The Lora is trained on MidJourney Images that use my "Personalised Style" + Pixel ArtWorks without the trigger words ArsMJStylePony, Pixel Art, manyP, fewP You can find v2 of this model here: <a target="_blank" rel="ugc" href="https://civarchive.com/models/629354?modelVersionId=703625">https://civarchive.com/models/629354?modelVersionId=703625</a> v2 have different color structure and the tagging described below works significantly better. IMPORTANT: All 8bit samples in the dataset are tagged with fewP . You can add it to the negative/positive depending on whether you prefer that type of Pixel Art . All "higher quality" 16bit+ samples are tagged with manyP Try lowering CFG if everything is too "melty"Will start adding visible effects at 0.35+ Works great in the 0.4 - 0.8 range, depending on your preference and Lora MixI recommend 1+ for best "Pixel Art" results, if you are not going to mix it with other Loras.Few of the images in the gallery are with the final epoch that was overbaked, but still produced interesting results. If anyone is interested I will upload it as well as a separate version.

Files with SHA256: d834886dcf23fd5f920ee866beb956f64ca429be0aee93eaa3b4dfa8023b5f72