Throughout our daily experiences, we learn regularities about contextual associations between objects and scenes which are used to form predictions about the likely features of the environment and facilitate perception of noisy visual input. Our recent studies showed that blurred objects that can be predicted based on contextual information appear subjectively sharper than the same objects that cannot. However, current predictive processing theories suggest that these effects may depend on the robustness of predictions. Additionally, there is neuroimaging evidence suggesting that object-based predictions can reciprocally influence the processing of scene context. Our study aimed to address these two hypotheses at the perceptual level using a blur matching task in two Experiments. In Experiment 1, participants (n = 65) saw two images depicting two versions of the same scene containing a blurred object and had to adjust the blur level of the right object to match the blur level of the left one. We manipulated the robustness of context-based predictions about the object in scene pairs by varying the phase coherence of their contextual information (allowing to form weak to very robust predictions). Blur-matching errors suggested that robustly-predicted objects were subjectively perceived as sharper than objectively similar objects benefiting from weaker predictions and this effect increased with the relative robustness of predictions. In Experiment 2, participants (n = 30) had to adjust the blur level of the right context to match the left one. This time, we manipulated object information allowing to form prediction about the context: one scene contained an intact object (predictable context), while the other had a phase-scrambled object (unpredictable context). Results showed that at objectively equal blur levels participants perceived predictable contexts as sharper than unpredictable ones, indicating that object-based predictions sharpen the perception of noisy contextual information. These findings further precise how predictions shape subjective perception.