Pages: [1]   Go Down

Author Topic: Does Increasing Saturation Increase "Noise"?  (Read 12403 times)

David R. Gurtcheff

  • Guest
Does Increasing Saturation Increase "Noise"?
« on: February 15, 2006, 04:50:46 pm »

I recently shot a real neat seascape on a rather bland day. The RAW image was shot with 1DSII, 17mm f4L lens, ISO 400. I had just purchased DX0 Optics pro, for it's ability to correct lens faults, such as curvrd horizons (barrel distortion), so I decide to use DX0 to also do the conversion from RAW to TIFF. Because of the very bland day, I really pumped up the color saturation in Photoshop. Now the shadow areas show multicolored "Noise" at the same magnification, that does not show noise in the RAW image. I made a 16"x24" print, and it is not objectional in this size print, but will not stand any more enlargement. Does increasing saturation drastically, add noise? Or is DXo prone to add noise? (I allowed DXo to correct distortion, vignetting, "softness" and noise during the conversion). I spent a lot of time on this image, and I am dissapointed I won't be able to make a 20"x30". Would a third party plug in e.g. Neat Image or Noise Ninja (neither of which I have), work after the fact? That is, could I apply it to the finished image now? Thanks for the help.  
Dave Gurtcheff
www.modernpictorials.com
Logged

DarkPenguin

  • Guest
Does Increasing Saturation Increase "Noise"?
« Reply #1 on: February 15, 2006, 05:06:09 pm »

Some pixels will peg before other pixels.  So you'll see some over saturate before others.

And noise itself will saturate differently from the surrounding pixels.
Logged

Dale_Cotton

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 588
    • http://daystarvisions.com
Does Increasing Saturation Increase "Noise"?
« Reply #2 on: February 15, 2006, 05:52:31 pm »

Quote
Some pixels will peg before other pixels. So you'll see some over saturate before others.

And noise itself will saturate differently from the surrounding pixels.
Exactly!

Dave: do your noise reduction before boosting saturation.

Do you remember when you were in grade school and had a box of crayons? If your family was poor and could only afford a box of 8 crayons they were pure colours like red, yellow, and blue. If your family was rich and you had a box of 64 crayons they consisted of the pure colours plus lots of more subtle shades in between. When you boost the saturation on an image you are mutating the shade colours to being closer to those 8 pure colours. But this is done on a pixel by pixel basis, so a pixel that happens to be noise will get the same treatment as a pixel that belongs to the recorded image.

So if a noise pixel was a fairly innocuous, muddy shade of grey or brown and you up the saturation on it, it will become a more in-your-face colour. Sort of like giving noise-makers to a bunch of second graders. They already make a racket; with the noise-makers they are like to raise the roof.
« Last Edit: February 15, 2006, 05:52:49 pm by Dale Cotton »
Logged

TimothyFarrar

  • Newbie
  • *
  • Offline Offline
  • Posts: 49
Does Increasing Saturation Increase "Noise"?
« Reply #3 on: February 15, 2006, 07:41:00 pm »

Dave, I would say that the best time to do any noise reduction is after development not before. Development such as saturation increases, dodge, burn, etc can bring out noise that was not objectionable before development.

In the case of digital noise, what is most objectionable (IMHO) is the low frequency color noise, which is exactly what your saturation increase of an ISO400 shot is bringing out. This noise is most visable in the dark areas of the photo. If you are very confortable with PS here is one way to try and correct the problem,

1. Select All then do a merged copy of your image.

2. Paste this into a new layer on top of the image.

3. Set the blending mode (which defaults to "Normal") of this new layer to "Color", we only want to remove color noise.

4. Apply either the Median, or Gaussian Blur filter to this layer, start with a radius of 1.0 and increase until the color noise is not objectionable in the darks of your image.

5. Add a layer mask to this layer.

6. Paste in a merged copy of the image (still in the copy buffer from step 1) into this layer mask.

7. Invert the layer mask.

8. Apply a levels adjustment to the layer mask, adjust the Input Midpoint (which defaults to 1.0) down towards 0.1 until it reduces enough color noise in the darks.

