If it takes too long to render in GitHub itself, take a look at it at Kaggle.
-
Notebook Link: https://www.kaggle.com/code/aneesh10/dcgan-on-pok-mon-images
-
Dataset Link: https://www.kaggle.com/datasets/kvpratama/pokemon-images-dataset
Also consider upvoting!
The learning is quite unstable, if we look at the intermediate output on the fixed noise. Also, the model is kind off overfitting with some attributes of a few pokémons. This might be corrected, with some additional regularization.