sd-webui-deforum/Deforum_Stable_Diffusion.ipynb

729 lines
33 KiB
Plaintext

{
"cells": [
{
"cell_type": "markdown",
"metadata": {
"id": "c442uQJ_gUgy"
},
"source": [
"# **Deforum Stable Diffusion v0.6**\n",
"[Stable Diffusion](https://github.com/CompVis/stable-diffusion) by Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser, Björn Ommer and the [Stability.ai](https://stability.ai/) Team. [K Diffusion](https://github.com/crowsonkb/k-diffusion) by [Katherine Crowson](https://twitter.com/RiversHaveWings). You need to get the ckpt file and put it on your Google Drive first to use this. It can be downloaded from [HuggingFace](https://huggingface.co/CompVis/stable-diffusion).\n",
"\n",
"Notebook by [deforum](https://discord.gg/upmXXsrwZc)"
]
},
{
"cell_type": "code",
"metadata": {
"cellView": "form",
"id": "2g-f7cQmf2Nt"
},
"source": [
"#@markdown **NVIDIA GPU**\n",
"import subprocess, os, sys\n",
"sub_p_res = subprocess.run(['nvidia-smi', '--query-gpu=name,memory.total,memory.free', '--format=csv,noheader'], stdout=subprocess.PIPE).stdout.decode('utf-8')\n",
"print(f\"{sub_p_res[:-1]}\")"
],
"outputs": [],
"execution_count": null
},
{
"cell_type": "markdown",
"metadata": {
"id": "T4knibRpAQ06"
},
"source": [
"# Setup"
]
},
{
"cell_type": "code",
"metadata": {
"cellView": "form",
"id": "TxIOPT0G5Lx1"
},
"source": [
"#@markdown **Model and Output Paths**\n",
"def get_model_output_paths():\n",
"\n",
" models_path = \"models\" #@param {type:\"string\"}\n",
" output_path = \"output\" #@param {type:\"string\"}\n",
"\n",
" #@markdown **Google Drive Path Variables (Optional)**\n",
" mount_google_drive = True #@param {type:\"boolean\"}\n",
" force_remount = False\n",
"\n",
" try:\n",
" ipy = get_ipython()\n",
" except:\n",
" ipy = 'could not get_ipython'\n",
"\n",
" if 'google.colab' in str(ipy):\n",
" if mount_google_drive:\n",
" from google.colab import drive # type: ignore\n",
" try:\n",
" drive_path = \"/content/drive\"\n",
" drive.mount(drive_path,force_remount=force_remount)\n",
" models_path_gdrive = \"/content/drive/MyDrive/AI/models\" #@param {type:\"string\"}\n",
" output_path_gdrive = \"/content/drive/MyDrive/AI/StableDiffusion\" #@param {type:\"string\"}\n",
" models_path = models_path_gdrive\n",
" output_path = output_path_gdrive\n",
" except:\n",
" print(\"..error mounting drive or with drive path variables\")\n",
" print(\"..reverting to default path variables\")\n",
"\n",
" models_path = os.path.abspath(models_path)\n",
" output_path = os.path.abspath(output_path)\n",
" os.makedirs(models_path, exist_ok=True)\n",
" os.makedirs(output_path, exist_ok=True)\n",
"\n",
" print(f\"models_path: {models_path}\")\n",
" print(f\"output_path: {output_path}\")\n",
"\n",
" return models_path, output_path\n",
"\n",
"models_path, output_path = get_model_output_paths()\n",
"\n",
"def setup_environment():\n",
"\n",
" print_subprocess = False\n",
"\n",
" try:\n",
" ipy = get_ipython()\n",
" except:\n",
" ipy = 'could not get_ipython'\n",
"\n",
" if 'google.colab' in str(ipy):\n",
" import subprocess, time\n",
" print(\"Setting up environment...\")\n",
" start_time = time.time()\n",
" all_process = [\n",
