Purpose <p>Granulation by roller compaction is an important unit operation in the production of pharmaceutical oral solid dosages. This process involves producing a densified “ribbon” of compressed powder between two counter rotating rolls that is then feed into a screening type mill and particulate matter of the order of 1–2&#xa0;mm is produced.</p> <p>In roller compaction, rate of production is generally maximised by increasing the roll speed, thus increasing the rate of ribbon production. However, such increases in production could result in issues with ribbon milling, potentially causing the mill to become overwhelmed. Knowledge of important milling parameters such as mill residence time and steady state hold up value as a function of roll and mill speed would allow for this parameter to be scaled. Traditional experiments towards this objective are problematic.</p> Methods <p>Presented here is a digital image processing approach that allows the continual monitoring of the mass of powder within a roller compactor’s mill utilising the equipment’s observation window. This allowed for an assessment of steady state mass hold up and the mean residence time in the mill. In this study, mean residence times were calculated to be between 45 and 120&#xa0;s.</p> Results <p>The approach allowed for the rapid accumulation of data sufficient to construct a, data driven model predicting both residence time and steady state hold up as a function of both roll speed and mill speed.</p> Conclusion <p>It is the authors assertion that this kind of image processing approach, has great utility in pharmaceutical manufacturing.</p>