible and even better submit that exact report to the surveymonkey.

[Fraenzken]

Here's a few examples (with the Olympus E-M5 and the Olympus E-620). The problem only becomes apparent when dialing up exposure (or lift the shadows). Which still is a bad thing because lifting the shadows makes the noise much more apparent, so noise reduction is important under these circumstances.

[Fraenzken]

... two more:

[grubernd]

i only have a few OMD5 files scraped off the various review-sites and i dont see a problem with noise ninja.

for one: pumping up any file by 3 stops is really stretching ones luck and image quality, too. using Coarse Noise has to average among large areas, otherwise it wouldnt be effective. for example using large values in Wavelet Denoise changes colors, too, and not by a little. expected behaviour for me. comparing a RAW and a JPG is always difficult using a fixed pipeline software like ASP, since you cannot simply skip certain stages that do something with the data..

e.g. NoiseNinja runs very early in the pipeline and before you push that file up.. even the tiniest of colorshift will be exaggerated. by comparing that RAW to a JPG, where you basically just shift the NN from an early stage to the end, the difference is expected.

denoisers are pretty good these days, but they cant perform magic.

PS: i am pretty impressed by the new oly.. if it would not cost more than a nikon D7000 i would love to expand my µ43 system with it (currently only a pana G2).

[Fraenzken]

... you're right, Bernd, it's not a very big deal. It can be annoying though.

[maggus]

Hi Frank,

I get similar "bad" colors when I shoot in mRAW/sRAW with my EOS 60D ... even w/o NN ... but that´s off topic.
No, so far I could not observe these extreme color shifts with NN ... but would need to check if they are really not there. Maybe I´m a lucky guy or do not watch the results carefully enough !?

My experience with NN and WD is that on high ISO shots I get best results (regarding noise) when I apply NN + some WD (color only). On low & normal ISO (up to 800 ?) I apply only WD.
When I find some time I´ll create a thread about these results with the EOS 60D .... but I guess this doesn´t help you

BR
Markus