Table of contents
If you need better color performance, I recommend you try Color Box
Here are some suggested settings for the parameters:
Run map demo (multiple map warning)
Night Sky YOZORA Model - "Achieving Ultimate Image Quality and Large Image Sizes (>1536 x 1024)" Trained by @YozoRaAru
If you need better color performance, I recommend you try Color Box
YOZORA is a model that pursues perfection and is full of personal preferences . I trained it using images I like, which gives it an unrivaled level of detail and finish. It can provide you with a very refined image of a character.
In addition to training, I also blended it with dozens of other high-quality models that I personally like, and I also experimented with layer blending. Among these models, I chose NovelAI's original model as the main weight because it can give it excellent adaptability to various cues. I haven't tried to test it with scenery+characters yet, but I'm sure it can perform well there too.
-
For best results , make sure your final resolution is > 1536×1024 . This is necessary to generate nice illustrations
Introduction to the stream-saving version: the word "dazzling", the more dazzling the tag, the better
Here are some suggested settings for the parameters:
-
Clip skip : It is recommended to set it to no less than 2. YOZORA already has a rich level of detail, setting it to 1 may make the image appear choppy.
-
Resolution : YOZORA may not be suitable for producing small images , as extreme levels of detail can become overcrowded. It is not recommended to use Hires.fix to generate large images . Larger resolutions equate to larger canvases, making it easier for YOZORA to capture details. 1536 x 1024 or higher is recommended for Hires.fix .
-
Negative tip : EasyNegative is recommended to provide a short and precise description.
The name YOZORA means "walking among the stars in the night sky". Hope she brings you the excitement and joy of starry sky.
Recommended download address
Night Sky YOZORA Style Model - AiTool.aihttps://aitool.ai/model/12262
Run map demo (multiple map warning)