" ['pip', 'install', 'torch==1.12.1+cu113', 'torchvision==0.13.1+cu113', '--extra-index-url', 'https://download.pytorch.org/whl/cu113'],\n",
" ['pip', 'install', 'omegaconf==2.2.3', 'einops==0.4.1', 'pytorch-lightning==1.7.4', 'torchmetrics==0.9.3', 'torchtext==0.13.1', 'transformers==4.21.2', 'kornia==0.6.7'],\n",
" ['git', 'clone', '-b', 'local', 'https://github.com/deforum/stable-diffusion'],\n",
" ['pip', 'install', 'accelerate', 'ftfy', 'jsonmerge', 'matplotlib', 'resize-right', 'timm', 'torchdiffeq'],\n",
" ]\n",
" for process in all_process:\n",
" running = subprocess.run(process,stdout=subprocess.PIPE).stdout.decode('utf-8')\n",
" if print_subprocess:\n",
" print(running)\n",
"\n",
" with open('stable-diffusion/src/k_diffusion/__init__.py', 'w') as f:\n",
" f.write('')\n",
"\n",
" sys.path.extend([\n",
" 'stable-diffusion/',\n",
" 'stable-diffusion/src',\n",
" ])\n",
"\n",
" end_time = time.time()\n",
" print(f\"Environment set up in {end_time-start_time:.0f} seconds\")\n",
" \n",
" else:\n",
"\n",
" sys.path.extend([\n",
" 'src'\n",
" ])\n",
"\n",
" return\n",
"\n",
"setup_environment()\n",
"\n",
"# import\n",
"import torch\n",
"import gc\n",
"import time\n",
"import random\n",
"from types import SimpleNamespace\n",
"\n",
"from helpers.save_images import get_output_folder\n",
"from helpers.settings import load_args\n",
"from helpers.render import render_animation, render_input_video, render_image_batch, render_interpolation\n",
"\n",
"#@markdown **Select and Load Model**\n",
"\n",
"def load_model():\n",
"\n",
" import requests\n",
" import torch\n",
" from ldm.util import instantiate_from_config\n",
" from omegaconf import OmegaConf\n",
" from transformers import logging\n",
" logging.set_verbosity_error()\n",
"\n",
" model_config = \"v1-inference.yaml\" #@param [\"custom\",\"v1-inference.yaml\"]\n",
" model_checkpoint = \"sd-v1-4.ckpt\" #@param [\"custom\",\"sd-v1-4-full-ema.ckpt\",\"sd-v1-4.ckpt\",\"sd-v1-3-full-ema.ckpt\",\"sd-v1-3.ckpt\",\"sd-v1-2-full-ema.ckpt\",\"sd-v1-2.ckpt\",\"sd-v1-1-full-ema.ckpt\",\"sd-v1-1.ckpt\", \"robo-diffusion-v1.ckpt\",\"wd-v1-3-float16.ckpt\"]\n",
"\n",
" custom_config_path = \"\" #@param {type:\"string\"}\n",
" custom_checkpoint_path = \"\" #@param {type:\"string\"}\n",
"\n",
" load_on_run_all = True\n",
" half_precision = True\n",
" check_sha256 = True\n",
"\n",
" try:\n",
" ipy = get_ipython()\n",
" except:\n",
" ipy = 'could not get_ipython'\n",
"\n",
" if 'google.colab' in str(ipy):\n",
" path_extend = \"stable-diffusion\"\n",
" else:\n",
" path_extend = \"\"\n",
"\n",
" model_map = {\n",
" \"sd-v1-4-full-ema.ckpt\": {\n",
" 'sha256': '14749efc0ae8ef0329391ad4436feb781b402f4fece4883c7ad8d10556d8a36a',\n",
" 'url': 'https://huggingface.co/CompVis/stable-diffusion-v-1-2-original/blob/main/sd-v1-4-full-ema.ckpt',\n",
" 'requires_login': True,\n",
" },\n",
" \"sd-v1-4.ckpt\": {\n",
" 'sha256': 'fe4efff1e174c627256e44ec2991ba279b3816e364b49f9be2abc0b3ff3f8556',\n",
" 'url': 'https://huggingface.co/CompVis/stable-diffusion-v-1-4-original/resolve/main/sd-v1-4.ckpt',\n",
" 'requires_login': True,\n",
" },\n",
" \"sd-v1-3-full-ema.ckpt\": {\n",
" 'sha256': '54632c6e8a36eecae65e36cb0595fab314e1a1545a65209f24fde221a8d4b2ca',\n",
