HVS ColorGIF 2.0 has many new features that have been designed specifically to aid the designer in minimizing the size of GIF files while maintaining excellent quality. The following suggestions will help you get the best results.
The HVS algorithm is designed to reduce the image to flat areas of color without the eye perceiving those areas as visible bands. Flat areas compress in LZW much better than complex dithered regions.
The LZW algorithm used in GIF compression is very sensitive to the number of colors in the color table, because this determines how many bits it allocates to store its codes. If possible, try to get the color count down to the next lower power of two. For example, if you're using 80 colors, 64 will probably cut the size significantly. The live preview in HVS ColorGIF has been designed to help you see the effects of using different color counts on the same RGB image, and immediately get feedback on image quality and resulting GIF size.
By dithering within a selection (such as a foreground gradient that is highly noticeable), you can often maintain visual image quality with far fewer colors. This will cut GIF size dramatically. When you're happy with the visual result, be sure to select the Delete Unused Colors option. Use as little dithering as possible. HVS ColorGIF is one of the few tools that allows you to interactively control Dither Intensity.
This can be a fun exercise that shows the power of HVS ColorGIF to improve web performance. The next time you're at a web page with banners at the top, save one of the banner GIFs on your local disk using your browser. Bring it in to your paint program and convert it to RGB. Run the HVS ColorGIF dialog, and see how many colors are actually in the image. What we frequently find is that the designer has included a color table of 64, 128 or even 256 colors, but the image only uses a few. There are often a few flat colors in the banner, but anti-aliased text or drop shadows unnecessarily take up a bunch of colors.
HVS ColorGIF's algorithm analyzes images to determine which colors the eye is actually going to notice, so it's great for simplifying the gradients found in text and drop shadows. Try reducing the image to a lot less colors (we've gone down to 8 in some banners) and see how much you can crunch the banner down. It's not uncommon to end up with an image 2-3 times smaller than the one that's busy soaking up web bandwidth on someone's high-profile commercial page. Finally, write to the webmaster and tell them how much faster their page would load if they used HVS ColorGIF. (OK, you don't have to do that part, but you might earn eternal sainthood from us and the web community if you did).
Note that on Macs, only the data fork size is important. You cannot get an accurate GIF size from the Finder. Use a shareware utility called Snitch if you want to see exact sizes. The GIF size value shown in the HVS ColorGIF preview is accurate.