Update app.py

app.py

@@ -1,266 +1,216 @@
 import spaces
-
-"""
-Copyright NewGenAI
-Code can't be included in commercial app used for monetary gain. No derivative code allowed.
-"""
+from datetime import datetime
 import gc
-import json
-import torch
-import tqdm
 import gradio as gr
+import numpy as np
 import random
-import time
-from datetime import datetime
+from pathlib import Path
 import os
 
+from diffusers import AutoencoderKLLTXVideo, LTXPipeline, LTXVideoTransformer3DModel
 from diffusers.utils import export_to_video
-from diffusers import LTXPipeline
 from transformers import T5EncoderModel, T5Tokenizer
-from pathlib import Path
-from datetime import datetime
-from huggingface_hub import hf_hub_download
-
-STATE_FILE = "LTX091_state.json"
-queue = []
-
-
-def load_state():
-    if os.path.exists(STATE_FILE):
-        with open(STATE_FILE, "r") as file:
-            return json.load(file)
-    return {}
-
-# Function to save the current state
-def save_state(state):
-    with open(STATE_FILE, "w") as file:
-        json.dump(state, file)
-
-# Load initial state
-initial_state = load_state()
-
-def add_to_queue(prompt, negative_prompt, height, width, num_frames, num_inference_steps, fps, seed):
-    task = {
-        "prompt": prompt,
-        "negative_prompt": negative_prompt,
-        "height": height,
-        "width": width,
-        "num_frames": num_frames,
-        "num_inference_steps": num_inference_steps,
-        "fps": fps,
-        "seed": seed,
-    }
-    queue.append(task)
-    return f"Task added to queue. Current queue length: {len(queue)}"
-
-def clear_queue():
-    queue.clear()
-    return "Queue cleared."
-
-def process_queue():
-    if not queue:
-        return "Queue is empty."
-
-    for i, task in tqdm(enumerate(queue)):
-        generate_video(**task)
-        time.sleep(1)  # Simulate processing time
-
-    queue.clear()
-    tqdm.close()
-    return "All tasks in the queue have been processed."
-
-def save_ui_state(prompt, negative_prompt, height, width, num_frames, num_inference_steps, fps, seed):
-    state = {
-        "prompt": prompt,
-        "negative_prompt": negative_prompt,
-        "height": height,
-        "width": width,
-        "num_frames": num_frames,
-        "num_inference_steps": num_inference_steps,
-        "fps": fps,
-        "seed": seed,
-    }
-    save_state(state)
-    return "State saved!"
-
-repo_id = "a-r-r-o-w/LTX-Video-0.9.1-diffusers"
-base_path = repo_id
-files_to_download = [
-        "model_index.json",
-        "scheduler/scheduler_config.json",
-        "text_encoder/config.json",
-        "text_encoder/model-00001-of-00004.safetensors",
-        "text_encoder/model-00002-of-00004.safetensors",
-        "text_encoder/model-00003-of-00004.safetensors",
-        "text_encoder/model-00004-of-00004.safetensors",
-        "text_encoder/model.safetensors.index.json",
-        "tokenizer/added_tokens.json",
-        "tokenizer/special_tokens_map.json",
-        "tokenizer/spiece.model",
-        "tokenizer/tokenizer_config.json",
-        "transformer/config.json",
-        "transformer/diffusion_pytorch_model.safetensors",
-        "vae/config.json",
-        "vae/diffusion_pytorch_model.safetensors",
-    ]
-os.makedirs(base_path, exist_ok=True)
-for file_path in files_to_download:
-    try:
-        # Create the full directory path for this file
-        full_dir = os.path.join(base_path, os.path.dirname(file_path))
-        os.makedirs(full_dir, exist_ok=True)
-        
-        # Download the file
-        downloaded_path = hf_hub_download(
-            repo_id=repo_id,
-            filename=file_path,
-            local_dir=base_path,
-        )
-        
-        print(f"Successfully downloaded: {file_path}")
-        
-    except Exception as e:
-        print(f"Error downloading {file_path}: {str(e)}")
-        raise
-
-# Download model from different repo
-try:
-    # Create the full directory path for this file
-    full_dir = os.path.join(base_path, os.path.dirname(file_path))
-    os.makedirs(full_dir, exist_ok=True)
-    
-    # Download the file
-    downloaded_path = hf_hub_download(
-        repo_id="Lightricks/LTX-Video",
-        filename="ltx-video-2b-v0.9.1.safetensors",
-        local_dir=repo_id,
-    )
-except Exception as e:
-    print(f"Error downloading 0.9.1 model: {str(e)}")
-    raise
+import torch
 
