app.py

import torch
from diffusers import AutoencoderKLWan, WanPipeline, UniPCMultistepScheduler
from diffusers.utils import export_to_video
import gradio as gr
import tempfile
import spaces
from huggingface_hub import hf_hub_download
import numpy as np
import random
import os
# LIGHT WEIGHT 1.3b
# MODEL_ID = "Wan-AI/Wan2.1-T2V-1.3B-Diffusers"
# LORA_REPO_ID = "Kijai/WanVideo_comfy"
# LORA_FILENAME = "Wan21_CausVid_bidirect2_T2V_1_3B_lora_rank32.safetensors"


MODEL_ID = "Wan-AI/Wan2.1-T2V-14B-Diffusers"

#MODEL_ID ="FastWan/FastWan_T2V_14B_480p_lora_rank_128_bf16.safetensors"


LORA_REPO_ID = "Kijai/WanVideo_comfy"
LORA_FILENAME = "Lightx2v/lightx2v_T2V_14B_cfg_step_distill_v2_lora_rank256_bf16.safetensors"
#LORA_FILENAME = "Pusa/Wan21_PusaV1_LoRA_14B_rank512_bf16.safetensors"




# LORA_REPO_ID = "RaphaelLiu/PusaV1"
# LORA_FILENAME="pusa_v1.safetensors"
#LORA_REPO_ID = "Kijai/WanVideo_comfy"
#LORA_FILENAME = "Wan21_CausVid_14B_T2V_lora_rank32.safetensors"








vae = AutoencoderKLWan.from_pretrained(MODEL_ID, subfolder="vae", torch_dtype=torch.float32)
pipe = WanPipeline.from_pretrained(
    MODEL_ID, vae=vae, torch_dtype=torch.bfloat16
)
pipe.scheduler = UniPCMultistepScheduler.from_config(pipe.scheduler.config, flow_shift=8.0)
pipe.to("cuda")

causvid_path = hf_hub_download(repo_id=LORA_REPO_ID, filename=LORA_FILENAME)
pipe.load_lora_weights(causvid_path, adapter_name="causvid_lora")
pipe.set_adapters(["causvid_lora"], adapter_weights=[0.95])
pipe.fuse_lora()

# MOD_VALUE = 32
# DEFAULT_H_SLIDER_VALUE =  512
# DEFAULT_W_SLIDER_VALUE =  896

# # Environment variable check
# IS_ORIGINAL_SPACE = os.environ.get("IS_ORIGINAL_SPACE", "True") == "True"

# # Original limits
# ORIGINAL_SLIDER_MIN_H, ORIGINAL_SLIDER_MAX_H = 128, 1280
# ORIGINAL_SLIDER_MIN_W, ORIGINAL_SLIDER_MAX_W = 128, 1280
# ORIGINAL_MAX_DURATION = round(81/24, 1)  # MAX_FRAMES_MODEL/FIXED_FPS

# # Limited space constants
# LIMITED_MAX_RESOLUTION = 640
# LIMITED_MAX_DURATION = 2.0
# LIMITED_MAX_STEPS = 4

# # Set limits based on environment variable
# if IS_ORIGINAL_SPACE:
#     SLIDER_MIN_H, SLIDER_MAX_H = 128, LIMITED_MAX_RESOLUTION
#     SLIDER_MIN_W, SLIDER_MAX_W = 128, LIMITED_MAX_RESOLUTION
#     MAX_DURATION = LIMITED_MAX_DURATION
#     MAX_STEPS = LIMITED_MAX_STEPS
# else:
#     SLIDER_MIN_H, SLIDER_MAX_H = ORIGINAL_SLIDER_MIN_H, ORIGINAL_SLIDER_MAX_H
#     SLIDER_MIN_W, SLIDER_MAX_W = ORIGINAL_SLIDER_MIN_W, ORIGINAL_SLIDER_MAX_W
#     MAX_DURATION = ORIGINAL_MAX_DURATION
#     MAX_STEPS = 8

# MAX_SEED = np.iinfo(np.int32).max

# FIXED_FPS = 24
# FIXED_OUTPUT_FPS = 18 # we downspeed the output video as a temporary "trick"
# MIN_FRAMES_MODEL = 8
# MAX_FRAMES_MODEL = 81 


#New math to make it High Res

MOD_VALUE = 32

# Defaults for higher-res generation
DEFAULT_H_SLIDER_VALUE = 768
DEFAULT_W_SLIDER_VALUE = 1344  # 16:9 friendly and divisible by MOD_VALUE

# Original Space = Hugging Face space with compute limits
IS_ORIGINAL_SPACE = os.environ.get("IS_ORIGINAL_SPACE", "True") == "True"

# Conservative limits for low-end environments
LIMITED_MAX_RESOLUTION = 640
LIMITED_MAX_DURATION = 2.0
LIMITED_MAX_STEPS = 4

