Rotating to any angle other than +/- 90 degrees or 180 degrees, always introduces some level of fuzziness. It is an intrinsic problem with raster/pixel data.

(Rotation by a multiple of 90 degrees is merely an axis transformation - the original pixels are simply moved to a new position. No interpolation required so there is no fuzziness (assuming correct handling of those special cases).)

You are correct, copying and rotating the original minimizes the effect and the errors are not propagated.

So if you had to rotate every stamp let's say 5 degree's you could in theory do this

Stamp at -85 on the elliptic, then (move) that stamp to 90.
Then you would create a new stamp at -80 and move to 90. Shouldn't this in theory place a stamp every 5 degrees and have them stay pretty sharp? And wouldn't this in theory work for any circle if it is indeed a perfect circle?

So a script could have a value choice by degrees in increments of a division of 360 degrees.And also a size for the elliptic.
The user places in a value and the script takes that value and creates a new stamp start number location for the first stamp placement.
The script creates a perfect circle, stamps the first stamp based on the user degree rotation input.
Say the user types a value of 15 degrees for rotation increments for each stamp.
The script takes that number, adds -90 to it and stamps the first stamp at -75.Because the script will ALWAYS rotate each stamp by 90 degrees it places or moves the first stamp to 15 degrees.Then it takes 15 and adds it to -75 and places the second stamp at -60.Then it adds 90 and moves the second stamp to 30 degrees.

Wouldn't that work?

Rod - Yes but...

A) the stamp wouldn't rotate around the center, keeping one aspect always pointed to the center. Draconian commented on this with the example I showed with the rectangles.

B) With concentric rings of stamped images, you quickly move into angles with multiple digits after the decimal point. But that doesn't really matter since the angles used internally are in radians not degrees, so they already are floating point numbers.

If the rotation of the stamp wasn't an issue, the easiest solution is simply to stamp at each location on the circle - the locations are easy enough to compute.

OK, I'm a fixin' to go get them thar boot on my feets cuz it's gettin' purty deep aroun' here.

Why don't we just blow off the dumb script. It really does sound like a real pain to do and I'm sure everyone has better things to do. It was just wishful thinking anyway.

the parameters seem to be limited to the point that the script would have to be too specific to make it a worthwhile effort. My feelin's aint hurt If we just let this one slide.