我们只需要将下面的"xxxx"，替换成所需的权重名，即可快速使用！

from ppdiffusers import *

pipe_text2img = StableDiffusionPipeline.from_pretrained("xxxx")
pipe_img2img = StableDiffusionImg2ImgPipeline.from_pretrained("xxxx")
pipe_inpaint_legacy = StableDiffusionInpaintPipelineLegacy.from_pretrained("xxxx")
pipe_mega = StableDiffusionMegaPipeline.from_pretrained("xxxx")

# pipe_mega.text2img() 等于 pipe_text2img()
# pipe_mega.img2img() 等于 pipe_img2img()
# pipe_mega.inpaint_legacy() 等于 pipe_inpaint_legacy()

import paddle
from ppdiffusers import StableDiffusionImg2ImgPipeline
from ppdiffusers.utils import load_image

pipe = StableDiffusionImg2ImgPipeline.from_pretrained("Linaqruf/anything-v3.0", safety_checker=None)

url = "https://paddlenlp.bj.bcebos.com/models/community/CompVis/data/image_Kurisu.png"
image = load_image(url).resize((512, 768))

# 设置随机种子，我们可以复现下面的结果！
paddle.seed(42)
prompt = "Kurisu Makise, looking at viewer, long hair, standing, 1girl, hair ornament, hair flower, cute, jacket, white flower, white dress"
negative_prompt = "lowres, bad anatomy, bad hands, text, error, missing fingers, extra digit, fewer digits, cropped, worst quality, low quality, normal quality, jpeg artifacts, signature, watermark, username, blurry"

image = pipe(prompt=prompt, negative_prompt=negative_prompt, image=image, strength=0.75, guidance_scale=7.5).images[0]
image.save("image_Kurisu_img2img.png")

import paddle
from ppdiffusers import StableDiffusionInpaintPipelineLegacy
from ppdiffusers.utils import load_image

# 可选模型权重
# CompVis/stable-diffusion-v1-4
# runwayml/stable-diffusion-v1-5
# stabilityai/stable-diffusion-2-base （原始策略 512x512）
# stabilityai/stable-diffusion-2 （v-objective 768x768）
# Linaqruf/anything-v3.0
# ......
img_url = "https://paddlenlp.bj.bcebos.com/models/community/CompVis/stable-diffusion-v1-4/overture-creations.png"
mask_url = "https://paddlenlp.bj.bcebos.com/models/community/CompVis/stable-diffusion-v1-4/overture-creations-mask.png"

image = load_image(img_url).resize((512, 512))
mask_image = load_image(mask_url).resize((512, 512))

pipe = StableDiffusionInpaintPipelineLegacy.from_pretrained("stabilityai/stable-diffusion-2-base", safety_checker=None)

# 设置随机种子，我们可以复现下面的结果！
paddle.seed(10245)
prompt = "a red cat sitting on a bench"
image = pipe(prompt=prompt, image=image, mask_image=mask_image, strength=0.75).images[0]

image.save("a_red_cat_legacy.png")

Tips: 下面的使用方法是新版本的代码，也是官方推荐的代码，注意必须配合 runwayml/stable-diffusion-inpainting 和 stabilityai/stable-diffusion-2-inpainting 才可正常使用。

import paddle
from ppdiffusers import StableDiffusionInpaintPipeline
from ppdiffusers.utils import load_image

# 可选模型权重
# runwayml/stable-diffusion-inpainting
# stabilityai/stable-diffusion-2-inpainting
img_url = "https://paddlenlp.bj.bcebos.com/models/community/CompVis/stable-diffusion-v1-4/overture-creations.png"
mask_url = "https://paddlenlp.bj.bcebos.com/models/community/CompVis/stable-diffusion-v1-4/overture-creations-mask.png"

image = load_image(img_url).resize((512, 512))
mask_image = load_image(mask_url).resize((512, 512))

pipe = StableDiffusionInpaintPipeline.from_pretrained("stabilityai/stable-diffusion-2-inpainting")

