In recent days, deep learning technologies have gained more and more interest in computer related task. In generative models, autoencoder (AE) has achieved a tremendous success in many fields, especially in image generation due to ability to produce highly realistic looking images. In synthetic image generation, this learning framework is very effective. In this paper, we take a broad view of autoencoder technique and compare and implement some autoencoders model with different architectures. The first autoencoder is simple autoencoder with only bottleneck layer, second autoencoder is with one hidden layer and bottleneck layer and third autoencoder is the convolutional autoencoder. We use convolutional layers in convolutional autoencoder. We have generated the images using these autoencoders and compare the training loss, validation loss, and accuracy.