it's a process for all over dark places, like post #139254, or post #139212. mainly aimed for pics which has both colors & monochromes.fireattack said:
It works fine for me, thanks again (I've installed all the plug-ins required). But I have a question about the filter on shadows: sometimes I found it's not very necessary -- the dark area looks well already, and to run this process just killed some narrow lines in dark area (notice that the dark area has been filtered once, by the midtone filter). So maybe I should skip this step in these cases?
most full colors usually unnecessary this process, or use it with manual selection for better quality. changing the fuzzness of the color range selection form 100 into 50, and expand 3 => 2 (it's for unselect details) may also work
Is this how it should be looking:
I ended up adding a separate detail layer ("detail") to mask the blush detail back in on the left (in post #141319).
I ended up adding a separate detail layer ("detail") to mask the blush detail back in on the left (in post #141319).
The mask contract, etc. actions tend to fail: if nothing is selected (eg. no white areas for highlights), the whole action will stop with a "contract is not currently available" dialog. PSが呆けてる
the log to error can't work, too (ﾉ `Д´)ﾉ ~┻━┻petopeto said:
The mask contract, etc. actions tend to fail: if nothing is selected (eg. no white areas for highlights), the whole action will stop with a "contract is not currently available" dialog. PSが呆けてる
it usually occured on waterpaints which don't have blacks. I just turned off the shadow process before batching
I forgot to mention. the reason I added the process of shadow is mainly aimed for reduceing paper texture, not screening. the parameters depend on paper type.
also, highlights = bleed and dusts, midtones = screening.
also, highlights = bleed and dusts, midtones = screening.
http://www.mediafire.com/?gyouzmoxqez
a trial process to speed up filtering.
each filters are applied with a selection converted from a mask made by the apply image method.
a trial process to speed up filtering.
each filters are applied with a selection converted from a mask made by the apply image method.
A selection difference between "apply image"and "color range" in photoshop CS5
the 2nd is just selected by "color range", parameters are luminous 100 fuzziness 200. the 3rd image is masked with "apply image" using the original gradient image. the 4th is as same as the 2nd, but the localized color clusters option on.
since the color range can't select smoothly enough, I always use the apply image method now (that means I don't use http://moe.imouto.org/forum/show/9679 anymore), and if needed, convert it to a selection instead of using the color range (I can use "expand" a mask/selection area after convert a mask to a selection. http://moe.imouto.org/forum/show/9970 is using this method)
it's OK to use the color range for narrow ranges (fuzziness under 50), but in this case, an action error "nothing is selected" often occurs.
the 2nd is just selected by "color range", parameters are luminous 100 fuzziness 200. the 3rd image is masked with "apply image" using the original gradient image. the 4th is as same as the 2nd, but the localized color clusters option on.
since the color range can't select smoothly enough, I always use the apply image method now (that means I don't use http://moe.imouto.org/forum/show/9679 anymore), and if needed, convert it to a selection instead of using the color range (I can use "expand" a mask/selection area after convert a mask to a selection. http://moe.imouto.org/forum/show/9970 is using this method)
it's OK to use the color range for narrow ranges (fuzziness under 50), but in this case, an action error "nothing is selected" often occurs.
Unsharpmask causes white rings around dark areas, and black rings around bright areas. it's because USM just enhances contrast with same value on any luminosities.
but I think, on brights should be more enhanced contrast brighter, and on shadows should be more enhanced contrast more darker.
Therefore, I made an action of 2 passes USM2 with luminance masks (the way is as same as I showed before). The 1st pass has more amount+ and masking darks out, and 2nd pass has more amount- and masking brights out.
I applied it on the pool #1747. it had a heavily AM screening, so I filtered strongly.
but I think, on brights should be more enhanced contrast brighter, and on shadows should be more enhanced contrast more darker.
Therefore, I made an action of 2 passes USM2 with luminance masks (the way is as same as I showed before). The 1st pass has more amount+ and masking darks out, and 2nd pass has more amount- and masking brights out.
I applied it on the pool #1747. it had a heavily AM screening, so I filtered strongly.
NI & NW have terrible ringing effects even they're used only chroma channels.
Following images have a luminosity layer over the filtered layer copied from the original (duplicate layer =>change layermode luminosity => select background => filtering)
Each filtering strengths are exaggerated intentionally.
on pool #1747, I used this chroma-surfaceblur instead.
Following images have a luminosity layer over the filtered layer copied from the original (duplicate layer =>change layermode luminosity => select background => filtering)
Each filtering strengths are exaggerated intentionally.
on pool #1747, I used this chroma-surfaceblur instead.
While printing and scanning, gray (and black/white) colors get saturated a bit. on the contrary, vivid colors are lost their saturation.
for this reason, saturation should raised only high saturated values, and decreased low saturated values on a scanning process.
The smartcurve plugin is able to do this operation, but Photoshop can't do this in the default. you need to download HSB/HSL plugin from adobe (when you using CS4 or CS5), then change channels from RGB into HSB or HSL, and use curves to the S channel.
for this reason, saturation should raised only high saturated values, and decreased low saturated values on a scanning process.
The smartcurve plugin is able to do this operation, but Photoshop can't do this in the default. you need to download HSB/HSL plugin from adobe (when you using CS4 or CS5), then change channels from RGB into HSB or HSL, and use curves to the S channel.