# 设置随机种子，我们可以复现下面的结果！
paddle.seed(1024)
prompt = "Face of a yellow cat, high resolution, sitting on a park bench"
image = pipe(prompt=prompt, image=image, mask_image=mask_image).images[0]

image.save("a_yellow_cat.png")

import paddle
from ppdiffusers import StableDiffusionUpscalePipeline
from ppdiffusers.utils import load_image

pipe = StableDiffusionUpscalePipeline.from_pretrained("stabilityai/stable-diffusion-x4-upscaler")

url = "https://paddlenlp.bj.bcebos.com/models/community/CompVis/data/low_res_cat.png"
# 我们人工将原始图片缩小成 128x128 分辨率，最终保存的图片会放大4倍！
low_res_img = load_image(url).resize((128, 128))

prompt = "a white cat"
image = pipe(prompt=prompt, image=low_res_img).images[0]

image.save("upscaled_white_cat.png")

import paddle
from ppdiffusers import LDMSuperResolutionPipeline
from ppdiffusers.utils import load_image

pipe = LDMSuperResolutionPipeline.from_pretrained("CompVis/ldm-super-resolution-4x-openimages")

url = "https://paddlenlp.bj.bcebos.com/models/community/CompVis/stable-diffusion-v1-4/overture-creations.png"

# 我们人工将原始图片缩小成 128x128 分辨率，最终保存的图片会放大4倍！
low_res_img = load_image(url).resize((128, 128))

image = pipe(image=low_res_img, num_inference_steps=100).images[0]

image.save("ldm-super-resolution-image.png")

注意：当前导出的vae encoder带有随机因素！

• CompVis/stable-diffusion-v1-4@fastdeploy
• runwayml/stable-diffusion-v1-5@fastdeploy
• runwayml/stable-diffusion-inpainting@fastdeploy
• stabilityai/stable-diffusion-2-base@fastdeploy
• stabilityai/stable-diffusion-2@fastdeploy
• stabilityai/stable-diffusion-2-inpainting@fastdeploy
• Linaqruf/anything-v3.0@fastdeploy
• hakurei/waifu-diffusion-v1-3@fastdeploy

import paddle
import fastdeploy as fd
from ppdiffusers import FastDeployStableDiffusionMegaPipeline
from ppdiffusers.utils import load_image

def create_runtime_option(device_id=0, backend="paddle", use_cuda_stream=True):
    option = fd.RuntimeOption()
    if backend == "paddle":
        option.use_paddle_backend()
    else:
        option.use_ort_backend()
    if device_id == -1:
        option.use_cpu()
    else:
        option.use_gpu(device_id)
        if use_cuda_stream:
            paddle_stream = paddle.device.cuda.current_stream(device_id).cuda_stream
            option.set_external_raw_stream(paddle_stream)
    return option

runtime_options = {
    "text_encoder": create_runtime_option(0, "paddle"),  # use gpu:0
    "vae_encoder": create_runtime_option(0, "paddle"),  # use gpu:0
    "vae_decoder": create_runtime_option(0, "paddle"),  # use gpu:0
    "unet": create_runtime_option(0, "paddle"),  # use gpu:0
}

fd_pipe = FastDeployStableDiffusionMegaPipeline.from_pretrained(
    "Linaqruf/anything-v3.0@fastdeploy", runtime_options=runtime_options
)

# text2img
prompt = "a portrait of shiba inu with a red cap growing on its head. intricate. lifelike. soft light. sony a 7 r iv 5 5 mm. cinematic post - processing "
image_text2img = fd_pipe.text2img(prompt=prompt, num_inference_steps=50).images[0]
image_text2img.save("image_text2img.png")

# img2img
url = "https://paddlenlp.bj.bcebos.com/models/community/CompVis/data/image_Kurisu.png"
image = load_image(url).resize((512, 512))
prompt = "Kurisu Makise, looking at viewer, long hair, standing, 1girl, hair ornament, hair flower, cute, jacket, white flower, white dress"
negative_prompt = "lowres, bad anatomy, bad hands, text, error, missing fingers, extra digit, fewer digits, cropped, worst quality, low quality, normal quality, jpeg artifacts, signature, watermark, username, blurry"

image_img2img = fd_pipe.img2img(
    prompt=prompt, negative_prompt=negative_prompt, image=image, strength=0.75, guidance_scale=7.5
).images[0]
image_img2img.save("image_img2img.png")

