import os import spaces import torch from diffusers.pipelines.wan.pipeline_wan_i2v import WanImageToVideoPipeline from diffusers.models.transformers.transformer_wan import WanTransformer3DModel from diffusers.utils.export_utils import export_to_video import gradio as gr import tempfile import numpy as np from PIL import Image import random import gc from torchao.quantization import quantize_ from torchao.quantization import Float8DynamicActivationFloat8WeightConfig, Int8WeightOnlyConfig import aoti # ========================================================= # MODEL CONFIGURATION # ========================================================= MODEL_ID = "Wan-AI/Wan2.2-I2V-A14B-Diffusers" HF_TOKEN = os.environ.get("HF_TOKEN") MAX_DIM = 832 MIN_DIM = 480 SQUARE_DIM = 640 MULTIPLE_OF = 16 MAX_SEED = np.iinfo(np.int32).max FIXED_FPS = 16 MIN_FRAMES_MODEL = 8 MAX_FRAMES_MODEL = 7720 MIN_DURATION = round(MIN_FRAMES_MODEL / FIXED_FPS, 1) MAX_DURATION = round(MAX_FRAMES_MODEL / FIXED_FPS, 1) # ========================================================= # LOAD PIPELINE # ========================================================= print("Loading pipeline components...") # Load models in bfloat16 transformer = WanTransformer3DModel.from_pretrained( MODEL_ID, subfolder="transformer", torch_dtype=torch.bfloat16, token=HF_TOKEN ) transformer_2 = WanTransformer3DModel.from_pretrained( MODEL_ID, subfolder="transformer_2", torch_dtype=torch.bfloat16, token=HF_TOKEN ) print("Assembling pipeline...") pipe = WanImageToVideoPipeline.from_pretrained( MODEL_ID, transformer=transformer, transformer_2=transformer_2, torch_dtype=torch.bfloat16, token=HF_TOKEN ) print("Moving to CUDA...") pipe = pipe.to("cuda") # ========================================================= # LOAD LORA ADAPTERS # ========================================================= print("Loading LoRA adapters...") try: pipe.load_lora_weights( "Kijai/WanVideo_comfy", weight_name="Lightx2v/lightx2v_I2V_14B_480p_cfg_step_distill_rank128_bf16.safetensors", adapter_name="lightx2v" ) pipe.load_lora_weights( "Kijai/WanVideo_comfy", weight_name="Lightx2v/lightx2v_I2V_14B_480p_cfg_step_distill_rank128_bf16.safetensors", adapter_name="lightx2v_2", load_into_transformer_2=True ) pipe.set_adapters(["lightx2v", "lightx2v_2"], adapter_weights=[1., 1.]) pipe.fuse_lora(adapter_names=["lightx2v"], lora_scale=3., components=["transformer"]) pipe.fuse_lora(adapter_names=["lightx2v_2"], lora_scale=1., components=["transformer_2"]) pipe.unload_lora_weights() print("LoRA loaded and fused successfully.") except Exception as e: print(f"Warning: Failed to load LoRA. Continuing without it. Error: {e}") # ========================================================= # QUANTIZATION & AOT OPTIMIZATION # ========================================================= print("Applying quantization...") torch.cuda.empty_cache() gc.collect() try: quantize_(pipe.text_encoder, Int8WeightOnlyConfig()) quantize_(pipe.transformer, Float8DynamicActivationFloat8WeightConfig()) quantize_(pipe.transformer_2, Float8DynamicActivationFloat8WeightConfig()) print("Loading AOTI blocks...") aoti.aoti_blocks_load(pipe.transformer, 'zerogpu-aoti/Wan2', variant='fp8da') aoti.aoti_blocks_load(pipe.transformer_2, 'zerogpu-aoti/Wan2', variant='fp8da') except Exception as e: print(f"Warning: Quantization/AOTI failed. Running in standard mode might OOM. Error: {e}") # ========================================================= # DEFAULT PROMPTS # ========================================================= default_prompt_i2v = "Make this image come alive with dynamic, cinematic human motion. Create smooth, natural, lifelike animation with fluid transitions, expressive body movement, realistic physics, and elegant camera flow. Deliver a polished, high-quality motion style that feels immersive, artistic, and visually captivating." default_negative_prompt = ( "low quality, worst quality, motion artifacts, unstable motion, jitter, frame jitter, wobbling limbs, motion distortion, inconsistent movement, robotic movement, animation-like motion, awkward transitions, incorrect body mechanics, unnatural posing, off-balance poses, broken motion paths, frozen frames, duplicated frames, frame skipping, warped motion, stretching artifacts bad anatomy, incorrect proportions, deformed body, twisted torso, broken joints, dislocated limbs, distorted neck, unnatural spine