If you are looking at large continuous structures, an open mesh, 200 (lines/inch) or less is best. However, the support may have flaws or is damaged, therefore an elastic and strong film is required. Formvar with a thin coating of carbon provides these features.
At the other extreme, resolution and high contrast are major consideration. So if the final published magnification is higher than x600k, ideally no support is used under the specimen, since any support lowers contrast and resolution, however little that may be. Fibrous structures or other material that may stretch across a small hole in a support film can provide that opportunity. Lacy or holey films and plain, very fine mesh grids (>1000 mesh) could also be used. Without a support immediately under the object of interest, dark field electron microscopy is also possible.
Many applications fall between the above two extremes. When looking at small particles, finer 300 mesh grids give better support; this results in less film breaking, and less specimen movement is likely. Many users would find the common Formvar with thin carbon adequate, but thin (or no) Formvar and a thicker (self supporting) carbon film will be stronger, but less elastic.
For a given electron density a Formvar film shows slightly more intrinsic structure and is weaker than a carbon film. Pure carbon films, if evaporated onto mica and transferred to grids, do not adhere well and will float off on any solution. The preferred method for producing carbon films is through evaporation onto Formvar and a thicker layer of carbon, ~60nm. Then the user can dissolve the Formvar in the laboratory - retaining a little Formvar on the grid bars which anchor the carbon film.
Many failures using the negative staining method and support films relate to too much or too little retention of the negative stain on the grid. A large black area is not likely to reveal any fine structures and heats up quickly under the beam, destroying that square of material. Even distribution is achieved by repeated applications and blotting of the specimen grid and the negative stain. For the final blotting it is important not to remove too little or too much of the staining solution.