# inpaint_legacy
img_url = "https://paddlenlp.bj.bcebos.com/models/community/CompVis/stable-diffusion-v1-4/overture-creations.png"
mask_url = "https://paddlenlp.bj.bcebos.com/models/community/CompVis/stable-diffusion-v1-4/overture-creations-mask.png"
image = load_image(img_url).resize((512, 512))
mask_image = load_image(mask_url).resize((512, 512))
prompt = "a red cat sitting on a bench"

image_inpaint_legacy = fd_pipe.inpaint_legacy(
    prompt=prompt, image=image, mask_image=mask_image, strength=0.75, num_inference_steps=50
).images[0]
image_inpaint_legacy.save("image_inpaint_legacy.png")

text_to_image_generation-stable_diffusion

from ppdiffusers import StableDiffusionPipeline

# 加载模型和scheduler
pipe = StableDiffusionPipeline.from_pretrained("runwayml/stable-diffusion-v1-5")

# 执行pipeline进行推理
prompt = "a photo of an astronaut riding a horse on mars"
image = pipe(prompt).images[0]

# 保存图片
image.save("astronaut_rides_horse_sd.png")

text_to_image_generation-stable_diffusion_xl

import paddle
from ppdiffusers import StableDiffusionXLPipeline

pipe = StableDiffusionXLPipeline.from_pretrained(
     "stabilityai/stable-diffusion-xl-base-1.0",
     paddle_dtype=paddle.float16,
     variant="fp16"
)
prompt = "a photo of an astronaut riding a horse on mars"
generator = paddle.Generator().manual_seed(42)
image = pipe(prompt=prompt, generator=generator, num_inference_steps=50).images[0]
image.save('sdxl_text2image.png')

text_to_image_generation-sdxl_base_with_refiner

from ppdiffusers import DiffusionPipeline
import paddle

# load both base & refiner
base = DiffusionPipeline.from_pretrained(
    "stabilityai/stable-diffusion-xl-base-1.0",
    paddle_dtype=paddle.float16,
)
refiner = DiffusionPipeline.from_pretrained(
    "stabilityai/stable-diffusion-xl-refiner-1.0",
    text_encoder_2=base.text_encoder_2,
    vae=base.vae,
    paddle_dtype=paddle.float16,
    variant="fp16",
)

# Define how many steps and what % of steps to be run on each experts (80/20) here
n_steps = 40
high_noise_frac = 0.8

prompt = "A majestic lion jumping from a big stone at night"
prompt = "a photo of an astronaut riding a horse on mars"
generator = paddle.Generator().manual_seed(42)

# run both experts
image = base(
    prompt=prompt,
    output_type="latent",
    generator=generator,
).images

image = refiner(
    prompt=prompt,
    image=image,
    generator=generator,
).images[0]
image.save('text_to_image_generation-sdxl-base-with-refiner-result.png')

text_to_image_generation-kandinsky2_2

from ppdiffusers import KandinskyV22Pipeline, KandinskyV22PriorPipeline

pipe_prior = KandinskyV22PriorPipeline.from_pretrained("kandinsky-community/kandinsky-2-2-prior")
prompt = "red cat, 4k photo"
out = pipe_prior(prompt)
image_emb = out.image_embeds
zero_image_emb = out.negative_image_embeds
pipe = KandinskyV22Pipeline.from_pretrained("kandinsky-community/kandinsky-2-2-decoder")
image = pipe(
    image_embeds=image_emb,
    negative_image_embeds=zero_image_emb,
    height=768,
    width=768,
    num_inference_steps=50,
).images
image[0].save("text_to_image_generation-kandinsky2_2-result-cat.png")

text_to_image_generation-unidiffuser

import paddle
from paddlenlp.trainer import set_seed

from ppdiffusers import UniDiffuserPipeline

model_id_or_path = "thu-ml/unidiffuser-v1"
pipe = UniDiffuserPipeline.from_pretrained(model_id_or_path, paddle_dtype=paddle.float16)
set_seed(42)

# Text variation can be performed with a text-to-image generation followed by a image-to-text generation:
# 1. Text-to-image generation
prompt = "an elephant under the sea"
sample = pipe(prompt=prompt, num_inference_steps=20, guidance_scale=8.0)
t2i_image = sample.images[0]
t2i_image.save("t2i_image.png")

text_to_image_generation-deepfloyd_if

import paddle

from ppdiffusers import DiffusionPipeline, IFPipeline, IFSuperResolutionPipeline
from ppdiffusers.utils import pd_to_pil

