So, I'm not a graphics designer but I need some help with my science project on image tampering. so basically I want to know how do expert graphics software users actually crop a part precisely in an image, because as far as I know there is not definite boundary around the corners. So how do you remove those extra outside the boundary colors? So do you like compromise with the shape of the cropping part itself or what ?
Some terms need to be defined. The main one is Anti-aliasing.
A digital image is made by little squares. If the resolution of the device is not high enough we can see them. This is known as saw border, especially on diagonal lines.
One way we use to smooth diagonal lines is anti-aliasing, which consists in using pixels of an intermediate color between those adjacent colors.
Every digital photography on the planet and the majority of raster-based images have some level of anti-aliasing or blurriness in them.
So to cut an image there are some approaches.
This is defined by the threshold of the selection tool. If it is near zero, it will choose only some specific colors.
But adjusting it, and turning on an anti-alias option it starts selecting intermediate colors.
The first option of this selections is making them semi-transparent. (This example images actually are transparent so you can use them to see the effects on other backgrounds)This is what most software will do.
A more advanced algorithm could also remove the contamination of the adjacent color, it could take the values of the near color you actually want to see and transform those semitransparent pixels.
This is how that zone looks without the transparency. The borders are actually harsh again because the aliasing is given by the transparency, not the color.
This last type of anti-aliasing by transparency is more common when the image is generated directly as transparent, like using a 2D vector program or a 3D render one.
But in real life, there is a compromise between this perfect transparent anti-aliasing and "just enough". This is why it is preferable to work with high-resolution images, so these defects are tiny, and hopefully will not show in the finished product.
One common mistake some people do is taking still images with a green background thinking that this will be easier to remove in post.
If you are taking a photo that needs to be cut to remove the background, is a lot easier to choose a flat background of the similar tone of the final result, with enough difference from the main object.
For video is a bit different. If you are shooting video, the green screen is one good option. Some video compositing programs handle this color very well.