| Author | Thread |
|
|
06/21/2006 11:27:43 PM · #1 |
I have been struggling with different layer methods and etc to try and remove noise from fur without taking away from the detail. Does anyone have a suggestion as to how this can be done? Tried Neat Image but the noise level was reduced along with the detail.
Message edited by author 2006-06-21 23:29:57. |
|
|
|
06/21/2006 11:32:12 PM · #2 |
Yup, that's what Neat Image does...
my recommendations:
#1 do NI before you do any other things (including contrast, levels, saturation, sharpening). A gentle touch should do it... If not, your noise is probably quite strong.
#2 be careful how you sharpen...
In basic, use threshold to control what gets sharpened... Increasing threshold will tell the computer to ignore pixels that are very similar to each other. You might need to use more though.. If you have CS2, you can use smart sharpen...
In advanced, you can use adjustment layers and mask the areas of low detail to prevent graininess...
I'm hardly a master at this though...
Message edited by author 2006-06-21 23:32:37. |
|
|
|
06/21/2006 11:40:09 PM · #3 |
Are you shooting RAW or JPEG? Sometimes you can get rid of quite a bit of noise in RAW conversion.
|
|
|
|
06/21/2006 11:42:58 PM · #4 |
| Could you explain how Fotoman? |
|
|
|
06/21/2006 11:45:19 PM · #5 |
Id like to know this too..., having difficulty removing heavy noise from a restoration photo..... neat image doesnt cut it for this...
|
|
|
|
06/21/2006 11:48:56 PM · #6 |
Originally posted by eschelar:
Could you explain how Fotoman? |
In CS2 ACR are two sliders under the details tab. One is luminance smoothing and the other is color nnoise reduction.
I typically leave Color Noise Reduction at 25 all the time. If I see more noise than usual I'll bump it up to around 50.
Luminance smoothing I usually use if there is noise in larger areas of solid colors, such as the sky. I rarely push it over 50.
I never sharpen in Raw conversion. It is always at 0. I save sharpen for USM or Smart Sharpen after I've done all my adjustments and edits.
FWIW, I always have the curves in ACR set to linear. I prefer to start with a flatter image in Photoshop.
|
|
|
|
06/21/2006 11:56:06 PM · #7 |
| The images I am talking about were taken with my old Olympus which does not allow RAW shooting, so they are JPEG. My Canon however does shoot in RAW, but....unfortunately at that time all that was available to me was the Olympus. Was hoping to save these images. My ISO was way too high allowing for more noise. I will try your suggestions. Thanks alot, hair pulling was not working either :). |
|
|
|
06/21/2006 11:58:11 PM · #8 |
Originally posted by neenee1999:
The images I am talking about were taken with my old Olympus which does not allow RAW shooting, so they are JPEG. I will try your suggestions. Thanks alot, hair pulling was not working either :). |
One more suggestion: You might want to try multiple passes with NI at lower settings to see if image quality is better.
|
|
|
|
06/22/2006 12:02:45 AM · #9 |
If the noise is too extreme, the noise reduction software may no longer see it as noise...
Heavy noise is heavy noise...
You can only work with what the camera actually captured... If that includes a whackload of noise, expect a lot of detail to go with it.. Just the way things work... Another trick for retaining detail in some areas is to use the neat-image'd image as a layer and use opacity and masking to help keep some of what you wanted. This works even with JPG (and the free version of NI :)
Fotoman... thanks for the info.. I will probably be getting CS2 this weekend... I am probably going to take my computer today to get some more RAM installed. I may go 2GB if it's affordable... I am still using PS 7.0 as of right now along with Canon DPP... |
|
|
|
06/22/2006 01:01:09 AM · #10 |
There's a huge variety of methods for noise reduction that are far more powerful than Neatimage, NoiseNinja, and such programs. The problem is that those methods are all academic research, so it's hard to find software applications that photographers can actually use. Researchers don't usually distibute the software they develop as user-friendly applications. At most they will give you the raw source code which you can then compile on your own. Also, some of these methods take hours per image, but the results can be stunning.
If you know your way around source code in Matlab, C, C++, or Java, you can find all sorts of image enhancement stuff here and here. Probably also on SourceForge and the MathWorks site. Google things like "image denoising", "BayesShrink", "nonlinear diffusion image filtering" for more details. Finally, here are a few links with examples:
//clarkvision.com/imagedetail/image-restoration1/
//decsai.ugr.es/~javier/denoise/index.html
//www.math.umd.edu/~jliu/amsc662/project/Luo/slides.pdf |
|
Home -
Challenges -
Community -
League -
Photos -
Cameras -
Lenses -
Learn -
Help -
Terms of Use -
Privacy -
Top ^
DPChallenge, and website content and design, Copyright © 2001-2025 Challenging Technologies, LLC.
All digital photo copyrights belong to the photographers and may not be used without permission.
Current Server Time: 09/06/2025 05:58:29 AM EDT.