# Generous limits for local or Pro spaces
ORIGINAL_SLIDER_MIN_H, ORIGINAL_SLIDER_MAX_H = 128, 1536
ORIGINAL_SLIDER_MIN_W, ORIGINAL_SLIDER_MAX_W = 128, 1536
ORIGINAL_MAX_DURATION = round(81 / 24, 1)  # 3.4 seconds
ORIGINAL_MAX_STEPS = 8

# Use limited or original (generous) settings
if IS_ORIGINAL_SPACE:
    SLIDER_MIN_H, SLIDER_MAX_H = 128, LIMITED_MAX_RESOLUTION
    SLIDER_MIN_W, SLIDER_MAX_W = 128, LIMITED_MAX_RESOLUTION
    MAX_DURATION = LIMITED_MAX_DURATION
    MAX_STEPS = LIMITED_MAX_STEPS
else:
    SLIDER_MIN_H, SLIDER_MAX_H = ORIGINAL_SLIDER_MIN_H, ORIGINAL_SLIDER_MAX_H
    SLIDER_MIN_W, SLIDER_MAX_W = ORIGINAL_SLIDER_MIN_W, ORIGINAL_SLIDER_MAX_W
    MAX_DURATION = ORIGINAL_MAX_DURATION
    MAX_STEPS = ORIGINAL_MAX_STEPS

MAX_SEED = np.iinfo(np.int32).max

FIXED_FPS = 24
FIXED_OUTPUT_FPS = 18  # reduce final video FPS to save space
MIN_FRAMES_MODEL = 8
MAX_FRAMES_MODEL = 81


default_prompt_t2v = "cinematic footage, group of pedestrians dancing in the streets of NYC, high quality breakdance, 4K, tiktok video, intricate details, instagram feel, dynamic camera, smooth dance motion, dimly lit, stylish, beautiful faces, smiling, music video"
default_negative_prompt = "Bright tones, overexposed, static, blurred details, subtitles, style, works, paintings, images, static, overall gray, worst quality, low quality, JPEG compression residue, ugly, incomplete, extra fingers, poorly drawn hands, poorly drawn faces, deformed, disfigured, misshapen limbs, fused fingers, still picture, messy background, three legs, many people in the background, walking backwards, watermark, text, signature"

def get_duration(prompt, height, width, 
                   negative_prompt, duration_seconds,
                   guidance_scale, steps,
                   seed, randomize_seed, 
                   progress):
    if steps > 4 and duration_seconds > 2:
        return 90
    elif steps > 4 or duration_seconds > 2:
        return 75
    else:
        return 60

@spaces.GPU(duration=get_duration)
def generate_video(prompt, height, width, 
                   negative_prompt=default_negative_prompt, duration_seconds = 2,
                   guidance_scale = 1, steps = 4,
                   seed = 42, randomize_seed = False, 
                   progress=gr.Progress(track_tqdm=True)):
    """
    Generate a video from a text prompt using the Wan 2.1 T2V model with CausVid LoRA.
    
    This function takes a text prompt and generates a video based on the provided
    prompt and parameters. It uses the Wan 2.1 1.3B Text-to-Video model with CausVid LoRA
    for fast generation in 3-8 steps.
    
    Args:
        prompt (str): Text prompt describing the desired video content.
        height (int): Target height for the output video. Will be adjusted to multiple of MOD_VALUE (32).
        width (int): Target width for the output video. Will be adjusted to multiple of MOD_VALUE (32).
        negative_prompt (str, optional): Negative prompt to avoid unwanted elements. 
            Defaults to default_negative_prompt (contains unwanted visual artifacts).
        duration_seconds (float, optional): Duration of the generated video in seconds.
            Defaults to 2. Clamped between MIN_FRAMES_MODEL/FIXED_FPS and MAX_FRAMES_MODEL/FIXED_FPS.
        guidance_scale (float, optional): Controls adherence to the prompt. Higher values = more adherence.
            Defaults to 1.0. Range: 0.0-20.0.
        steps (int, optional): Number of inference steps. More steps = higher quality but slower.
            Defaults to 4. Range: 1-30.
        seed (int, optional): Random seed for reproducible results. Defaults to 42.
            Range: 0 to MAX_SEED (2147483647).
        randomize_seed (bool, optional): Whether to use a random seed instead of the provided seed.
            Defaults to False.
        progress (gr.Progress, optional): Gradio progress tracker. Defaults to gr.Progress(track_tqdm=True).
    
    Returns:
        tuple: A tuple containing:
            - video_path (str): Path to the generated video file (.mp4)
            - current_seed (int): The seed used for generation (useful when randomize_seed=True)
    
    Raises:
        gr.Error: If prompt is empty or None.
    
    Note:
        - Frame count is calculated as duration_seconds * FIXED_FPS (24)
        - Output dimensions are adjusted to be multiples of MOD_VALUE (32)
        - The function uses GPU acceleration via the @spaces.GPU decorator
        - Generation time varies based on steps and duration (see get_duration function)
    """
    if not prompt or prompt.strip() == "":
        raise gr.Error("Please enter a text prompt. Try to use long and precise descriptions.")