# Stage 1: generate images
pipe = IFPipeline.from_pretrained("DeepFloyd/IF-I-XL-v1.0", variant="fp16", paddle_dtype=paddle.float16)
pipe.enable_xformers_memory_efficient_attention()
prompt = 'a photo of a kangaroo wearing an orange hoodie and blue sunglasses standing in front of the eiffel tower holding a sign that says "very deep learning"'
prompt_embeds, negative_embeds = pipe.encode_prompt(prompt)
image = pipe(
    prompt_embeds=prompt_embeds,
    negative_prompt_embeds=negative_embeds,
    output_type="pd",
).images

# save intermediate image
pil_image = pd_to_pil(image)
pil_image[0].save("text_to_image_generation-deepfloyd_if-result-if_stage_I.png")
# save gpu memory
pipe.to(paddle_device="cpu")

# Stage 2: super resolution stage1
super_res_1_pipe = IFSuperResolutionPipeline.from_pretrained(
    "DeepFloyd/IF-II-L-v1.0", text_encoder=None, variant="fp16", paddle_dtype=paddle.float16
)
super_res_1_pipe.enable_xformers_memory_efficient_attention()

image = super_res_1_pipe(
    image=image,
    prompt_embeds=prompt_embeds,
    negative_prompt_embeds=negative_embeds,
    output_type="pd",
).images
# save intermediate image
pil_image = pd_to_pil(image)
pil_image[0].save("text_to_image_generation-deepfloyd_if-result-if_stage_II.png")
# save gpu memory
super_res_1_pipe.to(paddle_device="cpu")

text_guided_image_upscaling-stable_diffusion_2

from ppdiffusers import StableDiffusionUpscalePipeline
from ppdiffusers.utils import load_image

pipe = StableDiffusionUpscalePipeline.from_pretrained("stabilityai/stable-diffusion-x4-upscaler")

url = "https://paddlenlp.bj.bcebos.com/models/community/CompVis/data/low_res_cat.png"
low_res_img = load_image(url).resize((128, 128))

prompt = "a white cat"
upscaled_image = pipe(prompt=prompt, image=low_res_img).images[0]
upscaled_image.save("upsampled_cat_sd2.png")

原图像 生成图像

text_guided_image_inpainting-stable_diffusion_2

import paddle

from ppdiffusers import PaintByExamplePipeline
from ppdiffusers.utils import load_image

img_url = "https://paddlenlp.bj.bcebos.com/models/community/Fantasy-Studio/data/image_example_1.png"
mask_url = "https://paddlenlp.bj.bcebos.com/models/community/Fantasy-Studio/data/mask_example_1.png"
example_url = "https://paddlenlp.bj.bcebos.com/models/community/Fantasy-Studio/data/reference_example_1.jpeg"

init_image = load_image(img_url).resize((512, 512))
mask_image = load_image(mask_url).resize((512, 512))
example_image = load_image(example_url).resize((512, 512))

pipe = PaintByExamplePipeline.from_pretrained("Fantasy-Studio/Paint-by-Example")

# 使用fp16加快生成速度
with paddle.amp.auto_cast(True):
    image = pipe(image=init_image, mask_image=mask_image, example_image=example_image).images[0]
image.save("image_guided_image_inpainting-paint_by_example-result.png")

原图像

掩码图像

参考图像 生成图像

text_guided_image_inpainting-kandinsky2_2

import numpy as np
import paddle

from ppdiffusers import KandinskyV22InpaintPipeline, KandinskyV22PriorPipeline
from ppdiffusers.utils import load_image

pipe_prior = KandinskyV22PriorPipeline.from_pretrained(
    "kandinsky-community/kandinsky-2-2-prior", paddle_dtype=paddle.float16
)
prompt = "a hat"
image_emb, zero_image_emb = pipe_prior(prompt, return_dict=False)
pipe = KandinskyV22InpaintPipeline.from_pretrained(
    "kandinsky-community/kandinsky-2-2-decoder-inpaint", paddle_dtype=paddle.float16
)
init_image = load_image(
    "https://huggingface.co/datasets/hf-internal-testing/diffusers-images/resolve/main/kandinsky/cat.png"
)
mask = np.zeros((768, 768), dtype=np.float32)
mask[:250, 250:-250] = 1
out = pipe(
    image=init_image,
    mask_image=mask,
    image_embeds=image_emb,
    negative_image_embeds=zero_image_emb,
    height=768,
    width=768,
    num_inference_steps=50,
)
image = out.images[0]
image.save("text_guided_image_inpainting-kandinsky2_2-result-cat_with_hat.png")

