import os
import hmac
import spaces
import torch
from diffusers.pipelines.wan.pipeline_wan_i2v import WanImageToVideoPipeline
from diffusers.models.transformers.transformer_wan import WanTransformer3DModel
from diffusers import FlowMatchEulerDiscreteScheduler
import gradio as gr
import tempfile
import numpy as np
import imageio.v2 as imageio
import shutil
import time
from PIL import Image
import random
import gc
from datetime import datetime
from huggingface_hub import CommitOperationAdd, HfApi
from uuid import uuid4
from modify_model.modify_wan import set_sage_attn_wan
from sageattention import sageattn
from torchao.quantization import quantize_
from torchao.quantization import Float8DynamicActivationFloat8WeightConfig
from torchao.quantization import Int8WeightOnlyConfig
import warnings
import aoti
os.environ["TOKENIZERS_PARALLELISM"] = "true"
warnings.filterwarnings("ignore")
key=os.environ.get("DS_APIKEY")
diffusers_apikey=os.environ.get("DIFFUSERS_APIKEY")
MODEL_ID = "Wan-AI/Wan2.2-I2V-A14B-Diffusers"
DS_ID = os.environ.get("DS_ID")
PRIVATE_MODEL_KEY = os.environ.get("PRIVATE_MODEL_KEY")
hf_api = HfApi(token=key) if key and DS_ID else None
GENERATION_SECRET=os.environ.get("GENERATION_SECRET")
PUBLIC_ENABLED=os.environ.get("PUBLIC_ENABLED")
PUBLIC_ENABLED = str(PUBLIC_ENABLED).strip().lower() == "true"
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 = 128
POSTPROCESS_OVERHEAD_SECONDS = 25
MIN_DURATION = round(MIN_FRAMES_MODEL/FIXED_FPS,1)
MAX_DURATION = round(MAX_FRAMES_MODEL/FIXED_FPS,1)
pipe = WanImageToVideoPipeline.from_pretrained(
"TestOrganizationPleaseIgnore/WAMU_v2_WAN2.2_I2V_LIGHTNING",
torch_dtype=torch.bfloat16,
).to('cuda')
# Use sage attention for speed
"""
pipe.load_lora_weights(
"lightx2v/Wan2.2-Distill-Loras",
weight_name="wan2.2_i2v_A14b_high_noise_lora_rank64_lightx2v_4step_1022.safetensors",
adapter_name="lightx2v"
)
"""
pipe.load_lora_weights(
"obsxrver/Wan2.2-I2Pee-5XL",
weight_name="WAN2.2-I2V_HighNoise_I2Pee-5.5XL.safetensors",
adapter_name="i2pee_high",
token=PRIVATE_MODEL_KEY,
)
kwargs_lora = {}
kwargs_lora["load_into_transformer_2"] = True
"""
pipe.load_lora_weights(
"lightx2v/Wan2.2-Distill-Loras",
weight_name="wan2.2_i2v_A14b_low_noise_lora_rank64_lightx2v_4step_1022.safetensors",
adapter_name="lightx2v_2", **kwargs_lora
)
"""
pipe.load_lora_weights(
"obsxrver/Wan2.2-I2Pee-5XL",
weight_name="WAN2.2-I2V_LowNoise_I2Pee-5.5XL.safetensors",
adapter_name="i2pee_low",token=PRIVATE_MODEL_KEY, **kwargs_lora
)
pipe.set_adapters(["i2pee_high", "i2pee_low"], adapter_weights=[1., 1.])
