Video game designers do all they can to make their graphics look OK when rendered on video card hardware, but movie models have FAR higher polygon counts and texture resolutions than any real-time rendering hardware can handle. Also, film resolution is thousands of pixels by thousands of pixels doing full raytracing (as mentioned earlier - if a person stands next to another the light reflecting off one person subtly affects the other and vice versa.) In order to do lighting and subtle effects not possible with specialized rendering hardware often each frame is rendered over and over as many layers - one layer for lighting, one layer for shadows, one layer for this detail, one layer for that detail then they are composited.
Also, video cards are built with limits and certain capabilities. If movie makers want to do a new hair or water effect they need to be able to write a program to generate the image they want, not work within the confines of pre-determined hardware rendering abilities.
Games can do a lot to look danged good, but when it comes down to it, video cards can't begin to replicate what hollywood render farms do in the slightest. If you rendered Final Fantasy on an Nvidia card you'd wind up with something that looks like Reboot 😛