原图像 生成图像

image_to_image_text_guided_generation-stable_diffusion

import paddle

from ppdiffusers import StableDiffusionImg2ImgPipeline
from ppdiffusers.utils import load_image

# 加载pipeline
pipe = StableDiffusionImg2ImgPipeline.from_pretrained("runwayml/stable-diffusion-v1-5")

# 下载初始图片
url = "https://paddlenlp.bj.bcebos.com/models/community/CompVis/stable-diffusion-v1-4/sketch-mountains-input.png"

init_image = load_image(url).resize((768, 512))

prompt = "A fantasy landscape, trending on artstation"
# 使用fp16加快生成速度
with paddle.amp.auto_cast(True):
    image = pipe(prompt=prompt, image=init_image, strength=0.75, guidance_scale=7.5).images[0]

image.save("fantasy_landscape.png")

原图像 生成图像

image_to_image_text_guided_generation-stable_diffusion_xl

import paddle
from ppdiffusers import StableDiffusionXLImg2ImgPipeline
from ppdiffusers.utils import load_image

pipe = StableDiffusionXLImg2ImgPipeline.from_pretrained(
    "stabilityai/stable-diffusion-xl-refiner-1.0",
    paddle_dtype=paddle.float16,
    # from_hf_hub=True,
    # from_diffusers=True,
    variant="fp16"
)
url = "https://paddlenlp.bj.bcebos.com/models/community/westfish/develop-0-19-3/000000009.png"
init_image = load_image(url).convert("RGB")
prompt = "a photo of an astronaut riding a horse on mars"
image = pipe(prompt, image=init_image).images[0]
image.save('sdxl_image2image.png')

原图像 生成图像

image_to_image_text_guided_generation-kandinsky2_2

import paddle

from ppdiffusers import KandinskyV22Img2ImgPipeline, KandinskyV22PriorPipeline
from ppdiffusers.utils import load_image

pipe_prior = KandinskyV22PriorPipeline.from_pretrained(
    "kandinsky-community/kandinsky-2-2-prior", paddle_dtype=paddle.float16
)
prompt = "A red cartoon frog, 4k"
image_emb, zero_image_emb = pipe_prior(prompt, return_dict=False)
pipe = KandinskyV22Img2ImgPipeline.from_pretrained(
    "kandinsky-community/kandinsky-2-2-decoder", paddle_dtype=paddle.float16
)

init_image = load_image(
    "https://hf-mirror.com/datasets/hf-internal-testing/diffusers-images/resolve/main/kandinsky/frog.png"
)
image = pipe(
    image=init_image,
    image_embeds=image_emb,
    negative_image_embeds=zero_image_emb,
    height=768,
    width=768,
    num_inference_steps=100,
    strength=0.2,
).images
image[0].save("image_to_image_text_guided_generation-kandinsky2_2-result-red_frog.png")

原图像 生成图像

dual_text_and_image_guided_generation-versatile_diffusion

from ppdiffusers import VersatileDiffusionDualGuidedPipeline
from ppdiffusers.utils import load_image

url = "https://paddlenlp.bj.bcebos.com/models/community/CompVis/data/benz.jpg"
image = load_image(url)
text = "a red car in the sun"

pipe = VersatileDiffusionDualGuidedPipeline.from_pretrained("shi-labs/versatile-diffusion")
pipe.remove_unused_weights()

text_to_image_strength = 0.75
image = pipe(prompt=text, image=image, text_to_image_strength=text_to_image_strength).images[0]
image.save("versatile-diffusion-red_car.png")

