Update README.md

README.md

@@ -200,9 +200,9 @@ images[0].save(f"your image save path, png format is usually better than jpg or
 ## Evaluation Data
 The test data is randomly sample from popular wallpaper anime images(pixiv, nijijourney and so on), the purpose of the project is to letting everyone can draw an anime Illustration.
 We select 100 images and generate text with waifu-tagger[https://huggingface.co/spaces/SmilingWolf/wd-tagger] and generate 4 images per prompt, totally 400 images generated, the images
-should be 1024 * 1024 or same bucket resolution to acheive the best performance. We caculate the Laion Aesthetic Score to measure the beauty and the PerceptualSimilarity to measure the 
-control ability, we find the quality of images have a good consistency with the meric values. We compare our methods with other SOTA huggingface models and list the result below. We are 
-the models that have highest aesthectic score, and can generate visually appealing images if you prompt it properly.
+The images resolution should be 1024 * 1024 or same bucket for SDXL and 512 * 768 or same bucket for SD1.5, we then resize sdxl-generated images to 512 * 768 or same bucket for fair comparison. 
+We caculate the Laion Aesthetic Score to measure the beauty and the PerceptualSimilarity to measure the control ability, we find the quality of images have a good consistency with the meric values. 
+We compare our methods with other SOTA huggingface models and list the result below. We are the models that have highest aesthectic score, and can generate visually appealing images if you prompt it properly.
 
 ## Quantitative Result
 | metric | xinsir/anime-painter | lllyasviel/control_v11p_sd15_scribble |