Easy-to-follow video tutorials help you learn software, creative, and business skills.Become a member
Once again for those you who are fans of sharpening theory, I've got another theory exercise here for you. I was telling you at the end of the previous exercise how the More Accurate checkbox is based on the notion of a Convolution Kernel. A Convolution Kernel allows you to compare neighboring pixels to each other and exaggerate their differences or downplay their differences as well if you want it to. It's a primitive sharpening effect. It does a great job of bringing out noise inside of a digital photograph, which is one of the reasons you don't really want to work this way, but it also does a great job of sharpening the minutia in the image, which is what the More Accurate checkbox does.
So I am to going to show you a Convolution Kernel. I am not suggesting this is the way you work once again, just a little bit of theory I am going to show you a Convolution Kernel and then we'll see how it compares to the better More Accurate checkbox. I am working inside the test shapes.PSD image which is found inside, once again, inside the 03 sharpen filters folder. I am going to go up to the Filter menu and I am going to choose Other and I am going to choose this guy right here, Custom. The Custom filter allows you to multiply the luminance levels of pixels inside of an image and then add them together or subtract them from each other. Do all this wacky stuff.
Here's the idea. It's little hard to understand if you've never seen it before' which 99% of Photoshop users havent. Basically whats going on is, remember I was telling you that all of the sharpening filters inside of Photoshop and Gaussian Blur, the Blur filters as well and the Averaging filters, are actually scrubbing through the image. So you imagine all these little scrubbing bubbles going through the image and at any nanosecond of time you are evaluating a single pixel inside the image and trying to decide how to change that pixel or at least that's what Photoshop is doing, its making a decision based on the luminance level of that pixel and the luminance level of the neighboring pixels on a channel-by-channel basis incidentally.
So whats going on inside of this 5x5 grid is were seeing a grid of 25 pixels. So 5 pixels wide 5 pixels tall, right there in the center that's that one pixel that's being evaluated at any given nanosecond in time, and right now according to the default settings we are seeing, multiply the luminance level of that one pixel by 5 and then look at the pixel directly above it and multiply it times -1, so subtract its luminance level and take the guy directly to the right and subtract him and take the guy below and subtract him and take the guy immediately to the left and subtract him.
So if you do the math here, which is pretty simple: 5-1-1-1-1, that's 5 minus 4, so that's 1. So in other words we are attempting to maintain a consistent level of brightness inside of the image. If I were to raise this value to six, so the math is of, its now 6-1-1-1-1, so 6 minus 4, it gives you 2. We're essentially doubling the brightness of the image. If I were to take this value down to 4, so that the sum is now 0 instead of 1, we are darkening the image.
Don't worry about Scale and Offset. This is just going to multiply the overall effect and this one is going to add or subtract luminance levels, you typically don't want to work with them, you want to leave them alone. Actually you probably want to give this filter the path in general, but you can't, there are things you can do with it, you can create these really cool like sort of Color Emboss effects if you want to and you can do some blurring and you can enter all kinds of values inside of these various option boxes and play it to your hearts content. All you want to make sure, those of you who are comfortable with arithmetic, you just want to make sure that all your numbers whether positI've or negatI've they all add up to one.
So the end result is one, so I could enter 9 here for example and then a -1 here and a -1 here and a -1 here and a -1 here and now we have eight -1s, so we have 9 minus 8 and that's going to give us 1. Once again, so we are maintaining a consistent level of brightness. Now when you have a positive value in the middle and negative value surrounding it, you get a sharpening effect and so the default value, I'll go ahead and press the Alt key or Option key in order to change the Cancel button to Reset and I'll click on it and that will give us the original values which are 5 surrounded by -1 that gives you the most basic sharpening effect that you can apply essentially inside of this dialog box, so this is the lowest level sharpening and you can see how Photoshop is going in there on a really granular basis.
This is before and this is after, and applying the sharpening on a pixel-by-pixel level essentially. So theres no such thing as radius or halos or any of that stuff, you don't have that kind of control. You are just exaggerating pixel differences. So I'll go ahead and click OK so that we have that saved up and I'll go ahead and zoom-in, and when I say, "Its saved up, its stored now inside the History palette, so we can come back to it. Then I am going to go to the File menu and I am going to choose the Revert command so I am bringing back the original version of the image inside History so we can work with a flat file which is important just because for demonstration proposes I am using a flat file.
Later we'll be using layers which is a better way to work. So having reverted the image, now I am going to go up to the Filter menu, I am going to choose Sharpen and then I am going to choose Smart Sharpen. The idea is, we're going to compare the effect of the More Accurate checkbox, and this time I am going to change the Amount value something like 200%. I think it would work better for us. Lets take this Radius value down to 2 pixels, we've got Lens Blur, now lets turn on More Accurate and you'll see the difference, you might be able to see it there, lets go ahead and zoom-in a little bit so that you can see it little more closely.
So this is what this image looks like without More Accurate, this is what it looks like with More Accurate. So it's bringing out even more information inside the texture detail. So it's over-highlighting essentially the texture and it's not doing much to the big details at all. So its highlighting that high-frequency information and ignoring the low-frequency information that's already been treated by the other values here and as a result we are making this sort of textury effect look almost leathery inside of this particular image.
So I would say with this image More Accurate could be a good thing if you decide to go there out, because this isn't a portrait shot, it's a high frequency shot, so that's a good thing. Whenever you want to exaggerate the high-frequency details, that's when you turn on More Accurate. And well see more examples of this when we take a look at sharpening details in a later Chapter, but for now there it is, I am going to click OK and just so that we can compare the two effects I am going to move this guy over a little bit, bring up the History palette and this is what the Custom Convolution Kernel looks like and this is what the Smart Sharpen effect looks like.
So if you were to compare them very closely you would see that the Custom effect is very detailed, but its also very choppy, we have some razor like lines going on and we have some jagged transitions as well, whereas Smart Sharpen is a little more rounded, its a little more defused and its ultI'mately a better effect, but again, More Accurate hails from days of Convolution Kernels because it gives us that control over sharpening the high-frequency details inside of the image.
Get unlimited access to all courses for just $25/month.Become a member
180 Video lessons · 76678 Viewers
64 Video lessons · 94740 Viewers
86 Video lessons · 62142 Viewers
103 Video lessons · 31583 Viewers
Access exercise files from a button right under the course name.
Search within course videos and transcripts, and jump right to the results.
Remove icons showing you already watched videos if you want to start over.
Make the video wide, narrow, full-screen, or pop the player out of the page into its own window.
Click on text in the transcript to jump to that spot in the video. As the video plays, the relevant spot in the transcript will be highlighted.