Feedback delay networks (FDNs) are recursive filters,which are widely used for artificial reverberation and decorrela-tion. One central challenge in the design of FDNs is the generationof sufficient echo density in the impulse response without compro-mising the computational efficiency. In a previous contribution,we have demonstrated that the echo density of an FDN can beincreased by introducing so-called delay feedback matrices whereeach matrix entry is a scalar gain and a delay. In this contribu-tion, we generalize the feedback matrix to arbitrary lossless filterfeedback matrices (FFMs). As a special case, we propose the velvetfeedback matrix, which can create dense impulse responses at aminimal computational cost. Further, FFMs can be used to emulatethe scattering effects of non-specular reflections. We demonstratethe effectiveness of FFMs in terms of echo density and modaldistribution.