原图像 生成图像

text_to_video_generation-lvdm

import paddle

from ppdiffusers import LVDMTextToVideoPipeline

# 加载模型和scheduler
pipe = LVDMTextToVideoPipeline.from_pretrained("westfish/lvdm_text2video_orig_webvid_2m")

# 执行pipeline进行推理
seed = 2013
generator = paddle.Generator().manual_seed(seed)
samples = pipe(
    prompt="cutting in kitchen",
    num_frames=16,
    height=256,
    width=256,
    num_inference_steps=50,
    generator=generator,
    guidance_scale=15,
    eta=1,
    save_dir=".",
    save_name="text_to_video_generation-lvdm-result-ddim_lvdm_text_to_video_ucf",
    encoder_type="2d",
    scale_factor=0.18215,
    shift_factor=0,
)

text_to_video_generation-synth

import imageio

from ppdiffusers import DPMSolverMultistepScheduler, TextToVideoSDPipeline

pipe = TextToVideoSDPipeline.from_pretrained("damo-vilab/text-to-video-ms-1.7b")
pipe.scheduler = DPMSolverMultistepScheduler.from_config(pipe.scheduler.config)

prompt = "An astronaut riding a horse."
video_frames = pipe(prompt, num_inference_steps=25).frames
imageio.mimsave("text_to_video_generation-synth-result-astronaut_riding_a_horse.mp4", video_frames, fps=8)

text_to_video_generation-synth with zeroscope_v2_XL

import imageio

from ppdiffusers import DPMSolverMultistepScheduler, TextToVideoSDPipeline

# from ppdiffusers.utils import export_to_video

pipe = TextToVideoSDPipeline.from_pretrained("cerspense/zeroscope_v2_XL")
pipe.scheduler = DPMSolverMultistepScheduler.from_config(pipe.scheduler.config)

prompt = "An astronaut riding a horse."
video_frames = pipe(prompt, num_inference_steps=50, height=320, width=576, num_frames=24).frames
imageio.mimsave("text_to_video_generation-synth-result-astronaut_riding_a_horse.mp4", video_frames, fps=8)

text_to_video_generation-zero

import imageio

# pip install imageio[ffmpeg]
import paddle

from ppdiffusers import TextToVideoZeroPipeline

model_id = "runwayml/stable-diffusion-v1-5"
pipe = TextToVideoZeroPipeline.from_pretrained(model_id, paddle_dtype=paddle.float16)

prompt = "A panda is playing guitar on times square"
result = pipe(prompt=prompt).images
result = [(r * 255).astype("uint8") for r in result]
imageio.mimsave("text_to_video_generation-zero-result-panda.mp4", result, fps=4)

text_to_audio_generation-audio_ldm

import paddle
import scipy

from ppdiffusers import AudioLDMPipeline

pipe = AudioLDMPipeline.from_pretrained("cvssp/audioldm", paddle_dtype=paddle.float16)

prompt = "Techno music with a strong, upbeat tempo and high melodic riffs"
audio = pipe(prompt, num_inference_steps=10, audio_length_in_s=5.0).audios[0]

output_path = "text_to_audio_generation-audio_ldm-techno.wav"
# save the audio sample as a .wav file
scipy.io.wavfile.write(output_path, rate=16000, data=audio)

  <a rel="nofollow" href="https://paddlenlp.bj.bcebos.com/models/community/westfish/develop_ppdiffusers_data/techno.wav">
        <img width="200 style=" src="https://user-images.githubusercontent.com/20476674/209344877-edbf1c24-f08d-4e3b-88a4-a27e1fd0a858.png" align="center"></a><br>

unconditional_image_generation-latent_diffusion_uncond

from ppdiffusers import LDMPipeline

# 加载模型和scheduler
pipe = LDMPipeline.from_pretrained("CompVis/ldm-celebahq-256")

# 执行pipeline进行推理
image = pipe(num_inference_steps=200).images[0]

# 保存图片
image.save("ldm_generated_image.png")

super_resolution-latent_diffusion

import paddle

from ppdiffusers import LDMSuperResolutionPipeline
from ppdiffusers.utils import load_image

# 加载pipeline
pipe = LDMSuperResolutionPipeline.from_pretrained("CompVis/ldm-super-resolution-4x-openimages")

# 下载初始图片
url = "https://paddlenlp.bj.bcebos.com/models/community/CompVis/stable-diffusion-v1-4/overture-creations.png"

init_image = load_image(url).resize((128, 128))
init_image.save("original-image.png")