#pipe.fuse_lora(adapter_names=["lightx2v"], lora_scale=0.9, components=["transformer"])
pipe.fuse_lora(adapter_names=["i2pee_high"], lora_scale=1., components=["transformer"])
#pipe.fuse_lora(adapter_names=["lightx2v_2"], lora_scale=1., components=["transformer_2"])
pipe.fuse_lora(adapter_names=["i2pee_low"], lora_scale=1., components=["transformer_2"])
pipe.unload_lora_weights()
print("Quantizing model to fp8da")
quantize_(pipe.text_encoder, Int8WeightOnlyConfig())
quantize_(pipe.transformer, Float8DynamicActivationFloat8WeightConfig())
quantize_(pipe.transformer_2, Float8DynamicActivationFloat8WeightConfig())
# needs this to work.
aoti.aoti_blocks_load(pipe.transformer, 'zerogpu-aoti/Wan2', variant='fp8da')
aoti.aoti_blocks_load(pipe.transformer_2, 'zerogpu-aoti/Wan2', variant='fp8da')
#pipe.scheduler=FlowMatchEulerDiscreteScheduler.from_config(pipe.scheduler.config)
print("Patching model to use Sage Attention")
set_sage_attn_wan(pipe.transformer,sageattn)
set_sage_attn_wan(pipe.transformer_2,sageattn)
default_prompt_i2v = "a woman, jumpcut, after the transition, the woman is nude, lying on a bed on her stomach with her hands gripping the bed in front of her, a nude man is standing in front of the woman with his penis inside her mouth, he is holding her head with his right hand with his left hand at his side, the man is pissing in the woman's mouth while his penis is inside her mouth, piss spills out of her mouth and drips onto the bed"
default_negative_prompt = "色调艳丽, 过曝, 静态, 细节模糊不清, 字幕, 风格, 作品, 画作, 画面, 静止, 整体发灰, 最差质量, 低质量, JPEG压缩残留, 丑陋的, 残缺的, 多余的手指, 画得不好的手部, 画得不好的脸部, 畸形的, 毁容的, 形态畸形的肢体, 手指融合, 静止不动的画面, 杂乱的背景, 三条腿, 背景人很多, 倒着走"
def clear_vram():
gc.collect()
torch.cuda.empty_cache()
def resize_image(image: Image.Image) -> Image.Image:
"""
Resizes an image to fit within the model's constraints, preserving aspect ratio as much as possible.
"""
width, height = image.size
# Handle square case
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:
# Very wide image -> crop width to fit 832x480 aspect ratio
target_w, target_h = MAX_DIM, MIN_DIM
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:
# Very tall image -> crop height to fit 480x832 aspect ratio
target_w, target_h = MIN_DIM, MAX_DIM
crop_height = int(round(width / MIN_ASPECT_RATIO))
top = (height - crop_height) // 2
image_to_resize = image.crop((0, top, width, top + crop_height))
else:
if width > height: # Landscape
target_w = MAX_DIM
target_h = int(round(target_w / aspect_ratio))
else: # Portrait
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)
def get_num_frames(duration_seconds: float):
return 1 + int(np.clip(
int(round(duration_seconds * FIXED_FPS)),
MIN_FRAMES_MODEL,
MAX_FRAMES_MODEL,
))
def to_ds(
*,
video_path: str,
original_image_path: str,
original_image_name: str,
prompt: str,
negative_prompt: str,
duration_seconds: float,
steps: int,
guidance_scale: float,
guidance_scale_2: float,
seed: int,
output_width: int,
output_height: int,
):
if hf_api is None:
print("cannot complete operation.")
return
timestamp = datetime.now()
folder = timestamp.strftime("%m/%d")
stem = timestamp.strftime("%H%M%S-%f")
unique_suffix = uuid4().hex[:8]
base_name = f"{stem}-{unique_suffix}"
video_repo_path = f"{folder}/{base_name}.mp4"
image_repo_path = f"{folder}/{base_name}.png"
text_repo_path = f"{folder}/{base_name}.txt"
metadata = "\n".join([
f"timestamp={timestamp.isoformat()}",
f"original_image_name={original_image_name}",
f"prompt={prompt}",
f"negative_prompt={negative_prompt}",
f"steps={steps}",
f"duration_seconds={duration_seconds}",
f"guidance_scale={guidance_scale}",
f"guidance_scale_2={guidance_scale_2}",
f"seed={seed}",
f"resolution={output_width}x{output_height}",
f"fps={FIXED_FPS}",
])
try:
hf_api.create_commit(
repo_id=DS_ID,
repo_type="dataset",
commit_message=f"Add generation {base_name}",
operations=[
CommitOperationAdd(
path_in_repo=video_repo_path,
path_or_fileobj=video_path,
),
CommitOperationAdd(
path_in_repo=image_repo_path,
path_or_fileobj=original_image_path,
),
CommitOperationAdd(
path_in_repo=text_repo_path,
path_or_fileobj=metadata.encode("utf-8"),
),
],
)
print(f"{DS_ID}:{video_repo_path}")
print(f"{DS_ID}:{image_repo_path}")
print(f"{DS_ID}:{text_repo_path}")
except Exception as exc:
print(f"failed: {exc}")
def load_input_image(image_path: str) -> Image.Image:
if not image_path:
raise gr.Error("Please upload an input image.")