" 'url': 'https://huggingface.co/CompVis/stable-diffusion-v-1-3-original/blob/main/sd-v1-3-full-ema.ckpt',\n",
" 'requires_login': True,\n",
" },\n",
" \"sd-v1-3.ckpt\": {\n",
" 'sha256': '2cff93af4dcc07c3e03110205988ff98481e86539c51a8098d4f2236e41f7f2f',\n",
" 'url': 'https://huggingface.co/CompVis/stable-diffusion-v-1-3-original/resolve/main/sd-v1-3.ckpt',\n",
" 'requires_login': True,\n",
" },\n",
" \"sd-v1-2-full-ema.ckpt\": {\n",
" 'sha256': 'bc5086a904d7b9d13d2a7bccf38f089824755be7261c7399d92e555e1e9ac69a',\n",
" 'url': 'https://huggingface.co/CompVis/stable-diffusion-v-1-2-original/blob/main/sd-v1-2-full-ema.ckpt',\n",
" 'requires_login': True,\n",
" },\n",
" \"sd-v1-2.ckpt\": {\n",
" 'sha256': '3b87d30facd5bafca1cbed71cfb86648aad75d1c264663c0cc78c7aea8daec0d',\n",
" 'url': 'https://huggingface.co/CompVis/stable-diffusion-v-1-2-original/resolve/main/sd-v1-2.ckpt',\n",
" 'requires_login': True,\n",
" },\n",
" \"sd-v1-1-full-ema.ckpt\": {\n",
" 'sha256': 'efdeb5dc418a025d9a8cc0a8617e106c69044bc2925abecc8a254b2910d69829',\n",
" 'url':'https://huggingface.co/CompVis/stable-diffusion-v-1-1-original/resolve/main/sd-v1-1-full-ema.ckpt',\n",
" 'requires_login': True,\n",
" },\n",
" \"sd-v1-1.ckpt\": {\n",
" 'sha256': '86cd1d3ccb044d7ba8db743d717c9bac603c4043508ad2571383f954390f3cea',\n",
" 'url': 'https://huggingface.co/CompVis/stable-diffusion-v-1-1-original/resolve/main/sd-v1-1.ckpt',\n",
" 'requires_login': True,\n",
" },\n",
" \"robo-diffusion-v1.ckpt\": {\n",
" 'sha256': '244dbe0dcb55c761bde9c2ac0e9b46cc9705ebfe5f1f3a7cc46251573ea14e16',\n",
" 'url': 'https://huggingface.co/nousr/robo-diffusion/resolve/main/models/robo-diffusion-v1.ckpt',\n",
" 'requires_login': False,\n",
" },\n",
" \"wd-v1-3-float16.ckpt\": {\n",
" 'sha256': '4afab9126057859b34d13d6207d90221d0b017b7580469ea70cee37757a29edd',\n",
" 'url': 'https://huggingface.co/hakurei/waifu-diffusion-v1-3/resolve/main/wd-v1-3-float16.ckpt',\n",
" 'requires_login': False,\n",
" },\n",
" }\n",
"\n",
" # config path\n",
" ckpt_config_path = custom_config_path if model_config == \"custom\" else os.path.join(models_path, model_config)\n",
" if os.path.exists(ckpt_config_path):\n",
" print(f\"{ckpt_config_path} exists\")\n",
" else:\n",
" ckpt_config_path = os.path.join(path_extend,\"configs\",\"v1-inference.yaml\")\n",
" \n",
" ckpt_config_path = os.path.abspath(ckpt_config_path)\n",
"\n",
" # checkpoint path or download\n",
" ckpt_path = custom_checkpoint_path if model_checkpoint == \"custom\" else os.path.join(models_path, model_checkpoint)\n",
" ckpt_valid = True\n",
" if os.path.exists(ckpt_path):\n",
" pass\n",
" elif 'url' in model_map[model_checkpoint]:\n",
" url = model_map[model_checkpoint]['url']\n",
"\n",
" # CLI dialogue to authenticate download\n",
" if model_map[model_checkpoint]['requires_login']:\n",
" print(\"This model requires an authentication token\")\n",
" print(\"Please ensure you have accepted the terms of service before continuing.\")\n",
"\n",
" username = input(\"[What is your huggingface username?]: \")\n",
" token = input(\"[What is your huggingface token?]: \")\n",
"\n",
" _, path = url.split(\"https://\")\n",
"\n",
" url = f\"https://{username}:{token}@{path}\"\n",
"\n",
" # contact server for model\n",
