|
This is Anya Taylor-Joy 3 vector embedding for version 1.5 of stable diffusion. This model was made with a dataset of 28 images, a learning rate of 0.5:50,0.05:100,0.005:400, a batch size of 7 and a Gradient accumulation steps of 4. I modified the filewords created by Blip in order to have a more precise description to increase the flexibility of the model. I used a textural inversion template with only [name], [filewords] inside. The trigger name of the template is ANAT0J3 , but it can be changed to whatever you want by changing the file name. The example images were made using these two models: https://civitai.com/models/4201/realistic-vision-v12 https://civitai.com/models/3666/protogen-x34-photorealism-official-release This is my first embedding, I hope you like it. I would be glad to see your creations if you want to share them. I would also like to thanks seedspiller for the information he provided in the description of his models which allowed me to create the one I am presenting here. | 
发表于 2023-6-7 22:44:01