Caption Evaluation Method (CEM) was originally developed to evaluate the streetscape of a neighborhood high street (Koga T. et al.). The aim was to grasp the diversity of ordinary people’s image about townscape. The procedures of CEM are: 1) Participants walk around the objective area. 2) Each participant takes a photograph when he/she finds a landscape scene/element felt good or bad. 3) Afterwards, each participant puts captions on each of his/her own photographs. Participants are instructed that his/her captions on the scene/element should include “judgment (good/bad)”, “what kind of character”, “of which element (or part) of the scene”, made “what kind of impression”. Elements and captions obtained are usually used in the workshop of participants to make proposals for improving landscapes.In residential areas where townscape is well known to the residents, CEM has some advantages over previous popular methods such as Semantic Differential technique using color slides, which have limitation of the objective scenes and words used in the evaluation.As the development of CEM, two objectives are set in this study. 1) How do children react to landscape evaluation using CEM? 2) In shopping areas or shopping complex buildings where shoppers are not as familiar with landscapes as residential areas, how can we grasp the diversity of ordinary shoppers’ image? For the first objective, some developments are designed for children and the streetscape around an elementary school in the city center area is evaluated by the school children and neighborhood adults. Through the workshop, children were very interested in CEM and they made some good proposals, which are slightly different from the adults. For the second objective, Picture observing CEM is designed and a large shopping complex building is evaluated with both original (Environment experiencing) CEM and newly proposed CEM. More elements/scenes are noticed with Picture observing CEM than the original method. Picture observing CEM could be more useful to evaluate shopping complex buildings than the original CEM, though the difference between them should be studied further.