" print(f\"..attempting to download {model_checkpoint}...this may take a while\")\n",
" ckpt_request = requests.get(url)\n",
" request_status = ckpt_request.status_code\n",
"\n",
" # inform user of errors\n",
" if request_status == 403:\n",
" raise ConnectionRefusedError(\"You have not accepted the license for this model.\")\n",
" elif request_status == 404:\n",
" raise ConnectionError(\"Could not make contact with server\")\n",
" elif request_status != 200:\n",
" raise ConnectionError(f\"Some other error has ocurred - response code: {request_status}\")\n",
"\n",
" # write to model path\n",
" with open(os.path.join(models_path, model_checkpoint), 'wb') as model_file:\n",
" model_file.write(ckpt_request.content)\n",
" else:\n",
" print(f\"Please download model checkpoint and place in {os.path.join(models_path, model_checkpoint)}\")\n",
" ckpt_valid = False\n",
" \n",
" print(f\"config_path: {ckpt_config_path}\")\n",
" print(f\"ckpt_path: {ckpt_path}\")\n",
"\n",
" if check_sha256 and model_checkpoint != \"custom\" and ckpt_valid:\n",
" import hashlib\n",
" print(\"..checking sha256\")\n",
" with open(ckpt_path, \"rb\") as f:\n",
" bytes = f.read() \n",
" hash = hashlib.sha256(bytes).hexdigest()\n",
" del bytes\n",
" if model_map[model_checkpoint][\"sha256\"] == hash:\n",
" print(\"..hash is correct\")\n",
" else:\n",
" print(\"..hash in not correct\")\n",
" ckpt_valid = False\n",
"\n",
" def load_model_from_config(config, ckpt, verbose=False, device='cuda', half_precision=True,print_flag=False):\n",
" map_location = \"cuda\" # [\"cpu\", \"cuda\"]\n",
" print(f\"..loading model\")\n",
" pl_sd = torch.load(ckpt, map_location=map_location)\n",
" if \"global_step\" in pl_sd:\n",
" if print_flag:\n",
" print(f\"Global Step: {pl_sd['global_step']}\")\n",
" sd = pl_sd[\"state_dict\"]\n",
" model = instantiate_from_config(config.model)\n",
" m, u = model.load_state_dict(sd, strict=False)\n",
" if print_flag:\n",
" if len(m) > 0 and verbose:\n",
" print(\"missing keys:\")\n",
" print(m)\n",
" if len(u) > 0 and verbose:\n",
" print(\"unexpected keys:\")\n",
" print(u)\n",
"\n",
" if half_precision:\n",
" model = model.half().to(device)\n",
" else:\n",
" model = model.to(device)\n",
" model.eval()\n",
" return model\n",
"\n",
" if load_on_run_all and ckpt_valid:\n",
" local_config = OmegaConf.load(f\"{ckpt_config_path}\")\n",
" model = load_model_from_config(local_config, f\"{ckpt_path}\", half_precision=half_precision)\n",
" device = torch.device(\"cuda\") if torch.cuda.is_available() else torch.device(\"cpu\")\n",
" model = model.to(device)\n",
"\n",
" return model, device\n",
"\n",
"model, device = load_model()"
],
"outputs": [],
"execution_count": null
},
{
"cell_type": "markdown",
"metadata": {
"id": "ov3r4RD1tzsT"
},
"source": [
"# Settings"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "0j7rgxvLvfay"
},
"source": [
"### Animation Settings"
]
},
{
"cell_type": "code",
"metadata": {
"cellView": "form",
"id": "8HJN2TE3vh-J"
},
"source": [
"def DeforumAnimArgs():\n",
"\n",
" #@markdown ####**Animation:**\n",
" animation_mode = 'None' #@param ['None', '2D', '3D', 'Video Input', 'Interpolation'] {type:'string'}\n",
" max_frames = 1000 #@param {type:\"number\"}\n",
