I posted this in another area of the forum, but no one was able to give me any feedback. Perhaps here will be better.

I have written a shader to successfully do ambient occlusion in XNA and it has given me this result: https://gyazo.com/0bf260d8cdf9853a93440ca1a645eed7

The ambient occlusion here is not very intense and that is intentional. My question is should I recalculate my noise map every frame for the more "random" feel of indirect lighting? Or should I just use one static noise map like I currently am. Instead of even recalculating a noise map every time I could just create a coordinate offset every time for the sampling, I guess that's an option. Does anyone here have an opinion or some knowledge if this is commonly done? Or should I leave my shader as is?

I found that changing the random vector per frame ended up being too noisy, unless you have a strong blur implemented. Even with a static camera/scene, you'll still see the ao bouncing around because the random value changes. If you use the same random value, then in a static scene, the ao will also be static.