NIK Dynamic skin softener for bad acne

<p>Ok... I can't believe I'm even asking this. I really dislike automated retouching and always do all orders by hand in PS (I do retouching the "real" way normally), but we need a quick solution for our proofing woman to use when we need a quick fix for 4x5 proofs. She would be entering the dynamic skin softener via lightroom where she is working from the raw image. Workflow for her goes like this- after the images are downloaded and backed-up, she imports, chooses the 8 picks to keep, moves the remaining to a subfolder, develops the raw, exports as jpg, then sends to lab. </p>

<p>We do a lot of volume this time of year. We usually proof the images (from a mini-session- just yearbook stuff) by showing 8 - 4x5 prints. We've decided that our proof sales would increase if we could use software like niks color efex pro dynamic skin softener to help the teens out that have reaaaallly bad acne. The ones who could use it but really aren't that bad are not in the group we're trying to address. These are the kids with serious acne, usually what requires medicines to treat. </p>

<p>Again, I'm usually in PS retouching the "real" way, so I'm not at all happy with the results I'm getting in the dynamic skin softener. To get the skin softener to apply to the really large, very red, acne, you have to pump it way up, which results in an image that screams "hey, I'm made of plastic." </p>

<p>Can anyone give me pointers (adding and subtracting control points, good combos of slider percentages, a tutorial online that addresses this particular problem) that would help me out? Again, even the slightest bit of the filter screams "plastic" to me, but I need to develop a decent set of directions (starting points) so our proofing woman can learn and address what needs attention in a short amount of time (especially compared to the time it would take in ps to do it right). All I've found online is softening for people who don't really need it, nothing that deals with major acne. We purchased the software bundled with another piece we use often, so we want to stick with nik's color efex pro dynamic skin softener (or anything nik bundled in with that).</p>

<p>Thanks!</p>

<p>Not gonna do it. Sorry. The problem you describe is what you get when you try to smooth out blems with a softening program. You could try using a Hue/Sat layer, and desaturating the reds, then layer mask, stamp, run Nik, but I think the texture will still suck. I find I can get pretty quick with the Spot Healing brush, especially in CS5. (I make fun of the content-aware <em>fill</em>, but the brush works really well set to content-aware.) Keeping away from the plastic look may involve duping the spot-healed image layer, running Nik and then adjusting opacity.</p>

<p>just like Les says: healing brush, clone tool, possibly a bit of patch tool. perhaps a selective colour adjustment, lowering magenta slightly in the red and magenta sections, adjusting yellow and possibly black to get it right. i dont think NIK dynamic skin softener can do this, as you said the level of de-grunging required blows the rest of the skin to pieces, so you would have to be doing loads of masking, and if the image is going to require that much attention, you might as well do it with the proper tools and get much better results.</p>

<p>The retouching process in ps is not an issue for me. I'm fully confident in my abilities to retouch this sort of thing, but we were looking for a solution to integrate into our proofing workflow. We were looking at the nik software to keep the editing down to 1-2 minutes per image (again, this is only for proofs, not for final prints). We've actually settled on the "degrunge" method in ps instead of the nik software. Our proofing girl can load it, run the action, do the necessary adjustments and mask it in in just a couple of minutes, so it is working for us right now. </p>

<p>We proof these images, then if anything is ordered, it is properly retouched in ps for the final print. I would never use this method for anything other than the 4x5 proofs, but for that specific purpose , it is working fine. </p>

<p>Thanks for your replies.</p>

<p>Jen, take a look at Lee Varis' book, Skin <a rel="nofollow" href="http://www.amazon.com/dp/047004733X/nmphotonet-20" target="_blank">(link)</a>. In it, he describes a very nice (and fast) way to remove acne and blotchy / rosacea problems. Basically, one selects the "bad" colors and brings them closer to the desired "good" colors. He does it using the hue/sat tool in PS. I find it much faster and more accurate to accomplish the same thing using an inexpensive plugin called "Color Mechanic". All that is involved is (while within Color Mechanic), (a) click on a patch of good skin; (b) click on a patch of "bad" skin; © in their color wheel, pull an arrow half-way from the bad color to the good color. If the problem is severe or the lighting does not have a uniform color temp, repeat the above a few times. It takes a bit of practice, but once you / your assistant learns the technique it's exceptionally fast. No detailed masks or other selections are ever needed, and it preserves skin texture very well.</p>

<p>I described the process in a bit more detail in my Mar 03, 2010; 01:36 a.m. post in the thread, http://www.photo.net/digital-darkroom-forum/00VlrI.</p>

<p>Here is an example of using this technique on an image posted by another photo.netter in the portraits and fashion forum, asking for help:<br>

http://static.photo.net/attachments/bboard/00V/00V2BS-191693684.net<br>

The original poster improperly named the file. Just change the extension, "net", to "jpg" and it will work fine. For ease of comparison, I display it below.</p>

<p>A version that I processed using the technique described above is attached to the next message.</p>

<p>I hope you find this technique useful. It takes vastly longer to describe in words than it does to actually perform.</p>

<p>Tom M.</p>

<p> </p><div></div>

<p>Version retouched primarily using "Color Mechanic". I think I did a bit of sharpening and cc, as well.</p>

<p>Tom M</p>

<p>PS - I realize that this photo doesn't show the major acne situation that you asked about, but I don't want to publicly post pix of that sort. If you are interested, drop me a line and I'll privately email you a b4/after pair.</p><div></div>

<p>I don't do this kind of work, but I do use Color Mechanic for other purposes, and second the recommendation. You may find it useful.</p>

