- scale-up and / or rotate image prior to converting it to vectors
- saving a smaller-sized image displayed by Google image search (GIS) as the preview of the larger-sized image found on the linked website
- scaling and / or rotating distorts the bitmap and that loss of fidelity transfers to the resulting vectors
- unlike Hollywood in which scaling up a bitmap makes it "more in focus" scaling digitally "smears" the image. For example if you have a 10x10 pixel image and each pixel is a different colour (100 in total) which you scale up to 20x20 you will still only have 100 colours but each one of them will be represented by a 2x2 square instead a 1x1 square.
The problem with number 2 is that the image is not as large as it could be. In one video I watched recently the image at the linked-to site was 1880 pixels wide but the one the presenter saved from the GIS preview was 483 wide. Yet he was sure that he was saving the higher-resolution version.
Sample images to make two points clear:
In summary:
- don't scale / rotate images; scale or rotate vectors instead
- take the time to get the best resolution image you can (go to a linked website or scan at a higher pixel per unit of measure)
- bonus: avoid jpeg if you can (don't save to this format when scanning)