+from utils import install_packages
+
+torch.backends.cuda.matmul.allow_tf32 = True
+torch.backends.cudnn.allow_tf32 = True
+torch.jit._state.disable()
+torch.set_grad_enabled(False)
+
+gc.collect()
+torch.cuda.empty_cache()
+
+ckpt_path = Path("a-r-r-o-w/LTX-Video-0.9.1-diffusers")
+single_file_url = "https://huggingface.co/Lightricks/LTX-Video/ltx-video-2b-v0.9.1.safetensors"
+transformer = LTXVideoTransformer3DModel.from_single_file(
+    single_file_url, torch_dtype=torch.bfloat16
+)
+vae = AutoencoderKLLTXVideo.from_single_file(
+    single_file_url, torch_dtype=torch.bfloat16)
+vae.eval()
+vae = vae.to("cuda")
 
-single_file_url = repo_id+"/ltx-video-2b-v0.9.1.safetensors"
 text_encoder = T5EncoderModel.from_pretrained(
-  repo_id, subfolder="text_encoder", torch_dtype=torch.bfloat16
+    ckpt_path,
+    subfolder="text_encoder",
+    torch_dtype=torch.bfloat16
 )
+text_encoder.eval()
+text_encoder = text_encoder.to("cuda")
+
 tokenizer = T5Tokenizer.from_pretrained(
-  repo_id, subfolder="tokenizer", torch_dtype=torch.bfloat16
+    ckpt_path,
+    subfolder="tokenizer"
 )
-pipe = LTXPipeline.from_single_file(
-    single_file_url, 
-    text_encoder=text_encoder, 
-    tokenizer=tokenizer, 
+
+pipeline = LTXPipeline.from_single_file(
+    single_file_url,
+    transformer=transformer,
+    text_encoder=text_encoder,
+    tokenizer=tokenizer,
+    vae=vae,
     torch_dtype=torch.bfloat16
 )
-pipe.vae.enable_tiling()
-pipe.to("cuda")
-
-# pipe.load_lora_weights("TODO/TODO", adapter_name="ltx-lora")
-# pipe.set_adapters(["lrx-lora"], adapter_weights=[1.0])
-
-
-@spaces.GPU(duration=120)
-@torch.inference_mode()
-def generate_video(prompt, negative_prompt, height, width, num_frames, num_inference_steps, fps, seed, progress=gr.Progress(track_tqdm=True)):
-    INTERRUPT_PIPELINE = False
-    progress_steps = []
-    
-    # Randomize seed if seed is 0
-    if seed == 0:
-        seed = randomize_seed()
+# pipeline.enable_model_cpu_offload()
+
+pipeline.vae.enable_tiling()
+pipeline.vae.enable_slicing()
+
+pipeline = pipeline.to("cuda")
+
+
+MAX_SEED = np.iinfo(np.int32).max
+MAX_IMAGE_SIZE = 1280
+
+
+@spaces.GPU()
+def infer(
+    prompt,
+    negative_prompt,
+    seed,
+    randomize_seed,
+    width=704,
+    height=448,
+    num_frames=129,
+    fps=24,
+    num_inference_steps=30,
+    progress=gr.Progress(track_tqdm=True),
+):
+    if randomize_seed:
+        seed = random.randint(0, MAX_SEED)
+
+    generator = torch.Generator(device='cuda').manual_seed(seed)
+
+    with torch.amp.autocast_mode.autocast('cuda', torch.bfloat16), torch.no_grad(), torch.inference_mode():
+        video = pipeline(
+            prompt=prompt,
+            negative_prompt=negative_prompt,
+            width=width,
+            height=height,
+            num_frames=num_frames,
+            # guidance_scale=guidance_scale,
+            num_inference_steps=num_inference_steps,
+            # decode_timestep=decode_timestep,
+            # decode_noise_scale=decode_noise_scale,
+            generator=generator,
+            # max_sequence_length=512,
+        ).frames[0]
 