<p> </p>

<p>Jen, here's a acne / rosacea shot, b4 and after processing with Color Mechanic that I took that I don't think the subject would have any problems about using here. No color correction (other than that in Color Mechanic) or other processing other than cropping, compositing the 2 cropped images, and sharpening after downsizing.</p>

<p>Tom M</p>

<p> </p><div></div>

<p>Tom,</p>

<p>Thanks for that. I do use that technique often with people who have bad skin. I've integrated that into another technique and it is working great for our purpose (again, for proofs). But thanks for posting the detailed explanation again- it should prove useful to anyone who is unfamiliar with the technique. </p>

<p>Jen for what you describe I would also take a look at the following plugin. Portraiture from imagenomic.com. I think this does a better job than efex but you it pays to clean up first with the healing brush but does a good job without. For sure hand work is best but for a plugin I think this is the best I have seen.</p>

<p>Tom: Poor sod wound up with green hair. :)</p>

<p>Thank you for your comment, but it surprises me as I'm not seeing any green on my monitors.</p>

<p>However, to double check that my monitor calibration had not gone south, I selected the largest rectangular swatch of the subject's hair that I could get, averaged it, and read off the RGB values as 66, 51, 30. I then fed these numbers to one of the on-line converters RGB -> HSL / HSV converters (ie, <a href="http://web.forret.com/tools/color.asp?R=66&G=51&B=30">http://web.forret.com/tools/color.asp?R=66&G=51&B=30</a> ). The hue came out to 35 degrees.</p>

<p>If you look at a color circle (e.g., <a href="http://en.wikipedia.org/wiki/File:Hsv-hexagons-to-circles.svg">http://en.wikipedia.org/wiki/File:Hsv-hexagons-to-circles.svg</a> or <a href="http://www.workwithcolor.com/doughnut-color-picker-01.htm">http://www.workwithcolor.com/doughnut-color-picker-01.htm</a>), you will see that a hue of 35 degrees is just about half-way between red (0 degrees) and yellow (60 degrees), ie, is an orange-ish hue.</p>

<p>Thoughts?</p>

<p>Tom</p>

<p>PS - When I get home, I'll form separate averages for just the highlights and the shadows (of the same area of hair), just to make sure that the highlight and shadow hues are all in the orange range. </p>

 

<p>What I find with a quick look in Photoshop is a significant drop in the red value in the hair. From readings like 89-63-41 on the left to 77-63-38 on the right. This would indicate an increase in cyan, which is giving the hair a greenish cast.</p>

<p>@Les - At the risk of stating the obvious, a decrease in red / increase in cyan starting from a strongly orange/red color does not make the final color cyan or green. It only makes it less red. </p>

<p>I suspect I know what your are seeing, but I think it's a bit of an optical illusion caused when one's eye compares the adjusted color of the hair with his (nearby) still-ruddy face. It's an interesting phenomena. I wonder if anyone else has $0.02 to contribute on this?</p>

<p>Cheers, </p>

<p>Tom M.</p>

<p>PS - Last night, unfortunately I didn't get around to calculating the hue of the shadows separately from the hue of the hair highlights. Hopefully, I'll find a second to do this tonight. There might be some curve crossing going on and this more detailed analysis would show it (and validate what you are seeing), whereas a simple average wouldn't do so. I'll let you know how it comes out. I'm interested in this myself and certainly don't want to be giving folks green hair unless it's St. Patty's day. ;-)</p>

<p> </p>

<p>Les, I think I finally know what's going on.</p>

<p>Attached is a composite of three images generated from a cropped area of the image that I tweaked using Color Mechanic to help the subject's appearance. This crop shows just the hair and forehead of my rosacea-ridden subject.</p>

<p>1. The lowest section of the composite is just the cropped section of the my tweaked image.</p>

<p>2. The middle section of the composite is the most interesting. I converted the crop from sRGB to HSL color space, extracted out the hue channel, expanded the region from green (95 degrees) to red-orange (21 degrees), and displayed it as a gray scale map. By analyzing the map using a threshold adjustment, I determined that the darkest grays in the image correspond to a hue of 27 degrees, whereas the lightest grays correspond to a hue of 60 degrees (ie, pure yellow).</p>

<p>The three color scales above the image show the colors that a given hue can become, depending on the saturation and lightness. The lowest color scale is the brightest and most saturated. The middle color scale is medium brightness and saturation, and the top color scale is for the least saturated and least bright. Note, in the top scale, hue still varies over the same range from L to R, but the saturation and lightness are both fixed at 50%.</p>

<p>What these color scales show (at least to my eye, on my monitor, and in sRGB color space) is that as the saturation and brightness decrease for hues around 60 degrees, the colors look progressively more murky and even slightly greenish, ie, the yellow that is supposed to be at 60 degrees becomes more and more difficult to see. I suspect this is what you were experiencing.</p>

<p>3. Finally, the top section of the attached composite highlights those regions in the cropped image that have hues from about 50 to 60 degrees (ie, the yellow hues closest to the greens).</p>

<p>My conclusion is that there is no curve crossing or other problems with the tweak using Color Mechanic, but the low brightness and saturation of some areas of the hair can make these areas appear as a murky brown-orange-green, even though a detailed analysis shows there are no areas in the image greener than a mid-yellow hue.</p>

<p>Cheers,</p>

<p>Tom M</p>

<p>PS - Don't feel obliged to go through my analysis (...as you can tell, I'm hyper-analytical...), but to be honest, I learned a lot from going through this exercise to try to respond thoughtfully to your comment / observation.</p><div></div>