Free play, make rational use

For the new recommended resolution for model use, the probability of generating image failure you get is reduced to recommend70%.

ResolutionRatioResolutionRatioResolutionRatio1024 x 10241:1 Squarerecommended1600×6405:21408×7042:1704×14081:21472×70423:11704×147211:23960×102415:161024×96016:15960×108815:171088×96017:15896×10887:91088×8969:7896×11527:91152×8969:7832×115213:18recommended1152×83218:13recommended832×121613:191216×83219:13768×12803:51280×7685:3768×13444:7recommended1344×7687:4recommended1536×64012:5recommended640×15365:12recommended

Following such resolution settings will give you the best experience, based on research testing by mj and stable AI

CategoryDescription WordsColorMonochromeColorfulStyleIllustrationCyberpunkInk wash painting styleTraditional Chinese painting styleFantasy Art, Anime

If your description words do not emphasize the background, then the background will be generated through style or solid color background,

directly generate large images, detailed pictures,normally no need for high-definition restoration and refiner,unless full body is needed, these types of prompts may not cause face crashes.

The normal basic XL model comes with many styles that you can try to summon, but the secondary element of two-dimensional subject will not be affected by the concepts you summon

Model Author: Rice not tastyModel Type: Diffusion-based text-to-image generative modelLicense: CC-BY-NC-SA-4.0Fine-tuned to Model: Animagine XL 3.0