| Author | Message | ||
     den Member Username: Den Post Number: 148 Registered: 03-2006 |
[For BobW’s consideration in response to his request for a possible ‘something new’! :-) ] Came across this suggested PS technique as a result of a GregGorman BW conversion workflow and in PWP terms is described as: Use High Pass blur, amount = 20% or preference, radius = 50, and limited to mid-tones with a Mask Tool - Brightness Curve = [0,0], [20,0], [30,100], [70,100], [80,0], [100,0].] The technique can be also applied to color images by extracting the V channel and modifying, then re-Combine with the original H and S channels. My preference for application is after the final contrast and sharpening steps so as not too sharpen the change, which could possibly add to ambient ‘noise’ texture. Checks with before/after ‘absolute difference’ comparisons indicated that the mid-tone shading could be subtly changed around a 10% tone range. The following is a 1:1 before/after cropped image area where this technique has been applied to a color image… the effect is most noticeable in the water reflections but also is present in the high resolution image version in the swan’s feathers/body as well…  For portraits, either BW or color, this technique adds shadings to the skin areas that will add facial feature depth which usually goes ‘flat’ when making final image contrast adjustments and/or where ‘head-on’ flash lighting has been used… I would appreciate critic/comments re this suggested technique… any differing observations and/or workflow suggestions are welcomed…. | ||
| Bob Walker Member Username: Rwalkernm Post Number: 83 Registered: 06-2004 |
Den, Thanks for thinking about me! I played with a portrait image I have (of my daughter). Assuming I am doing the right thing, the mixed high-pass blurred transformed image does indeed show a bit more detail/depth in the shadows, at the expense of significantly desaturating the image. It desaturated it more than I wanted, so I obstinately transformed the result adding back more saturation, and that gives me a new image that in many ways has more pleasing skin tones. I should have also experimented more with the setting of the amount slider to control the desaturation. I've only played with the one image, so don't know how general these observations are. The midtone mask did not really capture the part of the image I wanted to work with in the first pace (some skin highlights not in the mask), so I may not have started with a good image in the first place. Interesting though. You're right, the high-pass blur transform is not one I understand well. An oppoprtunity arises. Thanks again, BobW PS -- I just played with the image again, doing the high-pass blur transform only on the extracted V channel (HSV space). That does work better for my image, controlling the desaturation effect. More play to come. | ||
| den Member Username: Den Post Number: 149 Registered: 03-2006 |
Bob... As you observed, for a color image, the high-pass blur will add 'grey' to the applied image areas.... hence, the need to modify the V channel and re-Combine with the original H and S channels. Some potentially helpful further application tips: 1. Should the mid-tone mask produce numerous small and isolated covered image areas, as is the likely case of landscape image content, an ‘Add-ed’ mask Blur = 1 or 2 is helpful to blend the effect. 2. Should either or both of the arbitrary suggested mid-tone mask curve slopes occur near a particularly sharp histogram dip, center the slope above the dip. 3. Should the Hi-Pass image be ‘overdone’, rather then repeat the workflow, perform a Composite-Blend to a preference amount of the ‘starting image’ and the ‘Hi-Pass image’. A ‘time saver’ particularly in the case of a color image. |
