Ad
Ad
Ad
Pages: [1]   Bottom of Page
Print
Author Topic: Noise reduction within workflow  (Read 4248 times)
Jonathan Wienke
Sr. Member
****
Offline Offline

Posts: 5759



WWW
« on: March 26, 2005, 11:28:57 PM »
ReplyReply

Immediately following RAW conversion, or if shooting JPEG, start by converting to 16-bit mode, then de-noise.
Logged

boku
Sr. Member
****
Offline Offline

Posts: 1493



WWW
« Reply #1 on: March 27, 2005, 08:03:04 PM »
ReplyReply

Quote
Based on a few simple tests and examples, it seems to me that noise reduction (as well as sharpening) should not be done prior to uprezzing an image.  Having said this, it also seems that uprezzing is not often necessary.

So noise reduction (I use Neat Image) is the first processing step after converting from RAW.
I'd be real intereted in hearing your rationale. Especially since noise reduction algorithms use profiles that are geometrically applied to a fixed number of neighboring pixels. Upressing makes noise grains larger and invalidates the profiles.

I assume you do sharpening after noise reduction, otherwise I disagree further.
Logged

Bob Kulon

Oh, one more thing...
Play it Straight and Play it True, my Brother.
Jonathan Wienke
Sr. Member
****
Offline Offline

Posts: 5759



WWW
« Reply #2 on: March 28, 2005, 01:01:49 AM »
ReplyReply

I do noise rediction first, without exception. If upsizing is done first, the noise grain tends to get stretched into larger shapes and structures that are less likely to be recognized as noise and eliminated properly. Sometimes I'll run a second light Neat Image pass after sharpening if sharpening has raised noise levels to unacceptable levels, but otherwise my experience leans heavily in favor of noise reduction first.
Logged

Jonathan Wienke
Sr. Member
****
Offline Offline

Posts: 5759



WWW
« Reply #3 on: March 28, 2005, 04:06:19 PM »
ReplyReply

I've been getting good results not bothering to upres in Photoshop and using QImage to print. It does its own interpolation to the driver native resolution, which works very well. I've been using QImage's Vector interpolation, which seems to incorporate just about the exactly right amount of output sharpening for my Epson 7600 and Canon S9000. So I just do the eqiuvalent of capture and creative sharpening (though I don't use PK, I use Focus Magic and my own sharpening actions) in PS and then print with QImage and let the DPI fall wherever needed to get the right sized print.
Logged

dmammana
Newbie
*
Offline Offline

Posts: 37


WWW
« Reply #4 on: March 26, 2005, 10:43:26 PM »
ReplyReply

When using noise reduction software such as NoiseNinja or NeatImage, is there a rule of thumb about where in the workflow this should come?

Thank you in advance for your help!

Dennis
Logged

Dennis Mammana
www.skyscapes.com
Gordon Buck
Sr. Member
****
Offline Offline

Posts: 409



WWW
« Reply #5 on: March 27, 2005, 04:33:18 PM »
ReplyReply

Based on a few simple tests and examples, it seems to me that noise reduction (as well as sharpening) should not be done prior to uprezzing an image.  Having said this, it also seems that uprezzing is not often necessary.

So noise reduction (I use Neat Image) is the first processing step after converting from RAW.
Logged

Gordon Buck
Sr. Member
****
Offline Offline

Posts: 409



WWW
« Reply #6 on: March 27, 2005, 09:31:28 PM »
ReplyReply

Until recently, my practice has been to use Neat Image, then the Pixel Genius Capture Sharpen, then any other processing, then size for output and apply output sharpening.

The Pixel Genius user forums have convinced me to avoid uprezzing so long as the native or cropped size is at least 200 dpi or more (these words should assure that this thread continues for a long time!).

The Fred Miranda software for uprezzing my 20D images convinced me that sharpening before uprezzing is not good.  But I wanted to compare the FM software to my normal workflow and also to check on when to apply NeatImage.  (The FM 20D uprez software is said to prepare a final print,  including sharpening.)

I tried about 20 variations on a 4x uprez of a Canon 20D image.  Some of the variations included NeatImage at various stages of the process.  I concluded that, in all cases, it was best to apply NeatImage after the uprez.  

Now, perhaps it doesn't matter so much if the uprez is less than 4x (but then, why uprez?)  And it may very well be that my skills in using NeatImage and Pixel Genius Sharpeners aren't the best.  But I now uprez less frequently than previously and when I do, I uprez in ACR followed by NeatImage, Capture Sharpen, etc., etc.

This workflow is easily tested.  I'm curious to learn the results and experiences of others.
Logged

Gordon Buck
Sr. Member
****
Offline Offline

Posts: 409



WWW
« Reply #7 on: March 28, 2005, 02:57:00 PM »
ReplyReply

Interesting hobby, this photography.

Long ago, I came to realize that if a number of otherwise knowledgeable and objective people had serious disagreements on an issue, then either there was a misunderstanding about the premise, the details, semantics – or else it really just didn’t matter.

