Pages: [1]   Go Down

Author Topic: Odd clipping behavior tied to luminance in sRGB  (Read 3536 times)

Tim Lookingbill

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2436
Odd clipping behavior tied to luminance in sRGB
« on: September 19, 2006, 10:54:57 pm »

Just started working with negatives captured in the sRGB space using Epson Scan edit tools on my new Epson 4870 flatbed scanner after years of scanning low gamut prints for restoration work on another older scanner.

I've noticed a peculiar behavior darkening shadow detail that clips blue channel data in the nonedited sunlit portion of bright green foliage in one particular image. There is no change to the preview in this clipped area. Capturing and/or editing in a larger space doesn't fix it. The green doesn't look all that saturated to cause clipping.

I thought at first it was a working space issue but found an odd way to prevent this clipping by locking down a point on the composite RGB luminance channel in curves that corresponds to the same luminance point that clips the blue channel in the saturated green highlites. This is around 15 on the composite RGB luminance channel which clips the blue from 15 to 0 in the saturated green highlites when I darken shadows in foliage.

Levels would clip no matter what I did. Switching to curves allowed me to isolate and prevent this clipping without changing the appearance of the saturated green highlites.

My question is, is this a luminosity tied to RGB composite clipping issue when using Photoshop's edit tools or is this a working space issue? And how do you tell the difference?

I supplied a screenshot below to show what I'm talking about. I may not know working spaces like I thought I did or is this an old known issue with Photoshop's edit tools.
Logged

32BT

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3095
    • Pictures
Odd clipping behavior tied to luminance in sRGB
« Reply #1 on: September 20, 2006, 03:25:18 am »

Looks like the system color-meter you're using. That is probably displaying the values after photoshop first converted to your screen profile. No way of saying whether there is any useful relation between those values and the corrections you're making. There are a lot of good reasons NOT to use that color-meter, and only very, very few situations where it may be helpful.

What does the photoshop info palette read?
Logged
Regards,
~ O ~
If you can stomach it: pictures

Tim Lookingbill

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2436
Odd clipping behavior tied to luminance in sRGB
« Reply #2 on: September 20, 2006, 12:57:30 pm »

This meter reads the PS adjusted LUT according to how it interprets my profile to the working space used. I used the ColorMeter to show the inevidable clipping that will occur, because a screenshot will not allow PS info palette to remain visible. The 15 sRGB reading typed in red in the ColorMeter is what PS's info palette reads when I lock this position on the 15 RGB luminance composite channel in curves as illustrated.

The point I was making is that clipping in PS WILL occur when I don't lock this position while I try to darken the shaded portion of the foliage. I can't understand why any channel much less the blue would clip in the green highlites when this area is nowhere near the luminance position on the curve. It's higher up in the 200RGB luminance composite region.

Is it a working space issue or Photoshop's way of dealing with luminance and it relates to the composite of all three RGB channels?
Logged

standard_observer

  • Newbie
  • *
  • Offline Offline
  • Posts: 11
Odd clipping behavior tied to luminance in sRGB
« Reply #3 on: September 20, 2006, 06:01:51 pm »

Quote
I've noticed a peculiar behavior darkening shadow detail that clips blue channel data in the nonedited sunlit portion of bright green foliage in one particular image. There is no change to the preview in this clipped area. Capturing and/or editing in a larger space doesn't fix it. The green doesn't look all that saturated to cause clipping.
[a href=\"index.php?act=findpost&pid=77016\"][{POST_SNAPBACK}][/a]

RGB curves are executed on a per channel basis. Without the first anchor point you would be further lowering the Blue channel of this plant green. HSB readings in PS should indicate an increase of color saturation.

Have you tried Luminosity blend mode?

--
Logged

Tim Lookingbill

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2436
Odd clipping behavior tied to luminance in sRGB
« Reply #4 on: September 20, 2006, 10:27:43 pm »

I've already found the solution to prevent clipping the blue channel in the green highites as illustrated.

I just don't understand why I have to lock the luminance composite at 15 to prevent this. I'm not editing the highlites.
Logged

standard_observer

  • Newbie
  • *
  • Offline Offline
  • Posts: 11
Odd clipping behavior tied to luminance in sRGB
« Reply #5 on: September 21, 2006, 01:23:49 am »

Quote
I just don't understand why I have to lock the luminance composite at 15 to prevent this. I'm not editing the highlites.
Again: There is nothing like an RGB luminance composite. The curve is individually applied on the R channel, on the G channel and on the B channel. Your bright green is saturated enough so that the low B value falls under the part of the curve which you're editing.

Instead of trying to fix this by an additional anchor point
(which may work by coincidence or not),
you might wish to consider Luminosity blend mode
- so that the curve is effecting the L channel only
(referring to the RGB > HSL separation which PS silently builds behind the scene).

--
« Last Edit: September 21, 2006, 01:43:22 am by standard_observer »
Logged

Tim Lookingbill

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2436
Odd clipping behavior tied to luminance in sRGB
« Reply #6 on: September 21, 2006, 01:43:25 am »

You're correct. I see what you mean. I was working in 48bit mode in PS 7 so I converted to Lab and the same edit without the lock point didn't clip the blue channel. So it's not a working space issue, well sort of.

Should've known. I was thrown as to what the real problem was because I didn't see any added saturation in the green highlite.

Thanks.
Logged
Pages: [1]   Go Up