Basic quality embeddings made for the DogePoint with no particular style bias so it might work from sketch to photorealism without much trouble. Can't guarantee it'll work in different models with different block weights or clipskip training, so I recommend testing your prompt with and without it to be sure it'll go smooth. Most of the showcased models are made with the 30v positive and 35v negative, so this is my personal recommendation for now, since these current embedding are getting outdated and I'll soon decide a new definitive combo and update it.

DogeNAI - Embeddings

Files with SHA256: 624a08052fe4ab5aa8cd3278b2de282a78ba946b47b6a653a2a661b3906fe504