[BBB]

Perhaps it's asking too much to get the transformation in one image, given transformation art is niche enough it may not appear often in the training data. One could try separate generations for "She-Hulk beginning to transform," "She-Hulk in mid transformation," and "She-Hulk at full power"; or "Power Girl with a modest bust," "Power Girl with prominent breasts," and "Power Girl with huge bazongas!" etc. You may be able to use the same seed for all prompts to get a similar art style. Then the results could be edited together (by a human) into a sequence.