I operate an online service for converting home videos to digital. I started it because I couldn't find a single company that was digitizing NTSC video the "right" way.
Many of the posts here focus on de-interlacing, which is an important part of the process, but you also need to be concerned about how de-interlacing affects the motion cadence of the image. NTSC video is 60i, giving you a nice, smooth, 60hz motion cadence when viewed on a monitor that handles the interlaced signal correctly. Most de-interlacing, however, involves combining fields in some way, resulting in 30 progressive frames per second. You maximize the resolution this way, but you lose half of your temporal resolution. The alternative is usually field doubling, which maintains temporal resolution, but at the expense of spatial resolution.
Then there is also the matter of digitizing the analog signal itself. VHS uses a control track for a timing and image calibration signal (setting 0% and 100% alternatively). However, this control track is at the very edge of the tape and easily damaged, resulting in image issues. For best image quality you should use a TBC to regenerate a good control track signal. Color in NTSC works by carefully controlled chromatic interference, a very clever hack for CRT monitors. But, for digital sampling, there are a couple different ways to calculate the colors- The easy way and the not easy way. The easy way is an approximation, and what most composite video capture devices do. But this results in colors that aren't quite accurate to the original and a loss of chromatic resolution. It is possible to do the brute force math to calculate the physics involved in the chromatic interference, and get colors that are accurate to the original, but this involves a lot of parrel calculations that all must be done in real time during the capture.
To address these issues with the capture service I provide, I developed my own digitizing hardware and deinterlacing algorithm. It corrects the color the accurate, but computationally expensive way upon digitization for superior results. The de-interlacing algorithm maintains both spatial and temporal resolution, generating new fields via a neural network AI to get a full 60p signal.
The website for my service is
www.MagicMemoriesBox.com- On the site there is a video comparison showing the results of our capture system versus a typical digitization with off the shelf hardware.