When you scan something or have a file as a resolution of 300 dpi, that is usally great for printing, keeping in mind that you print it at 100%, the original size.
If you try to enlarge it you are reducing the dpi (dots per inch). There is no safe way to do this.
I think if you're trying to vectorise a raster image I think you are wasting your time. It would be better if you had the image at 300 dpi at the size you need to print it.
When you scan it you can scan up to 1200 dpi. That means you can print it at 4 times the original size, but it will be 300 dpi on ouput.
Similarly, if you scan it at 600 dpi you can get a print twice that size, at 300 dpi on output.
Vector images are a series of mathematical data. Here is a file that is for a circle of shamrocks it's the Illustrator file only it's in txt format and you can see all the maths going on.
The advantage with this is because every single piece of information is made up of a mathematical equation it only takes a simple multiplication to resize it to anysize needed.
Unlike a raster image, which is something that is scanned or used in pixel editing software, like photoshop. It's made up of pixels. Millions of pixels.
A pixel is the smallest amount of information that screen can hold. Pixels can vary in sizes, there is no one size for a pixel. If you have 300 dpi, or when on screen called PPI (pixels per inch) then you have a raster image. It goes through a process of interpolation when resized, and is very noticeable when you scale it up size.
All interpolation is maths too. But for pixels you can't stretch or skew them or multiply pixels. The application (phtoshop or other) has to keep the relation between the pixels the same. So if you have 300 dpi image and try to print it out at twice the size then what happens is that the pixels move twice as far apart and that leaves a gap in the image. So what the computer will do is work out the value of the pixels inbetween (by value I mean the colour content based between one pixel and the pixel beside it, it will insert a pixel based on the other two pixels).
There really would be no advantage of trying to vectorise scans.
The only time I would do this is if it's a logo that does not look like a photo. I would vectorise that in Illustrator. I'd draw it again, tracing over it with the pen tool.
But for an image like a sky or something, or someone's face, a room or anything that is life like, I would not try to vectorise at all. Plus when you do vectorise using the program like Streamline, or Illustrators own Live Trace button. The DPI is also traced. So even when you have it all vectorised, and you blow the picture up, you will still see a loss in quality.
The reason for this is that anything that is printed has a pattern, very small pattern, you can see this using a linen tester (
The pattern is a result of the printing process. It's called a ROSETTE. It's produced by:
Angles at which screens intersect with the horizontal line of the press sheet. The common screen angles for separations are black 45 degree, magenta 75 degree, yellow 90 degree and cyan 105 degree.
So if you were to scan an image you're scanning in all those little dots that are printed. You can see them when you scan. You can descreen the image, but it leaves a slight blur and you have to use an unsharpen mask or some sort of sharpening so it looks less blurry. But when you do this and try to vectorise it then you are vectorising the blur too.
The angles have to cross or they overlap, the colours can't be printed on top of each other, as they are transparent and they would combine to make a different colour, which is undesireable.
The only thing that doesn't have this rosette pattern is an original photograph. Something that you had processed from film.
Sorry for the lenghty explanation.
