An important step in the color image processing pipeline for modern digital cameras is the transform from camera response resulting from scene spectral radiance to objective colorimetric or related quantities. Knowledge of the camera spectral sensitivities is essential for building robust camera transforms for arbitrary capture and viewing conditions. Monochromator-based techniques are well-known but are cumbersome, slow, and impractical for routine use such as production-line camera calibration. It has been shown that with suitable characterization data it is possible to accurately estimate spectral sensitivities without the use of a monochromator. However, these methods are inherently highly metameric and the performance on likely scene capture spectra is unclear. In the current work, it is shown that combining highlydimensional characterization data with multidimensional analysis of typical camera responses produces spectral sensitivity estimates for a test camera that performs extremely well on likely scene spectra while minimizing the opportunity for camera observer metamerism.