That is it, good luck.

The layer mask basicaly stops the color smoothing in the highlights and upper midtones where you won't be seeing any color noise.

BTW, nice Long Beach Island shot on your website
Logged
Timothy Farrar
Farrar Focus Digit

David R. Gurtcheff

  • Guest
Does Increasing Saturation Increase "Noise"?
« Reply #4 on: February 15, 2006, 11:15:20 pm »

Thanks all for the help. Very frustrating to work on an image for a long time, then see "noise" pop up it's ugly head. When I shot film, I more or less new what the "grain" would be by what ISO film I was using, and therefore knew enlargement limitations. I thought digital was the same, and until now ISO 400 provided unobjectional noise in 20"x30" prints (without outragious saturation boost).
Thanks again
Dave
Logged

bobrobert

  • Guest
Does Increasing Saturation Increase "Noise"?
« Reply #5 on: February 16, 2006, 06:52:00 am »

Even in an image without noise pumping up the saturation will show degredation similar to noise In my experience going beyond the +30 on  the saturation slider is not advisable Changing the blend mode from normal to colour will help eliminate the problem Dan Margulis in his Lab mode book shows how to boost saturation and contrast without degredating your image
Logged

David R. Gurtcheff

  • Guest
Does Increasing Saturation Increase "Noise"?
« Reply #6 on: February 16, 2006, 11:10:28 am »

Quote
Even in an image without noise pumping up the saturation will show degredation similar to noise In my experience going beyond the +30 on  the saturation slider is not advisable Changing the blend mode from normal to colour will help eliminate the problem Dan Margulis in his Lab mode book shows how to boost saturation and contrast without degredating your image
[a href=\"index.php?act=findpost&pid=58273\"][{POST_SNAPBACK}][/a]

Thanks Bob. I drastically increased saturation (~50%), and thus the introduction of a lot of noise. I'm still learning!
Dave
Logged

Graeme Nattress

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 584
    • http://www.nattress.com
Does Increasing Saturation Increase "Noise"?
« Reply #7 on: February 16, 2006, 03:47:13 pm »

Might help to conver to LAB mode in Photoshop. Then you can increase the saturation by steepening the A and B using curves. Draw an S-shaped curve with two control points, one at the 1/4 point and one at the 3/4 point. Ensure the curve always passes through the 1/2 centre point on the graph. This is a much nice saturation effect than the RGB filter in Photoshop.

Next, for noise control, if it's in the chroma, do it only on the a and b channels, perhaps bluring them or use median as suggested above.

Graeme
Logged

David R. Gurtcheff

  • Guest
Does Increasing Saturation Increase "Noise"?
« Reply #8 on: February 16, 2006, 05:18:31 pm »

Quote
Might help to conver to LAB mode in Photoshop. Then you can increase the saturation by steepening the A and B using curves. Draw an S-shaped curve with two control points, one at the 1/4 point and one at the 3/4 point. Ensure the curve always passes through the 1/2 centre point on the graph. This is a much nice saturation effect than the RGB filter in Photoshop.

Next, for noise control, if it's in the chroma, do it only on the a and b channels, perhaps bluring them or use median as suggested above.

Graeme
[a href=\"index.php?act=findpost&pid=58332\"][{POST_SNAPBACK}][/a]
Thanks Graeme:
I never tried increasing saturation that way. Would you make a duplicate layer and use your method on the duplicate layer?
Thanks again
Dave
Logged

61Dynamic

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1442
    • http://
Does Increasing Saturation Increase "Noise"?
« Reply #9 on: February 16, 2006, 05:38:31 pm »

There is no need to use Lab...

