Existing methods of level generation using latent variable models such as\nVAEs and GANs do so in segments and produce the final level by stitching these\nseparately generated segments together. In this paper, we build on these\nmethods by training VAEs to learn a sequential model of segment generation such\nthat generated segments logically follow from prior segments. By further\ncombining the VAE with a classifier that determines whether to place the\ngenerated segment to the top, bottom, left or right of the previous segment, we\nobtain a pipeline that enables the generation of arbitrarily long levels that\nprogress in any of these four directions and are composed of segments that\nlogically follow one another. In addition to generating more coherent levels of\nnon-fixed length, this method also enables implicit blending of levels from\nseparate games that do not have similar orientation. We demonstrate our\napproach using levels from Super Mario Bros., Kid Icarus and Mega Man, showing\nthat our method produces levels that are more coherent than previous latent\nvariable-based approaches and are capable of blending levels across games.\n