Hi, this is a new formula, not an update, as it is not compatible. If it does what I expect, I still need some testing, itïs a major step forward, because now two way border values are available, which means, there also is an out-border. When this point is reached, the previous iteration is in use again. I did this in order to have a better separation and to a certain degree avoid distorted structures. Theoretically it should be possible to define the size of the inside image with pixel accuracy, as all border values are fractional. Therefor I skipped the lake effect, it was not compatible with most images anyway. The tricky real(p5) input is now easier to handle and the initial routine is more effective. That BUG still was there in the beginning, but when I changed all border inputs to be fractional (secondary inputs), it seemed to be gone. "Gallery" is a zoom into one "neuralgic" spot and renders ok. In the previous frm I had to use "round" functions to get rid of the "artefacts" which fractint adds to the inputs. I think in some rare events these caused a different value of a variable and the consequent errors. Jonathan, does this sound reasonable? Anyway I keep my fingers crossed and will do some more testing. Fractal Greetings, Al.