-    torch.cuda.empty_cache()
-    torch.cuda.synchronize()
-    # Generating the video <Does not support seed :( >
-    video = pipe(
-        prompt=prompt,
-        negative_prompt=negative_prompt,
-        width=width,
-        height=height,
-        num_frames=num_frames,
-        num_inference_steps=num_inference_steps,
-        generator=torch.Generator(device='cuda').manual_seed(seed),
-    ).frames[0]
-
-    # Create output filename based on prompt and timestamp
     timestamp = datetime.now().strftime("%Y%m%d_%H%M%S")
-    filename = f"{prompt[:10]}_{timestamp}.mp4"
-    
-    # Save the video to the output folder
-    os.makedirs("output_LTX091", exist_ok=True)
-    output_path = f"./output_LTX091/{filename}"
+    filename = f"output_{timestamp}.mp4"
+    os.makedirs("output", exist_ok=True)
+    output_path = f"./output/{filename}"
     export_to_video(video, output_path, fps=fps)
 
+    gc.collect
     torch.cuda.empty_cache()
-    gc.collect()
     return output_path
 
-# Gradio UI setup
-def randomize_seed():
-    return random.randint(0, 999999)
 
-with gr.Blocks() as demo:
-    with gr.Tabs():
-        with gr.Tab("Generate Video"):
-            with gr.Row():
-                prompt = gr.Textbox(label="Prompt", lines=3, value=initial_state.get("prompt", "A dramatic view of the pyramids at Giza during sunset."))
-                negative_prompt = gr.Textbox(label="Negative Prompt", lines=3, value=initial_state.get("negative_prompt", "worst quality, blurry, distorted"))
-            with gr.Row():
-                height = gr.Slider(label="Height", minimum=224, maximum=768, step=32, value=initial_state.get("height", 384))
-                width = gr.Slider(label="Width", minimum=320, maximum=1280, step=32, value=initial_state.get("width", 640))
-            with gr.Row():
-                num_frames = gr.Slider(label="Number of Frames", minimum=1, maximum=121, step=1, value=initial_state.get("num_frames", 49))
-                num_inference_steps = gr.Slider(label="Number of Inference Steps", minimum=1, maximum=30, step=1, value=initial_state.get("num_inference_steps", 20))
-            with gr.Row():
-                fps = gr.Slider(label="FPS", minimum=1, maximum=30, step=1, value=initial_state.get("fps", 16))
-                seed = gr.Number(label="Seed", value=initial_state.get("seed", 0))
-                random_seed_button = gr.Button("Randomize Seed")
-
-            output_video = gr.Video(label="Generated Video", show_label=True)
-            generate_button = gr.Button("Generate Video")
-            save_state_button = gr.Button("Save State")
-
-            random_seed_button.click(randomize_seed, outputs=seed)
-            generate_button.click(
-                generate_video,
-                inputs=[prompt, negative_prompt, height, width, num_frames, num_inference_steps, fps, seed],
-                outputs=output_video
+css = """
+#col-container {
+    margin: 0 auto;
+    max-width: 640px;
+}
+"""
+
+with gr.Blocks(css=css) as demo:
+    with gr.Column(elem_id="col-container"):
+        gr.Markdown(" # Text-to-Image Gradio Template")
+
+        with gr.Row():
+            prompt = gr.Textbox(
+                label="Prompt",
+                lines=3,
+                value=str("A woman with long brown hair and light skin smiles at another woman with long blonde hair. The woman with brown hair wears a black jacket and has a small, barely noticeable mole on her right cheek. The camera angle is a close-up, focused on the woman with brown hair's face. The lighting is warm and natural, likely from the setting sun, casting a soft glow on the scene. The scene appears to be real-life footage"),
             )
-            save_state_button.click(
-                save_ui_state,
-                inputs=[prompt, negative_prompt, height, width, num_frames, num_inference_steps, fps, seed],
-                outputs=gr.Text(label="State Status")
+
+            negative_prompt = gr.Textbox(
+                label="Negative prompt",
+                lines=3,
+                value=str("worst quality, blurry, distorted"),
             )
 