curvature, malformed hands, extra fingers, missing fingers, fused fingers, distorted legs, extra limbs, collapsed feet, floating feet, foot sliding, foot jitter, backward walking, unnatural gait blurry details, long exposure blur, ghosting, shadow trails, smearing, washed-out colors, overexposure, underexposure, excessive contrast, blown highlights, poorly rendered clothing, fabric glitches, texture warping, clothing merging with body, incorrect cloth physics ugly background, cluttered scene, crowded background, random objects, unwanted text, subtitles, logos, graffiti, grain, noise, static artifacts, compression noise, jpeg artifacts, image-like stillness, painting-like look, cartoon texture, low-resolution textures" ) # ========================================================= # IMAGE RESIZING LOGIC # ========================================================= def resize_image(image: Image.Image) -> Image.Image: width, height = image.size if width == height: return image.resize((SQUARE_DIM, SQUARE_DIM), Image.LANCZOS) aspect_ratio = width / height MAX_ASPECT_RATIO = MAX_DIM / MIN_DIM MIN_ASPECT_RATIO = MIN_DIM / MAX_DIM image_to_resize = image if aspect_ratio > MAX_ASPECT_RATIO: crop_width = int(round(height * MAX_ASPECT_RATIO)) left = (width - crop_width) // 2 image_to_resize = image.crop((left, 0, left + crop_width, height)) elif aspect_ratio < MIN_ASPECT_RATIO: crop_height = int(round(width / MIN_ASPECT_RATIO)) top = (height - crop_height) // 2 image_to_resize = image.crop((0, top, width, top + crop_height)) if width > height: target_w = MAX_DIM target_h = int(round(target_w / aspect_ratio)) else: target_h = MAX_DIM target_w = int(round(target_h * aspect_ratio)) final_w = round(target_w / MULTIPLE_OF) * MULTIPLE_OF final_h = round(target_h / MULTIPLE_OF) * MULTIPLE_OF final_w = max(MIN_DIM, min(MAX_DIM, final_w)) final_h = max(MIN_DIM, min(MAX_DIM, final_h)) return image_to_resize.resize((final_w, final_h), Image.LANCZOS) # ========================================================= # UTILITY FUNCTIONS # ========================================================= def get_num_frames(duration_seconds: float): return 1 + int(np.clip(int(round(duration_seconds * FIXED_FPS)), MIN_FRAMES_MODEL, MAX_FRAMES_MODEL)) # ========================================================= # MAIN GENERATION FUNCTION # ========================================================= @spaces.GPU(duration=180) def generate_video( input_image_path, # Receives file path now, not PIL object prompt, steps=4, negative_prompt=default_negative_prompt, duration_seconds=MAX_DURATION, guidance_scale=1, guidance_scale_2=1, seed=42, randomize_seed=False, progress=gr.Progress(track_tqdm=True), ): # Cleanup memory gc.collect() torch.cuda.empty_cache() try: # 1. Validation checks if not input_image_path: raise gr.Error("Please upload an input image.") if not os.path.exists(input_image_path): raise gr.Error("Image file not found! Please re-upload the image.") # 2. Manual Image Opening # We open it inside the function to avoid connection timeouts input_image = Image.open(input_image_path).convert("RGB") num_frames = get_num_frames(duration_seconds) current_seed = random.randint(0, MAX_SEED) if randomize_seed else int(seed) resized_image = resize_image(input_image) print(f"Generating video with seed: {current_seed}, frames: {num_frames}") output_frames_list = pipe( image=resized_image, prompt=prompt, negative_prompt=negative_prompt, height=resized_image.height, width=resized_image.width, num_frames=num_frames, guidance_scale=float(guidance_scale), guidance_scale_2=float(guidance_scale_2), num_inference_steps=int(steps), generator=torch.Generator(device="cuda").manual_seed(current_seed), ).frames[0] with tempfile.NamedTemporaryFile(suffix=".mp4", delete=False) as tmpfile: video_path = tmpfile.name export_to_video(output_frames_list, video_path, fps=FIXED_FPS) # Cleanup del output_frames_list del input_image del resized_image torch.cuda.empty_cache() return video_path, current_seed except Exception as e: print(f"Error during generation: {e}") raise gr.Error(f"Generation failed: {str(e)}") # ========================================================= # GRADIO UI # ========================================================= # Google Analytics Script ga_script = """ """ with gr.Blocks(theme=gr.themes.Soft(), head=ga_script) as demo: # --- PROFESSIONAL YOUTUBE EMBED SECTION --- gr.HTML("""
▶️

