Upload README.md

README.md

@@ -1,7 +1,7 @@
 ---
 license: creativeml-openrail-m
 base_model: Bingsu/my-korean-stable-diffusion-v1-5
-training_prompt: A rabbit is eating a watermelon on the table
+training_prompt: A man is surfing
 tags:
 - tune-a-video
 - text-to-video
@@ -11,20 +11,23 @@ inference: false
 ---
 
 # Tune-A-VideKO - Korean Stable Diffusion v1-5
-Github: [Kyujinpy/Tune-A-VideKO](https://github.com/KyujinHan/Tune-A-VideKO/tree/master)
+Github: [Kyujinpy/Tune-A-VideKO](https://github.com/KyujinHan/Tune-A-VideKO)
 
 ## Model Description
 - Base model: [Bingsu/my-korean-stable-diffusion-v1-5](https://huggingface.co/Bingsu/my-korean-stable-diffusion-v1-5)
-- Training prompt: A rabbit is eating a watermelon on the table
-![sample-train](sample/rabbit.gif)
+- Training prompt: A man is surfing
+![sample-train](sample/surfing.gif)
 
 ## Samples
 
-![sample-500](sample/video4.gif)
-Test prompt: 고양이가 해변에서 수박을 먹고 있습니다
+![sample-500](sample/video10.gif)
+Test prompt: 미키마우스가 서핑을 타고 있습니다
 
-![sample-500](sample/video5.gif)
-Test prompt: 강아지가 오렌지를 먹고 있습니다
+![sample-500](sample/video11.gif)
+Test prompt: 한 여자가 서핑을 타고 있습니다
+
+![sample-500](sample/video12.gif)
+Test prompt: 흰색 옷을 입은 남자가 바다를 걷고 있습니다
 
 ## Usage
 Clone the github repo
@@ -46,8 +49,8 @@ unet = UNet3DConditionModel.from_pretrained(unet_model_path, subfolder='unet', t
 pipe = TuneAVideoPipeline.from_pretrained(pretrained_model_path, unet=unet, torch_dtype=torch.float16).to("cuda")
 pipe.enable_xformers_memory_efficient_attention()
 
-prompt = "강아지가 만화 스타일로 상자를 먹고 있습니다"
-video = pipe(prompt, video_length=8, height=512, width=512, num_inference_steps=50, guidance_scale=12.5).videos
+prompt = "흰색 옷을 입은 남자가 바다를 걷고 있습니다"
+video = pipe(prompt, video_length=24, height=512, width=512, num_inference_steps=50, guidance_scale=12.5).videos
 
 save_videos_grid(video, f"./{prompt}.gif")
 ```