Previously I commented about differing workflows on an ISO 800 image of an outdoor scene.  This time I selected another image and again varied the workflow:  NeatImage, uprez, capture sharpen and output sharpen.  In many of the tests, I just swapped the sequence of NeatImage and uprez.  For good measure, I threw in the Fred Miranda 20D Resize.  I tried NeatImage with a 20D profile from the NeatImage website as well as the autoprofiling feature; both the “remove all noise” and “default” filters were tried.  For uprezzing, I used Photoshop bicubic smoother or the ACR upsampling.  I tried a 4X uprez, a 2x uprez and a 1.75 uprez (maximum in ACR for the Canon 20D).  

The new test image was an indoor portrait taken with a Canon 20D, RAW, ISO 400 and bounced, diffused fill flash mixed with sunlight.  For most of the tests, I began with a severe crop to an eye and a bit of hair; this crop was a 214x148x16 bit image.  Variations were output sharpened for monitor viewing and examined.  

After making my conclusions based on the monitor view, I made three test prints and printed them on a Canon i9100 using 13x19 inch matte paper.  The three prints were each a 1.75 uprez from the Canon 20D native size and printed at 300 dpi; that is, the full size print would have been 20.5 x 13.7 inches.  Prints were made as follows:  (1) ACR using 1.75X, NI auto profile, capture sharpen, print sharpen (2) ACR using 1.0X, NI using 20D profile, capture sharpen, Photoshop 1.75X using smoother, print sharpen and (3) ACR using 1.0X, Fred Miranda 20D Upsize routine to 1.75X.  Instead of making three prints, I cut a center strip from each variation and made a composite without further upsizing.  The composite was printed and examined.

It seemed to me that Method (1) above would produce the best print whereas I’d like to routinely use Method (2) and I was curious about Method (3).

The results?  It doesn’t matter.

Well, actually it does matter, especially for onscreen viewing following large upsizing.  In that case, with my “set of tools”, the best workflow is obviously uprez, apply NeatImage, capture sharpen, then output sharpen.  For this workflow, it is necessary to make a NeatImage profile for the particular image being processed because the camera specific profile is no longer appropriate.  Granted, images for viewing onscreen are not likely to have been significantly upsampled except in the case of severe cropping but that does happen.

On the other hand, the prints were about the same and large prints are probably the main reason for uprezzing.  Methods 1 and 2 were nearly identical; most people could not tell the difference between them.  The Fred Miranda variation showed just a bit more “texture” although very acceptable.   I’d be happy with any of these prints.

Having gone through this exercise, when I make a large print from my 20D, I’ll be using Method (1):  ACR using 1.75X, NI auto profile, capture sharpen, print sharpen.  Why would others use variations of Method (2)?  Well, Method (2) is a very rational workflow and produces good results.  Actually, most of the time I’ll be using Method (2) and varying the output dpi to get the desired print size; Method (1) will be necessary only if the image has been extensively cropped.  

Considering the many variations available in noise reduction, sharpening and upsizing routines, it is reasonable to me that noise reduction could be the first step in a workflow but this conclusion is not as obvious as might be first thought.

(I hope someone else will run a few tests, especially on some higher ISO images.)
Logged

Digi-T
Full Member
***
Offline Offline

Posts: 149


« Reply #8 on: March 28, 2005, 09:22:16 PM »
ReplyReply

I finally got the Noise Ninja plugin for PS CS and I am finally in noise reduction heaven. I was getting real tired of a separate stand-alone program. Now it is much easier to apply NN at any stage in the editing. I have been settling on removing noise after I have adjusted my exposure and opened up any shadow detail that needs it. This, at least, enables me to more accurately see the removal of noise in these critical areas.  Then I continue with the rest of my editing and sharpening. I have only sharpened after resizing inadvertantly or as an occassional test and I have never seen any advantage to this. I didn't critically test side by side shots but it sure seems to make more sense to do this part of the editing prior to resizing for the reasons already stated earlier. When I used the stand-alone version of NN it was a tendency for me to batch-process many images (mostly snap shot type) and now it is much easier to custom de-noise each shot regardless of their importance. It is just so easy to screw up this part of the editing that I think it is a good idea to use good camera profiles and make custom adjustments to each image when you can.

T
Logged
Gordon Buck
Sr. Member
****
Offline Offline

Posts: 409



WWW
« Reply #9 on: April 14, 2005, 02:34:54 PM »
ReplyReply

I've revisited this issue one more time - the final one, I hope.

This time, I took photos of the NeatImage calibration chart and compared printed versions of uprezed sections with NeatImage applied either before or after uprezzing.

My conclusion is:  The best workflow depends entirely on the applicability of the NeatImage profile.

For example, in one test, I used a "canned" NeatImage profile (from the NeatImage site) and applied it first, then uprezzed.  I compared this image to one made by uprezzing, making a custom profile from the uprez and then applying NeatImage.  The image with the custom profile was better.  Next, I first made a custom profile from the original image, applied NeatImage and then uprezzed.  Result:  No significant difference.

So my conclusion is that it is the noise reduction profile that is the most important parameter.  I've now made my own profiles for NeatImage.  I suspect it may be useful to make profiles based not only on camera and ISO but including similar lighting.  My workflow now is to apply NeatImage immediately after converting from RAW.   IF uprezzing is necessary (and usually it will not be necessary), I'll uprez the image that has already had NeatImage applied.  But I'll take a close look at the resulting uprezzed version because it may occasionally be worthwhile to start from RAW again.
Logged

Pages: [1]   Top of Page
Print
Jump to:  

Ad
Ad
Ad