In-situ computed tomography (CT) experiments on materials with time-dependent mechanical behaviour are affected by relaxationinduced motion, which can lead to image blur and motion-related artefacts if scans are initiated before relaxation-induced motion has subsided. Scan start times are therefore commonly defined based on force relaxation or force-gradient criteria, although these signals do not directly quantify image-relevant specimen motion. In this work, a radiography-based approach is presented to estimate relaxation-induced motion via pixel shifts from projection images acquired prior to CT scans. These projection-based pixel shift estimates of relaxation-induced motion are related to scan-specific image blur observed in the reconstructed volumes. Thereby, a direct link between specimen motion during the scan and CT image quality is established. The method is demonstrated for thermo-mechanically loaded specimens with pronounced temperature-dependent material behaviour, where relaxation-induced motion persists over extended time scales. The results show that projection-based pixel shift estimation provides a physically meaningful and experimentally accessible basis for defining scan start criteria. CT acquisition can be initiated based on an allowable level of relaxation-induced motion, rather than waiting for mechanical equilibrium to be reached. The proposed approach therefore offers a direct, image-related framework for scan timing in in-situ CT experiments on time-dependent materials.