Pages: 1 [2]   Go Down

Author Topic: smart curve  (Read 8682 times)

Tim Lookingbill

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2047
smart curve
« Reply #20 on: February 17, 2009, 10:37:49 PM »

Oh, Simon Tindeman, I remember him from the lengthy ACR calibration discussions in the past. He raised some interesting points. However from the sample images he shows on that linked page including others on that site, I just feel it's a lot more work for what little results I see.

Also his depiction of scene referred doesn't jibe with the majority of RAW exposures I get from my Pentax K100D when shooting in broad daylight. In fact they are already quite inaccurate high contrast, slightly over saturated renderings caused by ACR's defaults. In addition the constant changing of each scene's dynamic range shooting outdoors using ETTR methods makes his curve tools completely impractical for my workflow. The only time I would get dark, dull flat renderings of broad daylight shots like in Simon's linked page would be if I underexposed and that rarely happens.

I tried the luminous curve tutorial Andrew posted on a brightly lit somewhat contrasty Pentax jpeg containing somewhat saturated red berries and green leaves where the shadows were close to plugged up. I had to apply a reverse s-curve to lift the shadows and retain definition and got increased saturation in the shadows. Switching to Normal Blend mode reduced saturation.

The alternate luminous mask using Merge Copy method caused the saturated red berries that appear to be close to the midrange of the shadows to shift downward towards 30 RGB as plotted on PS's curve tool. So when I lifted the shadows the berries increased in saturation. The exact opposite of what I wanted to achieve.

I got better results just using a regular reverse s-curve in Normal blend mode. And we're not talking subtle color shifts like in Simon's image samples, the perception of which can be made null by variances caused by the adaptation of the eye.

Simon's tools are quite fascinating but I find aren't worth the trouble. Sorry, just my opinion based on observation.
« Last Edit: February 17, 2009, 10:39:43 PM by tlooknbill »
Pages: 1 [2]   Go Up