Hmm, we seem do be miles apart so let me try to clear up this whole process.
I started out in AI
I started out using a vectorized cartoony Mario and Luigi, with rendered raster Link and Samus.
The i took a lot of heat from board members (rightfully so) for not having the artwork mesh together, mixing cartoons and renders together.
My next task was to ditch the vectored Mario and Luigi's and find some nice rendered ones. I found a huge Raster Luigi render as you can see above. and a pretty big Raster Mario render.
Now that they were all Raster Renders , they seemed to mesh really well.
From here on I moved to FreeHand MX
Only problem was they still weren't high enough quality to use, especially at the size i wanted. That's when i started playing around with the Link Render, because that was the smallest image of the 4. I tested out different Re-sampling settings, (i.e. Bi-Cubic, Bi-Linear, Blurred, Nearest Neighbor) Turns out Bi-Cubic and Nearest Neighbor are terrible, Once re-sampled to 16X12 and 300 dpi, they were totally pixelated and flat out bad looking. But Bi-Linear and Blurred Looked great, Almost as good as the original image. I chose Bi-Linear because i thought Blurred had a negative connotation to it, although the looked virtually the same, it was just principal

. So anyway, Now I checked out the images and they looked good, they just had a little j-peg artifacting where 2 different colors meet up, ( links green hat/white background. This was easily taken care of with the Smudge tool.
Here is a little snippet so you can check to make sure my quality is up to snug
http://mywebpages.comcast.net/ldsfunaz/link-face-qualitycheck.bmp -LDS-