# 使用fp16加快生成速度
with paddle.amp.auto_cast(True):
    image = pipe(init_image, num_inference_steps=100, eta=1).images[0]

image.save("super-resolution-image.png")

原图像 生成图像

image_inpainting-repaint

from ppdiffusers import RePaintPipeline, RePaintScheduler
from ppdiffusers.utils import load_image

img_url = "https://paddlenlp.bj.bcebos.com/models/community/CompVis/data/celeba_hq_256.png"
mask_url = "https://paddlenlp.bj.bcebos.com/models/community/CompVis/data/mask_256.png"

# Load the original image and the mask as PIL images
original_image = load_image(img_url).resize((256, 256))
mask_image = load_image(mask_url).resize((256, 256))

scheduler = RePaintScheduler.from_pretrained("google/ddpm-ema-celebahq-256", subfolder="scheduler")
pipe = RePaintPipeline.from_pretrained("google/ddpm-ema-celebahq-256", scheduler=scheduler)

output = pipe(
    original_image=original_image,
    mask_image=mask_image,
    num_inference_steps=250,
    eta=0.0,
    jump_length=10,
    jump_n_sample=10,
)
inpainted_image = output.images[0]

inpainted_image.save("repaint-image.png")

原图像 mask图像 生成图像

image_variation-versatile_diffusion

from ppdiffusers import VersatileDiffusionImageVariationPipeline
from ppdiffusers.utils import load_image

url = "https://paddlenlp.bj.bcebos.com/models/community/CompVis/data/benz.jpg"
image = load_image(url)

pipe = VersatileDiffusionImageVariationPipeline.from_pretrained("shi-labs/versatile-diffusion")

image = pipe(image).images[0]
image.save("versatile-diffusion-car_variation.png")

原图像 生成图像

unconditional_audio_generation-audio_diffusion

from scipy.io.wavfile import write
from ppdiffusers import AudioDiffusionPipeline
import paddle

# 加载模型和scheduler
pipe = AudioDiffusionPipeline.from_pretrained("teticio/audio-diffusion-ddim-256")
pipe.set_progress_bar_config(disable=None)
generator = paddle.Generator().manual_seed(42)

output = pipe(generator=generator)
audio = output.audios[0]
image = output.images[0]

# 保存音频到本地
for i, audio in enumerate(audio):
    write(f"audio_diffusion_test{i}.wav", pipe.mel.config.sample_rate, audio.transpose())

# 保存图片
image.save("audio_diffusion_test.png")

  <a rel="nofollow" href="https://paddlenlp.bj.bcebos.com/models/community/teticio/data/audio_diffusion_test0.wav">
        <img width="200 style=" src="https://user-images.githubusercontent.com/20476674/209344877-edbf1c24-f08d-4e3b-88a4-a27e1fd0a858.png" align="center"></a><br>

unconditional_audio_generation-spectrogram_diffusion

import paddle
import scipy

from ppdiffusers import MidiProcessor, SpectrogramDiffusionPipeline
from ppdiffusers.utils.download_utils import ppdiffusers_url_download

# Download MIDI from: wget https://paddlenlp.bj.bcebos.com/models/community/junnyu/develop/beethoven_hammerklavier_2.mid
mid_file_path = ppdiffusers_url_download(
    "https://paddlenlp.bj.bcebos.com/models/community/junnyu/develop/beethoven_hammerklavier_2.mid", cache_dir="."
)
pipe = SpectrogramDiffusionPipeline.from_pretrained("google/music-spectrogram-diffusion", paddle_dtype=paddle.float16)
processor = MidiProcessor()
output = pipe(processor(mid_file_path))
audio = output.audios[0]

output_path = "unconditional_audio_generation-spectrogram_diffusion-result-beethoven_hammerklavier_2.wav"
# save the audio sample as a .wav file
scipy.io.wavfile.write(output_path, rate=16000, data=audio)

  <a rel="nofollow" href="https://paddlenlp.bj.bcebos.com/models/community/westfish/develop_ppdiffusers_data/beethoven_hammerklavier_2.wav">
        <img width="200 style=" src="https://user-images.githubusercontent.com/20476674/209344877-edbf1c24-f08d-4e3b-88a4-a27e1fd0a858.png" align="center"></a><br>