Update app.py

app.py

@@ -8,22 +8,25 @@ import requests
 import gradio as gr
 import numpy as np
 from lxml.html import fromstring
-from diffusers import AutoPipelineForText2Image
-#from tempfile import NamedTemporaryFile
 from pathos.threading import ThreadPool as Pool
-
-#model_id = "runwayml/stable-diffusion-v1-5"
-model_id = "kandinsky-community/kandinsky-3"
-#model_id = "stabilityai/stable-diffusion-3-medium-diffusers"
+from diffusers import AnimateDiffPipeline, MotionAdapter, EulerDiscreteScheduler
+from diffusers.utils import export_to_gif
+from huggingface_hub import hf_hub_download
+from safetensors.torch import load_file
 
 device = "cuda" if torch.cuda.is_available() else "cpu"
-if torch.cuda.is_available():
-    torch.cuda.max_memory_allocated(device=device)
-    pipe = AutoPipelineForText2Image.from_pretrained(model_id, torch_dtype=torch.float16, variant="fp16", use_safetensors=True, token=os.getenv('hf_token'))
-    pipe = pipe.to(device)
-else: 
-    pipe = AutoPipelineForText2Image.from_pretrained(model_id, use_safetensors=True, token=os.getenv('hf_token'))
-    pipe = pipe.to(device)
+dtype = torch.float16
+
+step = 8
+repo = "ByteDance/AnimateDiff-Lightning"
+ckpt = f"animatediff_lightning_{step}step_diffusers.safetensors"
+#base = "emilianJR/epiCRealism"
+base = "black-forest-labs/FLUX.1-dev"
+
+adapter = MotionAdapter().to(device, dtype)
+adapter.load_state_dict(load_file(hf_hub_download(repo ,ckpt), device=device))
+pipe = AnimateDiffPipeline.from_pretrained(base, motion_adapter=adapter, torch_dtype=dtype).to(device)
+pipe.scheduler = EulerDiscreteScheduler.from_config(pipe.scheduler.config, timestep_spacing="trailing", beta_schedule="linear")
 
 def translate(text,lang):
     
@@ -90,21 +93,21 @@ def Piper(_do,_dont):
         height=256,
         width=768,
         negative_prompt=_dont,
-        num_inference_steps=50,
+        num_inference_steps=step,
         guidance_scale=7
     )
 
 def infer(prompt_en,prompt2_en):
     name = generate_random_string(12)+".png"
     if prompt_en == "":
-        _do = 'photograph'
+        _do = 'film'
     else:
-        _do = f'photographed { prompt_en }'
+        _do = f'filmed { prompt_en }'
     if prompt2_en == "":
         _dont = 'complex scene, ugly human body, partial human body, smooth texture, fictional content, blurred content, amputated human body, distorted palm fingers, missing legs, unreal eyes, squinting eyes, text anywhere, prints anywhere'
     else:
         _dont = f'{prompt2_en} anywhere, complex scene, ugly human body, partial human body, smooth texture, fictional content, blurred content, amputated human body, distorted palm fingers, missing legs, unreal eyes, squinting eyes, text anywhere, prints anywhere'
-    image = Piper(_do,_dont).images[0].save(name)
+    export_to_gif(Piper(_do,_dont).frames[0],name)
     return name
 
 css="""
@@ -146,21 +149,16 @@ function custom(){
 }
 """
 
-if torch.cuda.is_available():
-    power_device = "GPU"
-else:
-    power_device = "CPU"
-
 with gr.Blocks(theme=gr.themes.Soft(),css=css,js=js) as demo:
     result = []
     with gr.Column(elem_id="col-container"):
         gr.Markdown(f"""
-            # מחולל תמונות {power_device}
+            # GIF AI
         """)
         with gr.Row():
             prompt = gr.Textbox(
                 elem_id="prompt",
-                placeholder="מה *כן* להוסיף",
+                placeholder="WHAT TO CREATE...",
                 container=False,
                 rtl=True,
                 max_lines=1
@@ -168,13 +166,13 @@ with gr.Blocks(theme=gr.themes.Soft(),css=css,js=js) as demo:
         with gr.Row():
             prompt2 = gr.Textbox(
                 elem_id="prompt2",
-                placeholder="מה *לא* להוסיף",
+                placeholder="WHAT TO AVOID...",
                 container=False,
                 rtl=True,
                 max_lines=1
             )
         with gr.Row():
-            run_button = gr.Button("התחלה",elem_classes="btn",scale=0)
+            run_button = gr.Button("START",elem_classes="btn",scale=0)
         with gr.Row():
             result.append(gr.Image(interactive=False,elem_classes="image-container", label="Result", show_label=False, type='filepath', show_share_button=False))
         with gr.Row():