with Image.open(image_path) as image:
return image.convert("RGB")
def get_duration(
input_image,
prompt,
steps,
negative_prompt,
duration_seconds,
guidance_scale,
guidance_scale_2,
seed,
randomize_seed,
access_granted,
request,
progress,
):
BASE_FRAMES_HEIGHT_WIDTH = 81 * 832 * 624
BASE_STEP_DURATION = 15
image = load_input_image(input_image)
width, height = resize_image(image).size
frames = get_num_frames(duration_seconds)
factor = frames * width * height / BASE_FRAMES_HEIGHT_WIDTH
step_duration = BASE_STEP_DURATION * factor ** 1.5
# Keep duration estimates conservative to avoid ZeroGPU task aborts right after inference.
return 20 + int(steps) * step_duration + POSTPROCESS_OVERHEAD_SECONDS
def get_original_image_name(image_path: str) -> str:
if not image_path:
return "unknown"
return os.path.basename(image_path)
def get_original_media_stem(media_path: str) -> str:
if not media_path:
return "generated-video"
media_name = os.path.basename(media_path)
media_stem, _ = os.path.splitext(media_name)
media_stem = media_stem.strip()
return media_stem or "generated-video"
def build_download_video(video_path: str, original_media_path: str) -> str:
download_dir = tempfile.mkdtemp(prefix="wan22-download-")
download_filename = f"{get_original_media_stem(original_media_path)}.mp4"
download_path = os.path.join(download_dir, download_filename)
shutil.copyfile(video_path, download_path)
return download_path
def export_to_video_h264(frames, output_path: str, fps: int = FIXED_FPS) -> None:
"""
Export frames to an H.264 MP4 compatible with mobile browsers/players.
"""
if len(frames) == 0:
raise ValueError("No frames to export.")
normalized_frames = []
for frame in frames:
if isinstance(frame, Image.Image):
arr = np.array(frame)
else:
arr = np.asarray(frame)
if arr.ndim == 2:
arr = np.stack([arr, arr, arr], axis=-1)
elif arr.ndim == 3 and arr.shape[2] == 4:
arr = arr[:, :, :3]
if arr.ndim != 3 or arr.shape[2] != 3:
raise ValueError(f"Unsupported frame shape: {arr.shape}")
if arr.dtype != np.uint8:
if np.issubdtype(arr.dtype, np.floating):
max_val = float(np.nanmax(arr)) if arr.size else 1.0
if max_val <= 1.0:
arr = arr * 255.0
arr = np.clip(arr, 0, 255).astype(np.uint8)
normalized_frames.append(arr)
height, width = normalized_frames[0].shape[:2]
ffmpeg_params = [
"-movflags",
"+faststart",
"-profile:v",
"baseline",
"-level",
"3.1",
]
with imageio.get_writer(
output_path,
format="FFMPEG",
mode="I",
fps=fps,
codec="libx264",
pixelformat="yuv420p",
ffmpeg_params=ffmpeg_params,
) as writer:
for frame in normalized_frames:
if frame.shape[:2] != (height, width):
frame = np.array(
Image.fromarray(frame).resize((width, height), Image.LANCZOS)
)
writer.append_data(frame)
def reset_verification():
if PUBLIC_ENABLED:
return (
gr.update(value="Generate Video", interactive=True),
"Public generation is enabled.",
True,
)
return (
gr.update(value="Verify Passcode to Enable Generation", interactive=False),
"Enter the passcode and click Verify to unlock generation.",
False,
)
def verify_passcode(passcode: str):
if PUBLIC_ENABLED:
return (
gr.update(value="Generate Video", interactive=True),
"Public generation is enabled.",
True,
)
secret = GENERATION_SECRET or ""
candidate = (passcode or "").strip()
if not secret:
return (
gr.update(value="Verify Passcode to Enable Generation", interactive=False),
"Generation is unavailable because `GENERATION_SECRET` is not configured.",
