A new paper on arXiv asks a deceptively simple question: can a volume that reproduces every external RGB view still have an incorrect density? In emission-absorption rendering, the authors show the answer is yes. Using an independently parameterized, physically bounded effective source, they construct volumes that are appearance-equivalent to a correct one but differ in density.
The result does not contradict the forward rendering model; it exposes an ambiguity in the inverse problem. Because the effective source can compensate for density changes, external color images do not uniquely determine the underlying volume. The authors emphasize that the source term must be physically bounded, which rules out some trivial or unphysical constructions.
For researchers using neural or classical volume reconstruction, the practical implication is that matching RGB views is necessary but not sufficient for recovering true density. The paper does not claim all reconstructions are wrong, but it shows that appearance-based validation alone can be misleading. The study is a theoretical contribution, and the authors do not provide a method for resolving the ambiguity in practice.