Puff pastry is characterized by a light and flaky texture due to its unique combination of fat and dough in a layered structure. It is commonly believed that the continuity of the layers is a feature essential to achieve the desirable texture after baking because fat layers would be impervious to vapor. The aim of the present study was to apply imaging techniques and image analysis to better understand the development of these structures (layers and bubbles) in croissant. A multi-scale study was attempted by combining techniques with distinct spatial resolutions: Confocal Laser Scanning Microscopy CLSM (pixel size 0.6µm), and Magnetic Resonance Imaging MRI (pixel size 500µm). CLSM was applied to pastry after sheeting. Thickness along each layer, number of fat layers, and number of ruptures in fat layers were calculated after thresholding and labeling of fat layers. MRI was applied to pastry during proving. Maps of gas, fat and dough content were calculated using a dedicated algorithm. The measured number of fat layers after sheeting was proportional to the expected one, although lower by 10%. Missing fat layers in CLSM images were attributed to large-sized ruptures (of few mm up to few cm), and were also observed in MRI images. Moreover fat fractures in CLSM images ranging from 10 to 100 µm increased in proportion at high number of sheets and could explain the loss of the impervious function of fat layers and the decrease in specific volume. Elastic recovery of pastry after sheeting was measured in CLSM images from the thickness of dough layers. Each dough layer contributed equally to global expansion during proving when its initial thickness remained above a threshold value, but expanded more slowly below this threshold. The largest bubbles in dough were also visualized in MRI images at the end of proving; they were not preferentially located near the fat layers.