rr, cc = polygon(y, x, (WIDTH, HEIGHT)) for i in range(3): data[0, rr, cc, i] += color # here's where you draw it. smooth. neat. xold = event.x yold = event.y data = np.array((np.random.random((1, HEIGHT, WIDTH, 3)) - 0.5) * 4, dtype=float) in_x = tf.placeholder(tf.float32, shape=(1, WIDTH, HEIGHT, 3)) with tf.variable_scope("gen", reuse=None) as scope: with tf.name_scope("1"): out_x = superres_model.superres_model(in_x) with tf.variable_scope("gen", reuse=True) as scope: for i in range(6): with tf.name_scope(str(i + 2)): out_x = superres_model.superres_model(out_x) restore_list = [x for x in tf.trainable_variables() if "ssskkds" not in x.name] saver = tf.train.Saver(var_list=restore_list) sess = tf.Session() saver.restore(sess, sys.argv[1]) window = Tk()
crops = [] for i in range(BATCH_SIZE): crops.append(tf.image.random_crop(jpg, [512, 512])) jpg = tf.pack(crops) #jpg = tf.image.resize_bilinear(jpg, [128, 128]) mix = tf.random_uniform([1]); net_in = mix*jpg + (1.0-mix)*tf.truncated_normal(jpg.get_shape(), dtype=tf.float32, stddev=1) l10 = net_in all_images = [] with tf.variable_scope("gen", reuse=None) as scope: with tf.name_scope("1"): l10 = superres_model.superres_model(l10) all_images.append(l10) with tf.variable_scope("gen", reuse=True) as scope: for i in range(2,8): with tf.name_scope(str(i)): l10 = superres_model.superres_model(l10) all_images.append(l10) measure_images = tf.concat(0, [image[:,:,:,0:3] for image in [all_images[i] for i in [0, 2, 4, 6]]]) feedback_images = tf.concat(0, [image[:,:,:,0:3] for image in [all_images[i] for i in [6]]]) show_images = tf.concat(0, [image[0:1,:,:,0:3] for image in all_images]) tf.image_summary("gen", measure_images, max_images=1) tf.image_summary("real", jpg, max_images=1)