add infos of each input field

README2.md

@@ -8,7 +8,7 @@ Please refer to our NeurIPS 2024 submission for more details about AdaFace:
 **AdaFace: A Versatile Face Encoder for Zero-Shot Diffusion Model Personalization** 
 </br>
 
-[![Hugging Face Spaces](https://img.shields.io/badge/%F0%9F%A4%97 Hugging Face-Spaces-yellow)](https://huggingface.co/spaces/adaface-neurips/adaface-animate)
+[![Hugging Face Spaces](https://img.shields.io/badge/%F0%9F%A4%97%20Hugging%20Face-Spaces-yellow)](https://huggingface.co/spaces/adaface-neurips/adaface-animate)
 
 This pipeline uses 4 pretrained models: [Stable Diffusion V1.5](https://huggingface.co/runwayml/stable-diffusion-v1-5), [AnimateDiff v3](https://github.com/guoyww/animatediff), [ID-Animator](https://github.com/ID-Animator/ID-Animator) and [AdaFace](https://huggingface.co/adaface-neurips/adaface).
 

app.py

@@ -229,13 +229,14 @@ with gr.Blocks(css=css) as demo:
     ❗️❗️❗️**Tips:**
     - You can upload one or more subject images for generating ID-specific video.
     - Try different parameter combinations for the best generation quality.
+    - Technical explanations and demo videos: [Readme](https://huggingface.co/spaces/adaface-neurips/adaface-animate/blob/main/README2.md).
         """
     )
 
     with gr.Row():
         with gr.Column():
             files = gr.File(
-                        label="Drag (Select) 1 or more photos of a person's face",
+                        label="Drag / Select 1 or more photos of a person's face",
                         file_types=["image"],
                         file_count="multiple"
                     )
@@ -245,7 +246,7 @@ with gr.Blocks(css=css) as demo:
                 remove_and_reupload = gr.ClearButton(value="Remove and upload subject images", components=files, size="sm")
 
             init_img_files = gr.File(
-                            label="Drag (Select) 1 image for initialization",
+                            label="[Optional] Select 1 image for initialization, or generate 3 images with the button below and select 1",
                             file_types=["image"],
                             file_count="multiple"
                     )
@@ -258,6 +259,7 @@ with gr.Blocks(css=css) as demo:
 
             init_image_strength = gr.Slider(
                     label="Init Image Strength",
+                    info="How much the init image should influence each frame. 0: no influence (scenes are more dynamic), 3: strongest influence (scenes are more static).",
                     minimum=0,
                     maximum=3,
                     step=0.25,
@@ -265,6 +267,7 @@ with gr.Blocks(css=css) as demo:
                 )
             init_image_final_weight = gr.Slider(
                     label="Final Weight of the Init Image",
+                    info="How much the init image should influence the end of the video",
                     minimum=0,
                     maximum=0.25,
                     step=0.025,
@@ -277,11 +280,12 @@ with gr.Blocks(css=css) as demo:
                 gen_init = gr.Button(value="Generate 3 new init images")
 
             prompt = gr.Textbox(label="Prompt",
-                    #    info="Try something like 'a photo of a man/woman img', 'img' is the trigger word.",
-                       placeholder="Iron Man soars through the clouds, his repulsors blazing.")
+                       info="Try something like 'man/woman walking on the beach'",
+                       placeholder="woman playing guitar on a boat, ocean waves")
            
             image_embed_scale = gr.Slider(
                     label="Image Embedding Scale",
+                    info="The scale of the ID-Animator image embedding (influencing coarse facial features and poses)",
                     minimum=0,
                     maximum=2,
                     step=0.1,
@@ -289,6 +293,7 @@ with gr.Blocks(css=css) as demo:
                 )
             attn_scale = gr.Slider(
                     label="Attention Processor Scale",
+                    info="The scale of the ID embeddings on the attention (the higher, the more focus on the face, less on the background)" ,
                     minimum=0,
                     maximum=2,
                     step=0.1,
@@ -296,6 +301,7 @@ with gr.Blocks(css=css) as demo:
                 )
             adaface_id_cfg_scale = gr.Slider(
                     label="AdaFace Embedding ID CFG Scale",
+                    info="The scale of the AdaFace ID embeddings (influencing fine facial features and details)",
                     minimum=0.5,
                     maximum=6,
                     step=0.25,
@@ -307,12 +313,15 @@ with gr.Blocks(css=css) as demo:
             with gr.Accordion(open=False, label="Advanced Options"):
                 video_length = gr.Slider(
                     label="video_length",
+                    info="Do not change, otherwise the video will be messy",
                     minimum=16,
                     maximum=21,
                     step=1,
                     value=16,
                 )
-                is_adaface_enabled = gr.Checkbox(label="Enable AdaFace", value=True)
+                is_adaface_enabled = gr.Checkbox(label="Enable AdaFace", 
+                                                 info="Enable AdaFace for better face details. If unchecked, it falls back to ID-Animator (https://huggingface.co/spaces/ID-Animator/ID-Animator).",
+                                                 value=True)
                 adaface_ckpt_path = gr.Textbox(
                     label="AdaFace ckpt Path", 
                     placeholder=args.adaface_ckpt_path,
@@ -321,6 +330,7 @@ with gr.Blocks(css=css) as demo:
 
                 adaface_power_scale = gr.Slider(
                         label="AdaFace Embedding Power Scale",
+                        info="Increase this scale slightly only if the face is defocused or the face details are not clear",
                         minimum=0.7,
                         maximum=1.3,
                         step=0.1,

requirements.txt

@@ -14,3 +14,5 @@ diffusers
 onnx>=1.16.0
 onnxruntime
 safetensors
+spaces
+protobuf==3.20