High pass sharpening in the LAB lightness channel.

[thormod]

<p>Is the subject possible?<br>

I use the LAB color space more and more, and I almost always sharpen in the Lightness channel - except when I need to use High pass to sharpen. I can't make it work on the Lightness channel only. I can sharpen the image globally in LAB using the high pass technique, but then I sharpen the a and b channels as well.<br>

The difficulty is Overlaying the filter on the lightness channel.<br>

Any ideas?<br>

Tom</p>

<p> </p>

[digitaldog]

<p>Hope you realize there’s no reason to do that routine in Lab for sharpening, just <em>Fade Luminosity </em>provides the same goal with more control and without the data loss associated with two color space conversions. <br>

In terms of Lab, there’s a concurrent discussion about it (and the idea of editing on a flat file) over at RetouchPro you might want to look over: http://www.retouchpro.com/forums/photo-retouching/38384-better-start-retouch-flat-image.html</p>

[thormod]

<p>Thanks Andrew. <br>

This is very preliminary.<br>

I have previously used Fade luminosity in RGB. It works, but it also impacts on color. Guess its because there is no luminosity layer in RGB.<br>

I tried Fade highpass Luminosity in LAB with an ugly result when changing blend to Overlay.<br>

Out of curiosity, I tried Fade highpass Color. This gave me a beautifully sharpened B&W image. Perfect for blending in overlay. :) Don't ask me why yet, cuz I don't quite understand it.<br>

Thanks for the link. Have browsed and looks very interesting. <br>

Tom</p>

<p>I tried </p>

[digitaldog]

<blockquote>

<p>I have previously used Fade luminosity in RGB. It works, but it also impacts on color.</p>

</blockquote>

<p>It shouldn’t, that’s the point of that blend mode. </p>

<blockquote>

<p>I tried Fade highpass Luminosity in LAB with an ugly result when changing blend to Overlay.<br /></p>

</blockquote>

<p>Again, the idea is to produce the same results in RGB using a blend mode that separates color; luminosity. Pointless to do this in Lab. </p>

[Tim_Lookingbill]

<blockquote>

<p>I have previously used Fade luminosity in RGB. It works, but it also impacts on color.</p>

</blockquote>

<p>What does it do to color? Reduce saturation or increase saturation? That's all I've seen it do working on Raw files out of ACR. Are you working on camera Raws or jpegs or tiffs off a scanner?</p>

<p>I've output sharpened directly in ACR to compensate softening from downsizing to 700 pixels long end for web posting in Photoshop comparted to High Pass sharpening and they produce the same results which means they reduce overall saturation just a bit which is the exact opposite of the way it used to be High Pass or USM sharpening on tiff scans.</p>

<p>ACR's sharpening already has a Luminosity blend effect built in which can't be seen unless viewing at 100% zoom at least in ACR 4.6 (PV2003). With ACR 6.7 (CS5 PV2010) the previews are better at showing this desaturation and fine detail imbuing a sort of neutral white haze or lightening especially if over cranking Amount and Detail. I've reduced this effect somewhat by increasing Radius from 1 to 2 and leaving Detail at 25 in ACR 6.7.</p>

<p>I never see any color issues other than that. What color impact are you talking about?</p>

[Tim_Lookingbill]

<p>I solved your issue with using Luminosity blend using High Pass on an Overlay blend layers.</p>

<p>First make two duplicate layers of the image and apply the normal High Pass on an Overlay Blend mode to the top layer (a tip: try switching to Hard, Soft, Vivid, Linear Light which may improve the look). Merge that top layer to the duplicate layer and switch that layer's blend mode to Luminosity or Normal and notice the slight saturation changes which is all I get on my Raw images coming out of ACR at default sharpening settings.</p>

[digitaldog]

<blockquote>

<p>What does it do to color? Reduce saturation or increase saturation?</p>

</blockquote>

<p>Should do neither. Should produce basically the same results of converting to Lab and sharpening on just the Lstar channel which is done to avoid color artifacts. But I’m with you, far better to do this all at the raw capture stage. </p>

[Tim_Lookingbill]

<p>Color artifacts? Like what? I've never seen color artifacts attributed to sharpening?</p>

<p>I could understand seeing that maybe on low MP cameras like my 2006 6MP Pentax DSLR PEF's which hasn't been my experience, but I would think it would be impossible to see on more modern higher rez cameras of today.</p>

