def Discriminator(image_shape, condition_shape, initial_size, latent_shape, base_filters, min_filters):
    # Initial Layers
    condition_input = Input(condition_shape)
    condition_latent = Dense(units=latent_shape)(condition_input)
    condition_latent = Activation('relu')(condition_latent)
    condition_model = Model(inputs=condition_input, outputs=condition_latent)

    filters = base_filters
    image_input = Input(image_shape)
    image = Conv2D(filters, kernel_size=4, strides=1)(image_input)

    for i in range(int(np.log2(initial_size / 4))):
        filters = filters // 2
        image = Conv2D(filters, kernel_size=3, strides=2)(image)
        image = Activation('relu')(image)

    image = Reshape()(image)
    image_latent = Dense(units=latent_shape)(image)
    image_model = Model(inputs=image_input, outputs=image_latent)

    dis = concatenate([condition_model.output, image_model.output])
    dis = Dense(units=filters)(dis)
    dis = Dense(units=filters)(dis)
    dis = Dense(units=1)(dis)
    dis = Activation('sigmoid')(dis)

    D = Model(inputs=[image_model.input, condition_model.input], outputs=dis, name='Discriminator')
    
    return D