While some think a quantum computer will give a stronger basis for random numbers, you don't need a quantum computer for quantum randomness. Given that some answers have already mentioned QKD, I'll focus instead on the complexity classes of quantum computing, and how they might be used to build cryptographic primitives.
Cryptography in general bases security claims on costs associated with cryptanalysis. For example, the Learning with Errors concept reduces a worst-case approximation to an average case for being able to solve lattice problems. It's understood that solving a lattice problem is "hard," meaning it is computationally infeasible to solve the problem in practice.
The most commonly referenced complexity classes are some form of $P$ and $NP$. These denote the time it takes to solve a respective problem, either in polynomial time or non-polynomial time.
$BQP$ is one of a few quantum complexity classes, which stands for bounded-error quantum polynomial time. While it is understood that:
$P \subseteq BPP \subseteq BQP \subseteq AWPP \subseteq PP \subseteq PSPACE$
The relationship between $BQP$ and $NP$ isn't fully understood yet. Shor's algorithm, which is one of the more famous instances of a quantum algorithm, is known to be in $BQP$ but believed to be outside $P$.
Impact on Cryptography
If the majority of cryptographic schemes make use, in some way, of the hardness of problems then it makes sense to assume that the $BQP$ class will also impact the security proofs and reductions in cryptography. A simple argument for this is as follows. Shor's algorithm is in $BQP$ and believed to be outside $P$. We know Shor's algorithm jeopardizes cryptography based on factorization. Thus, there are potentially many more algorithms in $BQP$ but not in $P$ that will jeopardize cryptographic schemes. That said, we also know that quantum computing is not universally better than a classical computer, there are certain problems where performance is equal.
Thus, it could be argued that with a new understanding of $BQP$, we may find different security reductions are possible based on problems in $BQP$. If it's discovered that $BQP$ intersects $NP$, I think this would support reductions of such $NP$ problems for use in cryptography.
we could also use it to develop primitives involving quantum algorithms in their implementation
Yes, we can use knowledge of quantum complexity to inform decisions on constructing new cryptographic primitives.
Have this been studied? are there any proposals?
The closest example of this I am aware of is a proposal that uses the Riemann sphere for the $l_1$-ball and $l_\infty$-ball. This method was proposed given the potential correspondence between the Bloch sphere and Riemann sphere. You can find a brief overview here.
does having quantum computers improve the Crypto world somehow?
Yes. The complexity associated with quantum computing will have additional impacts beyond quantum algorithms known to threaten some classical cryptography. The potential impact outside of cryptanalysis may come in the form of novel reductions in security proofs for cryptography that relies on $BQP$ problems. Further impacts may also come in the form of cryptography using the Riemann sphere as a foundation for primitives, allowing similar models between cryptography and quantum computing.