Spaces:
Sleeping
Sleeping
File size: 5,563 Bytes
f0533a5 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 |
{
"cells": [
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"import os\n",
"import json\n",
"import torch\n",
"import numpy as np\n",
"import PIL\n",
"from PIL import Image\n",
"from IPython.display import HTML\n",
"from pyramid_dit import PyramidDiTForVideoGeneration\n",
"from IPython.display import Image as ipython_image\n",
"from diffusers.utils import load_image, export_to_video, export_to_gif"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"variant='diffusion_transformer_768p' # For high resolution\n",
"# variant='diffusion_transformer_384p' # For low resolution\n",
"\n",
"model_path = \"/home/jinyang06/models/pyramid-flow\" # The downloaded checkpoint dir\n",
"model_dtype = 'bf16'\n",
"\n",
"device_id = 0\n",
"torch.cuda.set_device(device_id)\n",
"\n",
"model = PyramidDiTForVideoGeneration(\n",
" model_path,\n",
" model_dtype,\n",
" model_variant=variant,\n",
")\n",
"\n",
"model.vae.to(\"cuda\")\n",
"model.dit.to(\"cuda\")\n",
"model.text_encoder.to(\"cuda\")\n",
"\n",
"if model_dtype == \"bf16\":\n",
" torch_dtype = torch.bfloat16 \n",
"elif model_dtype == \"fp16\":\n",
" torch_dtype = torch.float16\n",
"else:\n",
" torch_dtype = torch.float32\n",
"\n",
"\n",
"def show_video(ori_path, rec_path, width=\"100%\"):\n",
" html = ''\n",
" if ori_path is not None:\n",
" html += f\"\"\"<video controls=\"\" name=\"media\" data-fullscreen-container=\"true\" width=\"{width}\">\n",
" <source src=\"{ori_path}\" type=\"video/mp4\">\n",
" </video>\n",
" \"\"\"\n",
" \n",
" html += f\"\"\"<video controls=\"\" name=\"media\" data-fullscreen-container=\"true\" width=\"{width}\">\n",
" <source src=\"{rec_path}\" type=\"video/mp4\">\n",
" </video>\n",
" \"\"\"\n",
" return HTML(html)"
]
},
{
"attachments": {},
"cell_type": "markdown",
"metadata": {},
"source": [
"#### Text-to-Video"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"prompt = \"A movie trailer featuring the adventures of the 30 year old space man wearing a red wool knitted motorcycle helmet, blue sky, salt desert, cinematic style, shot on 35mm film, vivid colors\"\n",
"\n",
"# used for 384p model variant\n",
"# width = 640\n",
"# height = 384\n",
"\n",
"# used for 768p model variant\n",
"width = 1280\n",
"height = 768\n",
"\n",
"temp = 16 # temp in [1, 31] <=> frame in [1, 241] <=> duration in [0, 10s]\n",
"\n",
"model.vae.enable_tiling()\n",
"\n",
"with torch.no_grad(), torch.cuda.amp.autocast(enabled=True if model_dtype != 'fp32' else False, dtype=torch_dtype):\n",
" frames = model.generate(\n",
" prompt=prompt,\n",
" num_inference_steps=[20, 20, 20],\n",
" video_num_inference_steps=[10, 10, 10],\n",
" height=height,\n",
" width=width,\n",
" temp=temp,\n",
" guidance_scale=9.0, # The guidance for the first frame\n",
" video_guidance_scale=5.0, # The guidance for the other video latent\n",
" output_type=\"pil\",\n",
" save_memory=True, # If you have enough GPU memory, set it to `False` to improve vae decoding speed\n",
" )\n",
"\n",
"export_to_video(frames, \"./text_to_video_sample.mp4\", fps=24)\n",
"show_video(None, \"./text_to_video_sample.mp4\", \"70%\")"
]
},
{
"attachments": {},
"cell_type": "markdown",
"metadata": {},
"source": [
"#### Image-to-Video"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"image_path = 'assets/the_great_wall.jpg'\n",
"image = Image.open(image_path).convert(\"RGB\")\n",
"\n",
"width = 1280\n",
"height = 768\n",
"temp = 16\n",
"\n",
"image = image.resize((width, height))\n",
"\n",
"display(image)\n",
"\n",
"prompt = \"FPV flying over the Great Wall\"\n",
"\n",
"with torch.no_grad(), torch.cuda.amp.autocast(enabled=True if model_dtype != 'fp32' else False, dtype=torch_dtype):\n",
" frames = model.generate_i2v(\n",
" prompt=prompt,\n",
" input_image=image,\n",
" num_inference_steps=[10, 10, 10],\n",
" temp=temp,\n",
" guidance_scale=7.0,\n",
" video_guidance_scale=4.0,\n",
" output_type=\"pil\",\n",
" save_memory=True, # If you have enough GPU memory, set it to `False` to improve vae decoding speed\n",
" )\n",
"\n",
"export_to_video(frames, \"./image_to_video_sample.mp4\", fps=24)\n",
"show_video(None, \"./image_to_video_sample.mp4\", \"70%\")"
]
}
],
"metadata": {
"kernelspec": {
"display_name": "Python 3",
"language": "python",
"name": "python3"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.8.10"
},
"orig_nbformat": 4
},
"nbformat": 4,
"nbformat_minor": 2
}
|