If you do this, or any other colorspace conversions, be sure to work in 16-bit. Otherwise, the conversions will lose color depth.
test. lowered saturation on gray, raised saturation on vivid
Saturation of HSL color space seems a bit unnatural. the color of bright skin points a very high value, even though it looks whitey. maybe Saturation of HSL is just a sub parameter to define Luminosity.
On the contrary, Saturation of HSB seems follow natural sense in my eyes. but Brightness is a bit differ from natural sense. The brightness=255=true white, or the highest saturated color, coz brightness=max(R,G,B).
I suggest to use HSB when your primary thought is Saturation, and use HSL when your primary thought is Luminosity
On the contrary, Saturation of HSB seems follow natural sense in my eyes. but Brightness is a bit differ from natural sense. The brightness=255=true white, or the highest saturated color, coz brightness=max(R,G,B).
I suggest to use HSB when your primary thought is Saturation, and use HSL when your primary thought is Luminosity
Since changing saturation of HSB changes luminosity, it may need to readjust contrasts and gammas of RGB later.
now my color adjustment flowchart is:
1) adjust color balance, levels, gammas roughly in RGB mode while scanning
2) filtering
3) adjust the saturation curve in HSB
4) adjust the brightness curve in RGB
now my color adjustment flowchart is:
1) adjust color balance, levels, gammas roughly in RGB mode while scanning
2) filtering
3) adjust the saturation curve in HSB
4) adjust the brightness curve in RGB
Grays pick a hue because of the color error of scanners, so why not just correct that in tone curve adjustments? Desaturating all grayish areas sounds quite dangerous to me.
Try adjusting gray~black with RGB tonecurves. not only it unexpectedly changing values depend on papers, also it causes awful hue shift. I was triyng adjusting saturation with RGB tonecurves in years, but keep failing.kiowa said:
Grays pick a hue because of the color error of scanners, so why not just correct that in tone curve adjustments? Desaturating all grayish areas sounds quite dangerous to me.
anyway the primary aim of using HSB curves is to saturate only vivid colors keeping grayish colors and skin colors not saturated. you know, just adding saturation brings oversaturated skin colors.
take a look at post #152369 and other scans of this pool implemented a trial HSB curves adjustment action. red is very vivid but the skin hasn't reddish tones. RGB tonecurves can't get these results (with dull colored poor magazine scans)
Hmm, I think I get your point. Skin color is a quite delicate matter since we humans are quite sensitive to it. So It's best to avoid messing with skin colors while raising overall saturation. Also, It's no good sacrificing overall color quality just to make grays look right, so simply desaturating them produces better and quicker results. Sounds quite reasonable, I'll give it a try.
After filtered shadows with a luminance mask, select black (of the original layer) and clear the selection of the masked layer can avoid diminishing the back lines.
PS: Batching couldn't work when selection is 0 (ﾉ `Д´)ﾉ ~┻━┻
make a mask > apply > make a new mask > apply again might work
PS: Batching couldn't work when selection is 0 (ﾉ `Д´)ﾉ ~┻━┻
make a mask > apply > make a new mask > apply again might work
threshold:2 surface blur with a luminance mask is the most effective way to reduce bleed through more than any other plug-in filters afaik in my experience.
What dpi should we resize at before posting?
To check it, I resized down 400dpi FM screening raw at some smaller resolution, then resized back to 400dpi.
If resolution is meaningless big, the quality difference must be a little.
To check it, I resized down 400dpi FM screening raw at some smaller resolution, then resized back to 400dpi.
If resolution is meaningless big, the quality difference must be a little.
Changing saturation of HSB doesn't keep same luminosity, but there is a simple way to keep it. just add a duplicate layer and change the layer mode luminosity before doing HSB color adjustment (・ω・)ノ unlike default HSL saturation adjustment, it prevent oversaturating high luminous colors like yellow, enable to keep natural skin colors.
Additional spot color: fluorescent pink.
Doujinshi often uses additional spot colors outside of process color (CMYK). the one widely used is fluorescent pink.
There are many doujinshis using this color. some are CYK + fluorescent pink, others are CMYK + fluorescent pink.
The problem is it seems there is no color profiles to scan fluorescent colors. I sometimes wonder some doujin covers became very pinkish and their histogram is unexpectedly messed. I guess they're using fluorescent pink.
Doujinshi often uses additional spot colors outside of process color (CMYK). the one widely used is fluorescent pink.
There are many doujinshis using this color. some are CYK + fluorescent pink, others are CMYK + fluorescent pink.
The problem is it seems there is no color profiles to scan fluorescent colors. I sometimes wonder some doujin covers became very pinkish and their histogram is unexpectedly messed. I guess they're using fluorescent pink.
A common problem when making color adjustments is luminosity (brightness).
Here, we're making this image more red:
When we use Curves and increase Red, we're also making the image much brighter. To correct this, you can also reduce green and blue, but that's a pain.
Instead, we can increase red while maintaining luminosity. Create a curves layer, and set its blending mode to Color. Now, when we increase red, the brightness of the image stays about the same. Only changes to color are applied and not changes to luminosity.
Here, we're making this image more red:
When we use Curves and increase Red, we're also making the image much brighter. To correct this, you can also reduce green and blue, but that's a pain.
Instead, we can increase red while maintaining luminosity. Create a curves layer, and set its blending mode to Color. Now, when we increase red, the brightness of the image stays about the same. Only changes to color are applied and not changes to luminosity.
fireattack