def Generator(noise_shape, condition_shape, output_shape, final_size, latent_shape=64, base_filters=512, min_filters=8):
    # Initial Layers
    condition_input = Input(condition_shape)
    condition_latent = Dense(units=latent_shape)(condition_input)
    condition_latent = Activation('relu')(condition_latent)
    condition_latent = Reshape()(condition_latent)
    condition_model = Model(inputs=condition_input, outputs=condition_latent)

    noise_input = Input(noise_shape)
    noise_latent = Dense(units=latent_shape)(noise_input)
    noise_latent = Activation('relu')(noise_latent)
    noise_latent = Reshape()(noise_latent)
    noise_model = Model(inputs=noise_input, outputs=noise_latent)

    filters = base_filters
    gen = condition_model.output
    noise = noise_model.output

    for i in range(int(np.log2(final_size / 4))):
        gen = concatenate([gen, noise])

        gen = Conv2D(filters, kernel_size=3, strides=1)(gen)
        gen = Activation('relu')(gen)
        gen = Conv2D(filters, kernel_size=3, strides=1)(gen)
        gen = Activation('relu')(gen)
        
        gen = UpSampling2D()(gen)
        noise = UpSampling2D()(noise)

        filters = int(filters / 2) if filters > min_filters else filters

    gen = Conv2D(output_shape, kernel_size=1, strides=1)(gen)
    gen = Activation('linear')(gen)

    G = Model(inputs=[noise_model.input, condition_model.input], outputs=gen, name='Generator')

    return G