No description provided.
alphaply changed pull request status to open

I have updated the image dataset, enlarging all images by a factor of two to 1024x1280 using the scaling algorithm provided by sdwebui. However, it's apparent that some areas of the images have also been scaled up, causing them to appear blurred. Please note that this is only a test. Nevertheless, it offers you a direct sdxl version of the LoRa training dataset. Feel free to remove any blurred images. In a few days, I will reorganize my dataset and submit a new pull request.

Carzit changed pull request status to merged

Sign up or log in to comment