a rambly quasi tutorial
So mylengrave asked how I go about resizing my images for icons and keeping details.
Under the cut is a lot of me rambling with very few pictures. Lots of telling. Not a lot of showing.
First first thing. Always save as PNG. That way JPEG never eats your graphic.
But the real first thing is I never use crop. I find it ends up sharpening the details of your image before you've even started colouring. When you're doing vibrant work, things tend to get grainy and pixellated easily. You don't want to start off with an already sharp base. Its easier to sharpen a slightly blurry image than to try and ease an oversharp one and still maintain detail.
I always take the original image I want and open up a new file at the size I want (almost always a 100 pixel square). I then drag the image into the new file and go about cropping by moving it around your 100px canvas. For resizing my images I use Free Transform (Command T or Edit > Free Transform, at least in CS6 on Mac). For resizing without skewing, always maintain aspect ratio. I also always start by cutting in halves or quarters. Shrinking down to 75%, 50%, 25%. Sometimes I'll go to 50% and then to 75%, it depends what I'm doing with the icon. If I don't end up with the image at the size I want it, then I'll start going to 80% or 85%. Some of that is just me being super anal, but I really do think that you maintain more detail when you're not shrinking to 67.3%.
For example, on this icon of Clary:
On the left we have the base I used the crop tool on. On the right we have the free transform base.
vs
I just whipped up a guestimation of the colouring on that icon and applied it to both the crop (left) and the transform (right).
vs
For my icons, once I've "cropped" them, I tend to make a duplicate layer (always dick around on an alternate layer, that way you don't get stuck when you run out of history) which I then use Unsharp Mask on. The settings I use for unsharp mask is almost always Amount 100%, Radius 0.3 Pixels, and Threshold 0 levels. If its an extra blurry picture I might go up to 0.5 pixels but never any higher. By doing your sharpening on a duplicate level you can play with the opacity of that layer, and even mask out parts that were sharp enough to begin with.
As you can see with the cropped icon on the left, treating it as a regular icon of mine results in it being oversharp. Especially around her hair. Whereas I could continue to add some sharpening to parts of her face in the right icon. I actually added another masked layer that I sharpened for this PSD. (Download Here.)
A slightly shorter rambling for mid to far crops like in these icons:
For both of these icons they don't have their original backgrounds. I prep my icons the same as I previously said, transforming down to the size I want. I find for extremely far crops you do lose facial details, I have not learned the magic way to keep them.
The best way to keep the general details of your subject is to resize and sharpen before you start cutting out. Again, I use duplicate layers and masks to accomplish removing the subject from their background, that way if you screw up you can go back and fix it.
As for using Topaz, I almost never use it. I find that it smooths out details and causes you to lose definition. A time I might use it is if the cap and colouring I was using in a close crop resulted in an overpixellated/sharp face. I'm pretty sure I used Topaz in this icon just for her cheek to smooth out some of the graininess. I definitely have some of a topaz layer present in the side of her hair, as well as I drew in some highlights to it (not that you would know).
Hopefully this completely unhelpful non tutorial is even vaguely helpful!
For keeping quality and definition in icons after editing them, I always try and use the most high quality screencaps possible. AND PNG FORMAT FOREVER.
Under the cut is a lot of me rambling with very few pictures. Lots of telling. Not a lot of showing.
First first thing. Always save as PNG. That way JPEG never eats your graphic.
But the real first thing is I never use crop. I find it ends up sharpening the details of your image before you've even started colouring. When you're doing vibrant work, things tend to get grainy and pixellated easily. You don't want to start off with an already sharp base. Its easier to sharpen a slightly blurry image than to try and ease an oversharp one and still maintain detail.
I always take the original image I want and open up a new file at the size I want (almost always a 100 pixel square). I then drag the image into the new file and go about cropping by moving it around your 100px canvas. For resizing my images I use Free Transform (Command T or Edit > Free Transform, at least in CS6 on Mac). For resizing without skewing, always maintain aspect ratio. I also always start by cutting in halves or quarters. Shrinking down to 75%, 50%, 25%. Sometimes I'll go to 50% and then to 75%, it depends what I'm doing with the icon. If I don't end up with the image at the size I want it, then I'll start going to 80% or 85%. Some of that is just me being super anal, but I really do think that you maintain more detail when you're not shrinking to 67.3%.
For example, on this icon of Clary:
On the left we have the base I used the crop tool on. On the right we have the free transform base.
vs
I just whipped up a guestimation of the colouring on that icon and applied it to both the crop (left) and the transform (right).
vs
For my icons, once I've "cropped" them, I tend to make a duplicate layer (always dick around on an alternate layer, that way you don't get stuck when you run out of history) which I then use Unsharp Mask on. The settings I use for unsharp mask is almost always Amount 100%, Radius 0.3 Pixels, and Threshold 0 levels. If its an extra blurry picture I might go up to 0.5 pixels but never any higher. By doing your sharpening on a duplicate level you can play with the opacity of that layer, and even mask out parts that were sharp enough to begin with.
As you can see with the cropped icon on the left, treating it as a regular icon of mine results in it being oversharp. Especially around her hair. Whereas I could continue to add some sharpening to parts of her face in the right icon. I actually added another masked layer that I sharpened for this PSD. (Download Here.)
A slightly shorter rambling for mid to far crops like in these icons:
For both of these icons they don't have their original backgrounds. I prep my icons the same as I previously said, transforming down to the size I want. I find for extremely far crops you do lose facial details, I have not learned the magic way to keep them.
The best way to keep the general details of your subject is to resize and sharpen before you start cutting out. Again, I use duplicate layers and masks to accomplish removing the subject from their background, that way if you screw up you can go back and fix it.
As for using Topaz, I almost never use it. I find that it smooths out details and causes you to lose definition. A time I might use it is if the cap and colouring I was using in a close crop resulted in an overpixellated/sharp face. I'm pretty sure I used Topaz in this icon just for her cheek to smooth out some of the graininess. I definitely have some of a topaz layer present in the side of her hair, as well as I drew in some highlights to it (not that you would know).
Hopefully this completely unhelpful non tutorial is even vaguely helpful!
For keeping quality and definition in icons after editing them, I always try and use the most high quality screencaps possible. AND PNG FORMAT FOREVER.