Conandy wrote:When you talk about a "high-rez aliased edge" vs "anti-aliased edge", can you tell me exactly what you mean, in terms of preparing the image initially?
Still early stages of learning, and I am extremely grateful for the time you all are giving me here.
Well, an aliased image is a hard-edged image within no transitional pixels to smooth it out. Take a look at the "A" letters on the wiki
here for an example.
A simplified explanation... when you downsample an image in Photoshop, it's very good at anti-aliasing images with complex mathematical formulas. MT is, comparatively, not very good. MT is probably aggressively discarding pixels for display, and the "hard" edges get more and more dominant/visible. So, you really want a softer edge so that even though it's discarding pixels, it's got more transitional pixels to use.
In *addition*, the images you're showing are particularly difficult to use because in addition to the edges of the token being high contrast single pixel edges, there isn't just an OUTSIDE edge... you've got an INSIDE edge all around the figure that MT can't anti-alias with anything because you have transparency inside the token. It's like a worst case scenario. Especially look at the edges of Serena... her weapon, her arms, and the thin gold rim around the token base. Imagine what happens when you're showing 1/16th of the image and you've had to make a decision 16 times to toss out information to display (pretty much what Phergus was explaining above.) It will get ugly fast. Much faster if there are high contrast, irregular edges to work with. Made worse, of course, on a high contrast display (i.e. the sharpness setting which creates contrast on edges.)
My issue wasn't with Phergus's analysis, on the face of it he's right, but I do think that the solution he offered was a bit draconian and might be discouraging to people trying to use MT on a TV. There are extenuating factors and some middle-ground options (past exploring all software/hardware optimizations along the way.)
So, some tips since you asked:
--Take a look at Token Tool here for example of more simplified token frames.
--I'd say to try not to use more than 4x the native resolution of the display size. So HD is 1920 pixels wide and you want a 40 inch width (approx 48 pixels an inch). So, rounding, I'd make them no more than 200dpi. I like to use 50-100 DPI for maps and 200 dpi for tokens.
--Think about the detail you're trying to display in 1". The 420 images you had shouldn't just be downsized to 200. You should consider the content of the 1". At the very least I'd probably cut out the middle 200 pixels and use that as the base. You won't see tiny rice bowls and gilded filigree detailing on the token frame inside a 1" window. Again, Phergus is totally right about that.
--Also, don't use transparency inside the token.
--Try to find lower contrast, softer edge images.
--Use your nifty high detail images as portraits if you want... not so good for tokens.
I did this to quickly to illustrate. Check out this token to see how it ends up looking. A happier medium between the TV and computer?
Keep in mind that this is still a 200dpi token. The outer edge of what I suggested above. I didn't down-sample your token, so the edge of the figure still isn't ideal for dithering. But I went ahead and used the 200x200 pixels in the middle of your original. Added a basic background (so MT had something to anti-alias that ugly edge with.) The resolution is still going to be dithered down into the 1" you want (i.e not displaying about 38000 pixels or 95% of the full image data!) I know that sounds crazy, but it should look a lot better/usable and still look decent on your laptop where you're zooming in closer.
Everyone has a different threshold for what's acceptable. You can try the token out at 50, 100, and 150 to see what that is for you.