-        with gr.Tab("Batch Processing"):
-            with gr.Row():
-                batch_prompt = gr.Textbox(label="Prompt", lines=3, value="A batch of videos depicting different landscapes.")
-                batch_negative_prompt = gr.Textbox(label="Negative Prompt", lines=3, value="low quality, inconsistent, jittery")
-            with gr.Row():
-                batch_height = gr.Slider(label="Height", minimum=224, maximum=768, step=32, value=384)
-                batch_width = gr.Slider(label="Width", minimum=320, maximum=1280, step=32, value=640)
-            with gr.Row():
-                batch_num_frames = gr.Slider(label="Number of Frames", minimum=1, maximum=121, step=1, value=49)
-                batch_num_inference_steps = gr.Slider(label="Number of Inference Steps", minimum=1, maximum=30, step=1, value=20)
+        with gr.Row():
+            run_button = gr.Button("Run", scale=0, variant="huggingface")
+
+        with gr.Row():
+            result = gr.Video(label="Result", show_label=False)
+
+        with gr.Accordion("Advanced Settings", open=False):
+            seed = gr.Slider(
+                label="Seed",
+                minimum=0,
+                maximum=MAX_SEED,
+                step=1,
+                value=0,
+            )
+
+            randomize_seed = gr.Checkbox(label="Randomize seed", value=True)
+
             with gr.Row():
-                batch_fps = gr.Slider(label="FPS", minimum=1, maximum=30, step=1, value=16)
-                batch_seed = gr.Number(label="Seed", value=0)
-                random_seed_batch_button = gr.Button("Randomize Seed")
+                width = gr.Slider(
+                    label="Width",
+                    minimum=256,
+                    maximum=MAX_IMAGE_SIZE,
+                    step=32,
+                    value=704,  # Replace with defaults that work for your model
+                )
+
+                height = gr.Slider(
+                    label="Height",
+                    minimum=256,
+                    maximum=MAX_IMAGE_SIZE,
+                    step=32,
+                    value=448,  # Replace with defaults that work for your model
+                )
 
-            add_to_queue_button = gr.Button("Add to Queue")
-            clear_queue_button = gr.Button("Clear Queue")
-            process_queue_button = gr.Button("Process Queue")
+            with gr.Row():
+                num_frames = gr.Slider(
+                    label="Number of frames",
+                    minimum=1,
+                    maximum=257,
+                    step=32,
+                    value=129,  # Replace with defaults that work for your model
+                )
+
+                fps = gr.Slider(
+                    label="Number of frames per second",
+                    minimum=1,
+                    maximum=30,
+                    step=1,
+                    value=24,  # Replace with defaults that work for your model
+                )
 
-            queue_status = gr.Text(label="Queue Status")
+            with gr.Row():
 
-            random_seed_batch_button.click(randomize_seed, outputs=batch_seed)
-            add_to_queue_button.click(
-                add_to_queue,
-                inputs=[batch_prompt, batch_negative_prompt, batch_height, batch_width, batch_num_frames, batch_num_inference_steps, batch_fps, batch_seed],
-                outputs=queue_status
-            )
-            clear_queue_button.click(clear_queue, outputs=queue_status)
-            process_queue_button.click(process_queue, outputs=queue_status)
+                num_inference_steps = gr.Slider(
+                    label="Number of inference steps",
+                    minimum=1,
+                    maximum=50,
+                    step=1,
+                    value=30,  # Replace with defaults that work for your model
+                )
+
+    gr.on(
+        triggers=[run_button.click, prompt.submit],
+        fn=infer,
+        inputs=[
+            prompt,
+            negative_prompt,
+            seed,
+            randomize_seed,
+            width,
+            height,
+            num_frames,
+            fps,
+            num_inference_steps,
+        ],
+        outputs=[result],
+    )
 
-demo.launch()
+if __name__ == "__main__":
+    install_packages()
+    demo.launch()