A decent cyclone with the design of the Penn State cyclone should be sufficient to capture 99.99% of material. Some of the shorter cyclones as seen on the new portable models from Grizzly and Laguna don't function as well, but still manage to capture 99% or so.
When we originally went to an approach of venting outside there was a series of tests to make sure that no significant chips were vented outside because of concerns about coating the neighbor's cars and boat. None were seen at that time. I'm also not aware of a way for a functioning cyclone to degrade to the point that the vented air would have more chips other than a collection backup.
Generally speaking the chips that don't fall out during the cyclone separation are going to be very small to the point of being hard to see with the naked eye. They might appear in large quantities, but otherwise should be invisible. Further they're going to be easily dispersed by the air currents outside the building.
A final filter is a compromise of the air quality to avoid problems with venting conditioned air outside the workshop. The best solution from a health prospective is to always vent outside. If excess heating costs is a concern then it might make sense to vent internally via a filter. Even then it's been my experience that the thermal mass of the workshop is usually sufficient to overcome the change in air circulation, unless the cyclone is left on for a long time.