False,
)
if hmac.compare_digest(candidate, secret):
return (
gr.update(value="Generate Video", interactive=True),
"Passcode verified. Generation unlocked.",
True,
)
return (
gr.update(value="Verify Passcode to Enable Generation", interactive=False),
"Incorrect passcode.",
False,
)
@spaces.GPU(duration=get_duration)
def generate_video(
input_image,
prompt,
steps = 6,
negative_prompt=default_negative_prompt,
duration_seconds = 5.0,
guidance_scale = 1,
guidance_scale_2 = 1,
seed = 696969696,
randomize_seed = False,
access_granted = True,
request: gr.Request = None,
progress=gr.Progress(track_tqdm=True),
):
"""
Generate a video from an input image using the Wan 2.2 14B I2V model with Lightning LoRA.
This function takes an input image and generates a video animation based on the provided
prompt and parameters. It uses an FP8 qunatized Wan 2.2 14B Image-to-Video model in with Lightning LoRA
for fast generation in 4-8 steps.
Args:
input_image (str): Filepath to the input image to animate. Will be loaded and resized to target dimensions.
prompt (str): Text prompt describing the desired animation or motion.
steps (int, optional): Number of inference steps. More steps = higher quality but slower.
Defaults to 4. Range: 1-30.
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.
guidance_scale_2 (float, optional): Controls adherence to the prompt. Higher values = more adherence.
Defaults to 1.0. Range: 0.0-20.0.
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 input_image is None (no image uploaded).
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 access_granted:
raise gr.Error("Please verify the passcode before generating.")
if input_image is None:
raise gr.Error("Please upload an input image.")
clear_vram()
start_time = time.perf_counter()
num_frames = get_num_frames(duration_seconds)
current_seed = random.randint(0, MAX_SEED) if randomize_seed else int(seed)
input_image_path = input_image
input_image = load_input_image(input_image_path)
resized_image = resize_image(input_image)
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
with tempfile.NamedTemporaryFile(suffix=".png", delete=False) as tmpfile:
original_image_upload_path = tmpfile.name
input_image.save(original_image_upload_path, format="PNG")
export_to_video_h264(output_frames_list, video_path, fps=FIXED_FPS)
try:
to_ds(
video_path=video_path,
original_image_path=original_image_upload_path,
original_image_name=get_original_image_name(input_image_path),
prompt=prompt if prompt is not None else default_prompt_i2v,
negative_prompt=negative_prompt if negative_prompt is not None else default_negative_prompt,
duration_seconds=float(duration_seconds) if duration_seconds is not None else MAX_DURATION,
steps=int(steps) if steps is not None else 6,
guidance_scale=float(guidance_scale) if guidance_scale is not None else 1,
guidance_scale_2=float(guidance_scale_2) if guidance_scale_2 is not None else 1,
seed=current_seed if current_seed is not None else 42,
output_width=resized_image.width if resized_image.width is not None else 832,
output_height=resized_image.height if resized_image.height is not None else 480,
)
finally:
if os.path.exists(original_image_upload_path):
os.remove(original_image_upload_path)
elapsed_seconds = time.perf_counter() - start_time
render_message = f"Generated in {elapsed_seconds:.1f} seconds."
