trained on hand-captioned images, better prompt adherence and flexibility










version 1.1 update - trained on largely the same dataset, but captioned. Output follows your prompt more closely and flexibly. It's minor, but definitely noticeable in a number of images.
Although well represented in the base model and by others's LoRAs, the eldritch vision of watercolor painting brings a stronger emphasis on the tactile sense of a heavyweight, rough paper substrate, and on versatility - this style should apply equally well to a broad range of subject matter.
You don't need particular keywords, but you can refer to 'watercolor painting' - the effect is often of a softer color and stronger paper texture. The model tends to respect 'minimalist' and 'white space' as concepts.
trained on hand-captioned images, better prompt adherence and flexibility
1. 轉载模型僅供學習與交流分享,其版權及最終解释權归原作者。
2. 模型原作者如需認領模型,請通過官方渠道联系海藝AI工作人員進行認證。我們致力於保護每一位創作者的權益。 點擊去認領