    # Apply limits based on environment variable
    if IS_ORIGINAL_SPACE:
        height = min(height, LIMITED_MAX_RESOLUTION)
        width = min(width, LIMITED_MAX_RESOLUTION)
        duration_seconds = min(duration_seconds, LIMITED_MAX_DURATION)
        steps = min(steps, LIMITED_MAX_STEPS)

    target_h = max(MOD_VALUE, (int(height) // MOD_VALUE) * MOD_VALUE)
    target_w = max(MOD_VALUE, (int(width) // MOD_VALUE) * MOD_VALUE)
    
    num_frames = np.clip(int(round(duration_seconds * FIXED_FPS)), MIN_FRAMES_MODEL, MAX_FRAMES_MODEL)
    
    current_seed = random.randint(0, MAX_SEED) if randomize_seed else int(seed)

    with torch.inference_mode():
        output_frames_list = pipe(
            prompt=prompt, negative_prompt=negative_prompt,
            height=target_h, width=target_w, num_frames=num_frames,
            guidance_scale=float(guidance_scale), 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_OUTPUT_FPS)
    return video_path, current_seed
    
    

with gr.Blocks(css="body { max-width: 100vw; overflow-x: hidden; }") as demo:
    gr.HTML('<meta name="viewport" content="width=device-width, initial-scale=1">')
    # ... your other components here ...
    gr.Markdown("# ⚡ InstaVideo")
    gr.Markdown("This Gradio space is a fork of [wan2-1-fast from multimodalart](https://huggingface.co/spaces/multimodalart/wan2-1-fast), and is powered by the Wan CausVid LoRA [from Kijai](https://huggingface.co/Kijai/WanVideo_comfy/blob/main/Wan21_CausVid_bidirect2_T2V_1_3B_lora_rank32.safetensors).")

    # Add notice for limited spaces
    if IS_ORIGINAL_SPACE:
        gr.Markdown("⚠️ **This free public demo limits the resolution to 640px, duration to 2s, and inference steps to 4. For full capabilities please duplicate this space.**")
    
    with gr.Row():
        with gr.Column():
            prompt_input = gr.Textbox(label="Prompt", value=default_prompt_t2v, placeholder="Describe the video you want to generate...")
            
            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, interactive=True)
                randomize_seed_checkbox = gr.Checkbox(label="Randomize seed", value=True, interactive=True)
                with gr.Row():
                    height_input = gr.Slider(
                        minimum=SLIDER_MIN_H, 
                        maximum=SLIDER_MAX_H, 
                        step=MOD_VALUE, 
                        value=min(DEFAULT_H_SLIDER_VALUE, SLIDER_MAX_H), 
                        label=f"Output Height (multiple of {MOD_VALUE})"
                    )
                    width_input = gr.Slider(
                        minimum=SLIDER_MIN_W, 
                        maximum=SLIDER_MAX_W, 
                        step=MOD_VALUE, 
                        value=min(DEFAULT_W_SLIDER_VALUE, SLIDER_MAX_W), 
                        label=f"Output Width (multiple of {MOD_VALUE})"
                    )
                duration_seconds_input = gr.Slider(
                    minimum=round(MIN_FRAMES_MODEL/FIXED_FPS,1), 
                    maximum=MAX_DURATION, 
                    step=0.1, 
                    value=2, 
                    label="Duration (seconds)", 
                    info=f"Clamped to model's {MIN_FRAMES_MODEL}-{MAX_FRAMES_MODEL} frames at {FIXED_FPS}fps."
                )
                steps_slider = gr.Slider(minimum=1, maximum=MAX_STEPS, step=1, value=4, label="Inference Steps") 
                guidance_scale_input = gr.Slider(minimum=0.0, maximum=20.0, step=0.5, value=1.0, label="Guidance Scale", visible=False)

            generate_button = gr.Button("Generate Video", variant="primary")
        with gr.Column():
            video_output = gr.Video(label="Generated Video", autoplay=True, interactive=False)
    
    ui_inputs = [
        prompt_input, height_input, width_input,
        negative_prompt_input, duration_seconds_input,
        guidance_scale_input, steps_slider, seed_input, randomize_seed_checkbox
    ]
    generate_button.click(fn=generate_video, inputs=ui_inputs, outputs=[video_output, seed_input])

    # Adjust examples based on space limits
    example_configs = [
        ["a majestic eagle soaring through mountain peaks, cinematic aerial view", 896, 512],
        ["a serene ocean wave crashing on a sandy beach at sunset", 448, 832],
        ["a field of flowers swaying in the wind, spring morning light", 512, 896],
    ]
    
    if IS_ORIGINAL_SPACE:
        # Limit example resolutions for limited spaces
        example_configs = [
            [example[0], min(example[1], LIMITED_MAX_RESOLUTION), min(example[2], LIMITED_MAX_RESOLUTION)]
            for example in example_configs
        ]
    
    gr.Examples(
        examples=example_configs,
        inputs=[prompt_input, height_input, width_input], 
        outputs=[video_output, seed_input], 
        fn=generate_video, 
        cache_examples="lazy"
    )

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