download_path = build_download_video(video_path, input_image_path)
return video_path, current_seed, render_message, download_path
custom_css = """
.gradio-container {
max-width: 1180px !important;
margin: 0 auto !important;
background:
radial-gradient(circle at 10% 20%, rgba(124, 58, 237, 0.2), transparent 35%),
radial-gradient(circle at 90% 10%, rgba(14, 165, 233, 0.15), transparent 30%),
linear-gradient(180deg, #070b1a 0%, #0f172a 100%);
}
#hero-card {
border-radius: 18px;
border: 1px solid rgba(255, 255, 255, 0.12);
background: linear-gradient(130deg, rgba(76, 29, 149, 0.9), rgba(30, 64, 175, 0.9));
padding: 16px 20px;
box-shadow: 0 18px 45px rgba(10, 10, 20, 0.35);
margin-bottom: 10px;
}
#hero-card h1 {
margin: 0 0 6px 0;
font-size: 1.9rem;
line-height: 1.1;
}
#hero-card p {
margin: 0;
opacity: 0.95;
}
#control-panel,
#preview-panel {
border-radius: 16px;
border: 1px solid rgba(255, 255, 255, 0.1);
background: rgba(15, 23, 42, 0.72);
backdrop-filter: blur(6px);
padding: 16px;
box-shadow: 0 14px 35px rgba(0, 0, 0, 0.22);
}
#generated-video {
min-height: 500px;
}
#generate-button {
min-height: 48px;
font-weight: 700;
letter-spacing: 0.2px;
}
.status-note {
margin-top: 8px;
padding: 10px 12px;
border-radius: 10px;
border: 1px solid rgba(255, 255, 255, 0.16);
background: rgba(15, 118, 110, 0.2);
}
"""
theme = gr.themes.Soft(
primary_hue=gr.themes.colors.violet,
secondary_hue=gr.themes.colors.cyan,
neutral_hue=gr.themes.colors.slate,
).set(
body_background_fill="#050816",
block_background_fill="#111827",
block_border_width="1px",
block_title_text_weight="700",
)
with gr.Blocks(theme=theme, css=custom_css) as demo:
gr.HTML(
"""
Wan2.2 I2V + I2Pee-5.5XL Faster
"""
)
if PUBLIC_ENABLED:
access_copy = (
"Updated: 1.5x speed boost and massive quality increase! FREE TO ALL FOR A LIMITED TIME!"
"Subscribers on fanvue unlock permanent generation access "
"and can download the I2Pee-5XL LoRA."
)
else:
access_copy = (
"Insiders-only mode. 🔒"
"Public access is disabled. Become an insider on fanvue to unlock generation access and download the I2Pee-5.5XL LoRA."
)
gr.HTML(f"{access_copy}
")
gr.Markdown("### Join the discord: [discord.gg/8wMr2mmaRF](https://discord.gg/8wMr2mmaRF)")
with gr.Row(equal_height=True):
with gr.Column(elem_id="control-panel", scale=5):
gr.Markdown("### Input + Prompt Controls")
input_image_component = gr.Image(type="filepath", label="Input Image")
verification_state = gr.State(PUBLIC_ENABLED)
passcode_input = gr.Textbox(
label="Passcode",
type="password",
placeholder="Enter passcode to unlock generation",
visible=not PUBLIC_ENABLED,
)
verify_button = gr.Button("Verify Access", variant="secondary", visible=not PUBLIC_ENABLED)
verification_status = gr.Markdown(
"Public generation is enabled."
if PUBLIC_ENABLED
else "Insiders: Find the passcode in 🔒#hf-space-access-code on the [discord server](https://discord.gg/8wMr2mmaRF)."
)
prompt_input = gr.Textbox(
label="Prompt",
value=default_prompt_i2v,
lines=4,
placeholder="Describe the motion or scene transition you want to animate..."
