def invert_color_func(params_dict): params01_vec_dst = [] for key in FurParam.ParamsColor: value = param_color_dict[key] val01 = FurParam.ConvertFurParam(key, value) params01_vec_dst.append(val01) return params01_vec_dst
## invert initial parameter dictionary as vector ############################################################ ## 1) random initial parameters [NOT SO USEFUL] #params01_vec_dst = np.random.random(15) ## 2) use initial parameters from Bayesian optimization #''' csv_ref_path = os.path.splitext(img_ref_path)[0]+"_bayesopt.csv" init_params_dict = FurParam.csv2dict(csv_ref_path) params01_vec_dst = [] #''' for key in FurParam.ParamsGeom: value = init_params_dict[key] val01 = FurParam.ConvertFurParam(key, value) params01_vec_dst.append(val01) ############################################################ ## enter optimization routine ############################################################ ## initialize elapsed time for each component as 0:00:00.000 t_total_start = datetime.now() ## run optimization on GEOMETRY parameters succeeded, shape_param_dict = GradientDescent( vgg_max_gray_gram, calc_cost_func, furRenderer.RenderFur, convert_param_func, params01_vec_dst, folder_root, imgFileExt )