Techniques are disclosed for providing guidance information for guiding a user to capture an image according to a predefined composition. A first user may provide information identifying a reference frame that defines a composition desired by the first user. The reference frame depicts a particular view of the real-world environment that the first user desires as the composition of the photo-to-be-taken by a second user. Based upon the reference frame and using augmented reality techniques, AR content is presented to a second user that guides the second user in positioning and orienting the image capture device such that the live view captured by the image capture device matches the particular view. When the live view aligns with the particular view, the second user may take a photo. The composition of such a photo matches the composition desired by the first user and represented by the reference frame.