Quantitative visualization of chemical processes and the consequent incorporation of images into process control loops was an idea that E.T. Woodburn et al. pioneered in mineral processing. This approach is based on the notion that many chemical processes possess intrinsic visual indicators which are otherwise difficult to instrument. Stirred vessels used as chemical reactors have the same potential to benefit from quantitative imaging, although, in this case, not just for control purposes but also as a powerful adjunct to model development and validation. We outline developments pursued at UMIST which seek to utilize image reconstruction modelling of mixing to interpret simple tracer diagnostic tests and thereby to predict chemical reactor behaviour under semi-batch operation. Progress from small two-dimensional axisymmetric networks-of-zones, through larger three-dimensional networks-of-zones to computational fluid dynamics predictions in full three dimensions are outlined. Problems requiring more advanced techniques and models to link macro-scale mixing to micro-mixing form a wide set of challenges for future research