DrSlony wrote:Yeah but they're extremely inefficient. Each marker, according to the documentation, targets 1 pixel, and antighosting-blend then avoids cutting through whatever it thinks is the object that pixel denotes. That's the problem, you can't tell it what the object is, so the only safe way is to spatter plenty of them.
Already opened a feature request for a polygonal mask.
DrSlony wrote:According to the documentation it does, and I'm guessing that antighost-blend uses some kind of seam carving or other shape-detection to cut along meaningful edges. Then, if it decides to discard a specific shape, and there happens to be a "keep" marker on it, it looks for a different edge.
I pointed out the documentation in post 1, and the rest is through observation.
DrSlony wrote:I didn't write the effect was on a single pixel, I wrote the markers, despite being big, targeted only a single pixel. It would be more intuitive and easier to use if we could mask the (un)wanted area, not to rely on pure luck that the algo will cut the exact region we had in mind, but that's a feature-request off-topic for this thread.
DrSlony wrote:It's unfortunate that there is 0 response from Kolor regarding this simple yet very time-saving feature.
Related requests or requests resolved by a polygonal mask tool:
DrSlony wrote:Wouldn't you consider them all obsolete with a polygonal masking tool?
DrSlony wrote: I did use more markers than needed because one never knows how many are needed. When I used less, placed as the documentation describes, there were patches appearing where I did not want them. I'd rather spray 50 clicks in 20 seconds and be sure, than spend many minutes trying to find those magical fewer spots.
DrSlony wrote:The algorithm examines the clicked pixel and masks the contiguous area. What it deems to be the area is never revealed to the user - the documentation hints that it's an area of uniform color and, as I understand it, bordered by a high contrast edge. You can keep placing single points and waiting for the final preview to update, hoping each time that your idea of an "area" is the same as the program's, but in the end it's still a guessing game.
DrSlony wrote:It is not you I seek to convince about the validity of requesting a masking tool that lets us mask a human-defined area instead of a point.
Users browsing this forum: marzipano and 1 guest