Frequently Asked Questions
1. Should I keep the marked area as small as possible?
2. Should I put in lots of lines?
3. Is it better to use lots of individual dots?
4. How far away from the marked area should I place the dots?
5. I am zooming into the image to mark and add the lines. How do I move the image without unzooming?
6. What’s the .NET Framework requirement and do I have it ?
1. Should I keep the marked area as small as possible by zooming in to high magnification and reducing the marker size so that I could follow the outline exactly or is a bit of overlap better?
Usually it is better to leave a little overlap. It sometimes happens that if you mark the area too precisely, then there's actually a shadow or a thin edge left in the unmarked area - and it stays in the result.
2. Should I put in lots of lines, both for the individual objects and their outlines or even areas of similar sky? This results in lines crossing one another.
The guiding lines are usually best for connecting the edges (i.e. continuation of outline, or object edge etc). Lines are sometimes useful for connecting areas of similar sky when the marked area goes through a lot of different sky gradients and you want to preserve their layout etc. Most of the time, it is probably best to not cross the lines.
This is how they work : For every individual guiding line (it can consist of multiple connected straight lines), part of it goes through marked and part of it through unmarked area - and the unmarked (known) background pixels along the line are pasted into the marked (unknown) area along the line.
3. Is it better to use lots of individual dots (line end-points) along the line when the edge is curved and not a straight line? Is it a good idea to use very short lines ?
The individual lines (not the line end-points) are used for pasting pixels. I.e., if half of the line goes through unmarked area, then those pixel patches will be pasted into the line pixels in the marked area. If the line consists of several line segments then similarly the pasting extends over all line segments, and the pixel patches are also rotated based on the angles of the individual line segments. So it is probably better not to have the lines too short.
4. How far away from the marked area should I place the dots? If there is a very defined straight edge, is it better to extend the lines further?
It is best to start/end the lines further away from the marked area as long as they still cover the kind of pixels/edge/texture which should be extended along the line in the marked area. The algorithm then has more pixels to sample from for this guided completion.
For example if you only cover a very short segment with the line outside of marked area, then the filled structure along the line in the marked area will tend to be very repetitive since it doesn't have enough sources where to sample from.
5. I am zooming into the image to mark and add the lines. A "hand" tool would be useful in order to drag the image around rather than having to zoom out and back in again which is how I had to move from one part of the image to another.
If you are in a zoomed-in mode, unclick the line/mark icons, then the mouse should show a pan cursor over the image and you can drag and move the image.
6. What’s the .NET Framework requirement and do I have it ?
If you have Windows Vista or Windows 7, then you automatically have this so you don’t have to worry. If you have Windows XP then it is still a good chance you already have it since many software applications rely on the .NET Framework, if not then Photoupz installer will automatically detect this and open a Microsoft download page for the .NET Framework. You can then easily install it.