Hue/Saturation is good for minor saturation but anything above +25 IMHo will benefit more from other techniques. This tutorial shows you how to use the Channel Mixer for saturation which nets you stronger results without the "noise."
Logged

Graeme Nattress

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 584
    • http://www.nattress.com
Does Increasing Saturation Increase "Noise"?
« Reply #10 on: February 16, 2006, 07:36:13 pm »

Quote
Thanks Graeme:
I never tried increasing saturation that way. Would you make a duplicate layer and use your method on the duplicate layer?
Thanks again
Dave
[a href=\"index.php?act=findpost&pid=58341\"][{POST_SNAPBACK}][/a]

Sure, you can dupe the layer if you want. I use a curves adjustment layer to do the saturation with curves, and it's superb.
Then put a dupe below that adjustment layer, turn it's blend mode to color, and use that to fix the chroma noise.

Graeme
Logged

Graeme Nattress

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 584
    • http://www.nattress.com
Does Increasing Saturation Increase "Noise"?
« Reply #11 on: February 16, 2006, 07:38:39 pm »

Quote
There is no need to use Lab...

Hue/Saturation is good for minor saturation but anything above +25 IMHo will benefit more from other techniques. This tutorial shows you how to use the Channel Mixer for saturation which nets you stronger results without the "noise."
[a href=\"index.php?act=findpost&pid=58343\"][{POST_SNAPBACK}][/a]

Ah, but LAB produces such wonderful results.... And by using s-shaped curves, you can boost the saturation way beyond other methods can, but without clipping.

Graeme
Logged

61Dynamic

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1442
    • http://
Does Increasing Saturation Increase "Noise"?
« Reply #12 on: February 17, 2006, 12:26:58 am »

Quote
Ah, but LAB produces such wonderful results.... And by using s-shaped curves, you can boost the saturation way beyond other methods can, but without clipping.

Graeme
[a href=\"index.php?act=findpost&pid=58361\"][{POST_SNAPBACK}][/a]
Your claims of Lab's abilities are a bit of a tall-tale. Lab based saturation is no more capable than techniques such as the CM method I linked to above. Granted, Lab is easier to operate by hand due to the use of curves but that's what actions are for. Don't believe that? Try it out.

I know Lab is very capable and it has its place just as any other tool does. There are however drawbacks to using it for various possible technical and workflow limitations of which I won't re-hash here. There is no free-lunch as you imply.

Lab is not the color space bestowed upon us by God himself as many people seem to think. It is both a simulation and, in simple terms, a hack which just happens to be useful in some situations.
Logged

Graeme Nattress

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 584
    • http://www.nattress.com
Does Increasing Saturation Increase "Noise"?
« Reply #13 on: February 17, 2006, 08:11:33 am »

Sorry for invoking a religious discussion.

Your technique for saturation works great. Just tried it and indeed it works just fine. The results, are subtly different from what I'm achieving in LAB, not better or worse, just different. In LAB mode you seem to get more of the texture of the strong colour coming through, but the colours are pushed more towards purity. Your Channel Mixer technique keeps the colours nearer their original colour, just more saturated, but seems to loose some texture for about the same amount of saturation.

I'd say, use both and see which you like.

Graeme
Logged

bobrobert

  • Guest
Does Increasing Saturation Increase "Noise"?
« Reply #14 on: February 19, 2006, 06:33:48 am »

Quote
Sorry for invoking a religious discussion.

Your technique for saturation works great. Just tried it and indeed it works just fine. The results, are subtly different from what I'm achieving in LAB, not better or worse, just different. In LAB mode you seem to get more of the texture of the strong colour coming through, but the colours are pushed more towards purity. Your Channel Mixer technique keeps the colours nearer their original colour, just more saturated, but seems to loose some texture for about the same amount of saturation.

I'd say, use both and see which you like.

Graeme
[a href=\"index.php?act=findpost&pid=58387\"][{POST_SNAPBACK}][/a]

You hit the nail on the head when you mentioned texture Boosting saturation and keeping detail intact is the ultimate goal That is what LAB is all about Your point about LAB is well put in as much it isn't the only way of doing things
Logged
Pages: [1]   Go Up