Paparazzi IMU question

[PB4]

haven't yet found the answer to this question, I know paparazzi can use both IR sensor and accelerometers at the same time but why use IR sensors if accelerometers can give you attitude ?
I am only interested in attitude information, I don't really like IR sensors because of their inherent limitations (clouds, flying next to a cliff..)
I know the main problems with acc/gyro is that error gets bigger over time, thus I was thinking of an hybrid system where IR sensors would be used to fine tune the acc/gyro derived attitude when IR sensor info is considered reliable (for example when vertical diff is greater than threshold, which would indicate out of clouds) so equiped my setup could fly through clouds for a limit of time defined by the error rate of acc/gyros derived attitude.

[zik]

I think you've missed something here - accelerometers don't give you attitude. They give you acceleration, which when you're doing a coordinated turn doesn't point down to the ground, it points down relative to the plane.

Believe me, if accelerometers gave attitude everyone would be using them instead of the more complex and more expensive stuff we do use.

[PB4]

ok sorry I took shortcuts in my post, by acc/gyro I meant using a set of accelerometers, Gyrometers and magnetometers, then by integration compute attitude.
If only attitude is needed I agree we can get rid of accelerometers since only Gyros should be enough, however I read about some using accelerometers only thus I don't discard them just yet.
So to rephrase, can I use paparazzi with only an IMU do compute attitude? if so what is the typical bias based on the most comon system (IDG300 is it?), and then, if I throw in some IR sensors would I be able to still fly based on IMU info only if IR sensor information becomes unreliable (entering cloud) ?

I realise a good small bias IMU would be horribly expensive for us hobbyists, thus why not take benefit of both systems ?