Let $\mathcal{H}^k$ denote the $k$-dimensional Hausdorff measure and $2 \le m < n$ be two integers. Suppose $f : \mathbf{R}^m \to \mathbf{R}^{n-m}$ is continuous and define $F : \mathbf{R}^m \to \mathbf{R}^n$ by setting $F(x) = (x,f(x))$ for $x \in \mathbf{R}^m$. Assume that the graph of $f$ has locally finite $\mathcal{H}^m$ measure, i.e.,
$$ \mathcal{H}^m(K \cap \operatorname{image}F) < \infty \quad \text{for any compact set $K \subseteq \mathbf{R}^n$}\,. $$
$\bullet$ Does it follow that $f$ is of bounded variation, i.e., $f \in BV$?
Assume additionally that $f$ has the Lusin N property, i.e.,
$$ \mathcal{H}^m(Z) = 0 \quad \Rightarrow \quad \mathcal{H}^m(F[Z]) = 0 \,. $$
$\bullet$ Does it follow that $f$ is weakly differentiable?
The answer to both questions is YES. For the proof one can employ the characterisation of BV by integration of essential variation on lines (see see 4.5.10 in Federer 1969). I wrote down the details.