How can we eliminate false positives on a compare?

Running 3.2.0.5, we get a lot of false positives on our schema compare report. I do lots of ignores on the filter (Ignore stats, ignore fill factor and index padding, ignore permissions, ignore WITH NOCHECK, ignore file groups, ignore white space, and ignore extended properties). We also produce a report that does not include the objects that are the same (we take away the = on the reports), yet we get a huge report with 80% of the tables have no difference between the environments.

What is the solution? It seems quite counterproductive to use a tool to automate the compare if I then have to go into every table it outputs to a report in order to see if there is truly a difference or not. We are considering any upgrade, but if this does not work in our software, why should we upgrade and pay more money for perhaps the same problem? I think you'll agree this is fundamental to the product and should have been working by version 3.

Thanks in advance for any help. This is slowing our release process quite a bit. I'll be happy to send the htm report and the options used to create it if that will help.