<p>Saturation induced issues with sharpening in ACR on my images are on account of preview zoom levels which don't show up viewing at 100% zoom, but at that magnification it's hard to attest overall saturation of the entire image until opened in Photoshop at 50-25% zoom view. 25% view in ACR 6.7 shows saturation as normal where it just looks a bit crispy increasing the Detail sharpening slider. Opening the finished image in Photoshop at 25% view looks less saturated on account of the crispy's appear more pronounced in the form of high frequency edge halos like on tree foliage that have gone to white.</p>

<p>From seeing this happen so often I finally figured out that ACR's default sharpening for my camera makes the image look natural viewed at 100% needing no further sharpening, but I just can't leave it alone and try to squeeze out as much sharpness as possible which creates the issue described above. I didn't realize there's a visual adaptive effect with sharpening much as it is with judging saturation, WB and contrast.</p>

[digitaldog]

<blockquote>

<p>Color artifacts? Like what? I've never seen color artifacts attributed to sharpening?</p>

</blockquote>

<p>Looks like colored noise around edges. Over do the sharpening on the edges and zoom in. Then use Fade Luminosity, the colors should disappear. Kind of like using NR on color noise in LR.<br>

And this 'technique' of sharpening predates raw quite a bit. So I wouldn’t expect you'd see it when sharpening in a good raw processor. In fact, the OP is talking about editing rendered images in PS, not raw. Hence converting to Lab. </p>

[Tim_Lookingbill]

<blockquote>

<p>In fact, the OP is talking about editing rendered images in PS, not raw.</p>

</blockquote>

<p>Images rendered from what? What software renders images with colored noise around edges that are brought out by sharpening in Photoshop in RGB space?</p>

<p>This topic is as old as Moses' toes going back at least a decade in discussions in which I participated and even back then I didn't see the point examining posted examples. It was so subtle and minute that it wasn't even worth the extra effort to go to these lengths of converting to Lab just to sharpen. But hey, that's not the first time that's happened on the internet.</p>

[thormod]

<p>Sorry Andrew, My Bad. I blame it on trying to many variations with to many layers, at the same time.<br>

Fade luminosity does not impact colors in RGB.</p>

<p>Thanks for your solution Tim. I'll get back to you when wife lets me open PS. We're on vacation and I am supposed to enjoy myself outdoors.</p>

[digitaldog]

<blockquote>

<p>Images rendered from what?</p>

</blockquote>

<p>You’ll have to ask the OP. Point is, lots of folks convert RGB data to Lab to sharpen the L channel and it’s totally unnecessary. The “<em>We <strong>must</strong> sharpen after converting to Lab</em>” is yet another digital imaging urban legend. <br /> And correct me if I’m wrong, but some people still have images from scanners or dog forbid JPEG’s out of a camera they want to sharpen where ‘rendering’ from raw isn’t pertinent.</p>

[thormod]

<p>I have never stated that “<em>We <strong>must</strong> sharpen after converting to Lab</em>” That does not make sence.<br>

Still, sharpening the lightness chan in LAB makes sence to the subjective me - some times.<br>

Most of the time I use USM in RGB, if not USM - then prolly HighPass in RGB.<br>

Yet I find myself often in LAB, and it saves time sharpening the L chan - now that I know how to do with HighPass also:)<br>

P.S. Remember, I'm a retired old fart. Did some BW dark room in my teens, but it has been "Point and Shoot" since the introduction of color. Took up "serious" photo around 2005 and am autodidact, thanks to the internet.<br>

I may claim strange things, but it will be out of ignorance - never malice!</p>

[thormod]

<p>So, now a shot at answering "Rendered from what" As far as I understand, in PS, it is the sharpening halos that are the issue. They can take on weird colors, both the Lighten one and the darken one. Thats why I at the moment think it is a good ideah to sharpen in LAB L. You don't impact on colors.</p>

[digitaldog]

<blockquote>

<p>I have never stated that “<em>We <strong>must</strong> sharpen after converting to Lab</em>” That does not make sence.</p>

</blockquote>

<p>I never said <strong>you</strong> did. </p>

[digitaldog]

<blockquote>

<p>As far as I understand, in PS, it is the sharpening halos that are the issue. They can take on weird colors, both the Lighten one and the darken one.</p>

</blockquote>

<p>They <em>can</em> but will not if you simply use the Luminosity Blend mode (fade) just after USM. You don't impact colors.</p>

[thormod]

<blockquote>

<p><strong>They <em>can</em> but will not if you simply use the Luminosity Blend mode (fade) just after USM. You don't impact colors.</strong></p>

</blockquote>

<p>You are probably right, because I never really understood "Fade Luminosity". Its like the "Blend if" sliders - a mystery</p>

<p>Tom</p>