Imagination Engineering

Mastering AI & Creative Tech

""") # --------------------------------------------- gr.Markdown("# 🚀 Dream Wan 2.2 Faster Pro (14B) — Ultra Fast I2V with Lightning LoRA") gr.Markdown("Optimized FP8 quantized pipeline with AoT blocks & 4-step fast inference ⚡") with gr.Row(): with gr.Column(): # CHANGE: type="filepath" fixes the file not found error input_image_component = gr.Image(type="filepath", label="Input Image") prompt_input = gr.Textbox(label="Prompt", value=default_prompt_i2v) duration_seconds_input = gr.Slider( minimum=MIN_DURATION, maximum=MAX_DURATION, step=0.1, value=3.5, label="Duration (seconds)", info=f"Model range: {MIN_FRAMES_MODEL}-{MAX_FRAMES_MODEL} frames at {FIXED_FPS}fps." ) with gr.Accordion("Advanced Settings", open=False): negative_prompt_input = gr.Textbox(label="Negative Prompt", value=default_negative_prompt, lines=3) seed_input = gr.Slider(label="Seed", minimum=0, maximum=MAX_SEED, step=1, value=42) randomize_seed_checkbox = gr.Checkbox(label="Randomize seed", value=True) steps_slider = gr.Slider(minimum=1, maximum=30, step=1, value=6, label="Inference Steps") guidance_scale_input = gr.Slider(minimum=0.0, maximum=10.0, step=0.5, value=1, label="Guidance Scale (high noise)") guidance_scale_2_input = gr.Slider(minimum=0.0, maximum=10.0, step=0.5, value=1, label="Guidance Scale 2 (low noise)") generate_button = gr.Button("🎬 Generate Video", variant="primary") with gr.Column(): video_output = gr.Video(label="Generated Video", autoplay=True) ui_inputs = [ input_image_component, prompt_input, steps_slider, negative_prompt_input, duration_seconds_input, guidance_scale_input, guidance_scale_2_input, seed_input, randomize_seed_checkbox ] generate_button.click(fn=generate_video, inputs=ui_inputs, outputs=[video_output, seed_input]) # --- BOTTOM ADVERTISEMENT BANNER --- gr.HTML("""

✨ New: Dream Hub Pro (All-in-One)

Access all your pro tools (Wan2.1, Qwen, Audio, Video Enhance) in one place!

""") if __name__ == "__main__": demo.queue().launch()