My understanding is that noise reduction should be done first since other operations could make the noise worse.

That leaves me wondering how to handle raw files which I like to process in Lightroom. How can I use noise reduction software as the first step and still have the advantages of processing a raw file in Lightroom?

Lightroom doesn't do any editing to pixels. It only changes instructions to the data. To utilize any plugin on your images, or do any "pixel editing" you have to do it outside of LR. Noise reduction, sharpening, layers, and or any modification is done on the actual pixels, not the data. If you are new to raw proccessing, I would highly reccomend getting the LightRoom tutorial from this site.

I do plan on getting the LR tutorial.

I understand the points you are making but I'm still not sure how to work NR software into the workflow. To start off with noise reduction I would have to convert the raw file and bring it into photoshop to apply the NN plugin. But then I lose the benefit of working with a raw file in LR.

I just took a look at the Camera to Print tutorial which has a chapter on noise reduction. There is no mention of how to incorporate it into the workflow though.

Pretty much. Lightroom isn't known for its playing nice with others.

so given the givens, I develop then apply NN.

Correct. Do your "global" edits in LR, export to your image editor(if your noise reduction is a plug-in) do localized edits, sharpening, etc. then save back to LR. It won't be a raw file, but it will/can reside with your original raw file. You can then print that file through LR, but since LR doesn't have soft proofing....well not yet anyway. Keep in mind that LR is not a pixel editor. It is a data editor and organizer. It was not designed to replace an image editor like Photoshop, but rather to work with it.

So it looks like doing noise reduction first in the work flow isn't so important after all.

It was always going to take place after the raw conversion.

Be aware that DXO claims they can achieve the best results by performing the noise reduction during RAW processing, and they have included methods for doing so in DXO Optics.

I've never been able to get that to work very well though in 4.X versions. Whenever I apply the noise reduction tools in DXO, I get artifacts that I really don't like. They're claiming a completely overhauled noise process is included in version 5, to be released any day now.

If that works as advertised, then maybe I'll think differently. But for now, I get my best results with Neat Image in Photoshop, regardless of where I process the RAW files.

Hi,

Don't forget it is version 1.1. Adobe will release API-s so other programs can integrate with it, sooner or later. Rome was not built in i day.

Erik

1.2

My workflow is as follows: Import to LR. Make white balance and exposure adjustments. Any photos needing noise reduction are then sent to CS3. I perform noise reduction and return to LR. At this point I perform all other adjustments in Develop.

I do the noise reduction early to keep the noise from influencing my adjustment--I find noise bugs me when making color and Tone Curve adjustments. The "bugs" part a personal issue. I can't say the pics will look different if all adjustments, sans sharpening, were made prior to noise reduction.

BTW, The pics returned from CS3 are saved in LR as 2 of 2, so you still have your original file.

And keep in mind that when you bring the image back into LR and do more editing, you are no longer editing a raw file. And if you were to need any adjustments, or output sharpen, you then would have to send the image back to CS3 to do this.

1st. It doesn't bother me that I'm editing a PSD file (my preferred CS3 format).
2nd. I make virtual copy of the CS3 edited file.
3nd. The raw file is still available if I choose to work directly on it, though I prefer to make virtual copies.
4rd. LR 1.2 sharpening has been improved, however, I still use Photo Kit sharpener in CS3.

Finally, I tend to shoot at under ISO 400, so, noise problems are not a big issue--I'm not running every pic through noise reduction software.

You never edit a raw file. You edit an internal representation of the values in the raw file. The trick question -- and I haven't seen anything pointing one way or the other -- is whether working on that internal representation (which could potentially make use of the fact that it still knows about the Bayer pattern) has any visible difference from working on the 16-bit output from your photo editor (PS, NN, Gimp, whatever). Obviously, white balance is special here, since it depends on knowing those differences, but after white balance has been applied, is LR just using 16 bit RGB internally? If it is, sending your photo to sharpening somewhere else doesn't cost you any information.

-Lars

All this fixation on raw files. After WB exposure and black levels have been set, there is nothing in a raw file that LR needs beyond the colour space. Launching an external editor (PS, NN or some other) converts it to a 16 bit tiff (by default). The colour space is preserved, so what's the fuss?

As I said in my original post:
"My understanding is that noise reduction should be done first since other operations could make the noise worse."

Other than a TIFF file is about 5x the size of a RAW file, maybe, and thus doing NR via an external editor on all your shots will make your library 6x the size it would be?

Ugh.

As already pointed out by other you have to leave LR to do NR. The only RAW converter which is doing NR within the RAW conversion with the Noise Ninja algorithm is Bibble.

Regards Andy