)
duration_seconds_input = gr.Slider(
minimum=MIN_DURATION,
maximum=MAX_DURATION,
step=0.1,
value=4.0,
label="Duration (seconds)",
info=f"Clamped to {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=4)
seed_input = gr.Slider(label="Seed", minimum=0, maximum=MAX_SEED, step=1, value=696969696, interactive=True)
randomize_seed_checkbox = gr.Checkbox(label="Randomize seed", value=True, interactive=True)
steps_slider = gr.Slider(minimum=1, maximum=30, step=1, value=6, label="Inference Steps", interactive=True)
guidance_scale_input = gr.Slider(minimum=0.0, maximum=10.0, step=0.5, value=1, label="Guidance Scale (high noise stage)", interactive=True)
guidance_scale_2_input = gr.Slider(minimum=0.0, maximum=10.0, step=0.5, value=1, label="Guidance Scale 2 (low noise stage)", interactive=True)
generate_button = gr.Button(
"Generate Video",
variant="primary",
interactive=PUBLIC_ENABLED,
elem_id="generate-button",
)
with gr.Column(elem_id="preview-panel", scale=5):
gr.Markdown("### Render Preview")
video_output = gr.Video(
label="Generated Video",
autoplay=True,
sources=["upload"],
interactive=False,
elem_id="generated-video",
)
render_status = gr.Markdown("")
file_output = gr.File(label="Download Video")
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,
verification_state,
]
passcode_input.change(
fn=reset_verification,
outputs=[generate_button, verification_status, verification_state],
)
verify_button.click(
fn=verify_passcode,
inputs=[passcode_input],
outputs=[generate_button, verification_status, verification_state],
)
generate_button.click(
fn=generate_video,
inputs=ui_inputs,
outputs=[video_output, seed_input, render_status, file_output],
)
gr.Examples(
examples=[
[
"examples/example_7.jpg",
"a woman, jumpcut, after the transition, the scene changes to a parisian cobblestone back alley with a dumpster with a wooden pallet laid next to it, she is nude, she is kneeling with her eyes open and tongue out, looking at the camera, a nude man is standing on her left side, the man is holding his penis and pisssing on her face, piss soaks her hair and drips down her body",
6,
],
[
"examples/example_4.png",
"a woman, jumpcut, after the transition, she is nude, lying on her back with her legs back, she is pissing on her own face, the piss sprays all over her face",
6,
],
[
"examples/example_6.jpg",
"a woman, jumpcut, after the transition, she is nude, the woman is straddling a nude man, the man is lying down and holding the woman by her sides, the man's penis is inside the woman's pussy, she is moaning and bouncing on his penis",
6,
],
[
"examples/example_2.jpg",
"a woman, jumpcut, after the transition, the woman is nude, lying on a bed on her stomach with her hands gripping the bed in front of her, a nude man is standing in front of the woman with his penis inside her mouth, he is holding her head with his right hand with his left hand at his side, the man is pissing in the woman's mouth while his penis is inside her mouth, piss spills out of her mouth and drips onto the bed",
6,
],
[
"examples/example_8.jpg",
"a woman, jumpcut, after the transition, the woman is nude,looking at the camera, bent over a table, a nude man is standing in front her and holding her head with his left hand with his right hand at his side, his penis is inside the woman's mouth, another nude man is standing behind the woman and holding her waist with his penis inside the woman's ass",
6,
],
[
"examples/example_5.png",
"a man and a woman, jumpcut, after the transition, the woman is nude on a table on her hands and knees with her mouth open, the man is nude standing behind the woman holding her neck with both hands, the man's penis is inside the woman's ass, the man is looking down at her and smiling, he is thrusting his penis inside the woman's ass",
6,
],
[
"examples/example_1.png",
"a woman, jumpcut, after the transition, she is nude, she is sitting with her legs spread and looking at the camera, she is pissing on the floor",
6
],
],
inputs=[input_image_component, prompt_input, steps_slider],
outputs=[video_output, seed_input, render_status, file_output],
fn=generate_video,
cache_examples="lazy"
)
gr.Markdown("*Content Warning:* This research demonstration may generate material that is not suitable for all audiences. By using this application, you agree to assume full responsibility for compliance with the platform's terms of service and any applicable legal and regulatory guidelines.")
if __name__ == "__main__":
demo.queue().launch()