" border = 'replicate' #@param ['wrap', 'replicate'] {type:'string'}\n",
"\n",
" #@markdown ####**Motion Parameters:**\n",
" angle = \"0:(0)\"#@param {type:\"string\"}\n",
" zoom = \"0:(1.04)\"#@param {type:\"string\"}\n",
" translation_x = \"0:(10*sin(2*3.14*t/10))\"#@param {type:\"string\"}\n",
" translation_y = \"0:(0)\"#@param {type:\"string\"}\n",
" translation_z = \"0:(10)\"#@param {type:\"string\"}\n",
" rotation_3d_x = \"0:(0)\"#@param {type:\"string\"}\n",
" rotation_3d_y = \"0:(0)\"#@param {type:\"string\"}\n",
" rotation_3d_z = \"0:(0)\"#@param {type:\"string\"}\n",
" flip_2d_perspective = False #@param {type:\"boolean\"}\n",
" perspective_flip_theta = \"0:(0)\"#@param {type:\"string\"}\n",
" perspective_flip_phi = \"0:(t%15)\"#@param {type:\"string\"}\n",
" perspective_flip_gamma = \"0:(0)\"#@param {type:\"string\"}\n",
" perspective_flip_fv = \"0:(53)\"#@param {type:\"string\"}\n",
" noise_schedule = \"0: (0.02)\"#@param {type:\"string\"}\n",
" strength_schedule = \"0: (0.65)\"#@param {type:\"string\"}\n",
" contrast_schedule = \"0: (1.0)\"#@param {type:\"string\"}\n",
"\n",
" #@markdown ####**Coherence:**\n",
" color_coherence = 'Match Frame 0 LAB' #@param ['None', 'Match Frame 0 HSV', 'Match Frame 0 LAB', 'Match Frame 0 RGB'] {type:'string'}\n",
" diffusion_cadence = '1' #@param ['1','2','3','4','5','6','7','8'] {type:'string'}\n",
"\n",
" #@markdown ####**3D Depth Warping:**\n",
" use_depth_warping = True #@param {type:\"boolean\"}\n",
" midas_weight = 0.3#@param {type:\"number\"}\n",
" near_plane = 200\n",
" far_plane = 10000\n",
" fov = 40#@param {type:\"number\"}\n",
" padding_mode = 'border'#@param ['border', 'reflection', 'zeros'] {type:'string'}\n",
" sampling_mode = 'bicubic'#@param ['bicubic', 'bilinear', 'nearest'] {type:'string'}\n",
" save_depth_maps = False #@param {type:\"boolean\"}\n",
"\n",
" #@markdown ####**Video Input:**\n",
" video_init_path ='/content/video_in.mp4'#@param {type:\"string\"}\n",
" extract_nth_frame = 1#@param {type:\"number\"}\n",
" overwrite_extracted_frames = True #@param {type:\"boolean\"}\n",
" use_mask_video = False #@param {type:\"boolean\"}\n",
" video_mask_path ='/content/video_in.mp4'#@param {type:\"string\"}\n",
"\n",
" #@markdown ####**Interpolation:**\n",
" interpolate_key_frames = False #@param {type:\"boolean\"}\n",
" interpolate_x_frames = 4 #@param {type:\"number\"}\n",
" \n",
" #@markdown ####**Resume Animation:**\n",
" resume_from_timestring = False #@param {type:\"boolean\"}\n",
" resume_timestring = \"20220829210106\" #@param {type:\"string\"}\n",
"\n",
" return locals()"
],
"outputs": [],
"execution_count": null
},
{
"cell_type": "markdown",
"metadata": {
"id": "63UOJvU3xdPS"
},
"source": [
"### Prompts\n",
"`animation_mode: None` batches on list of *prompts*. `animation_mode: 2D` uses *animation_prompts* key frame sequence"
]
},
{
"cell_type": "code",
"metadata": {
"id": "2ujwkGZTcGev"
},
"source": [
"prompts = [\n",
" \"a beautiful forest by Asher Brown Durand, trending on Artstation\", # the first prompt I want\n",
" \"a beautiful portrait of a woman by Artgerm, trending on Artstation\", # the second prompt I want\n",
" #\"this prompt I don't want it I commented it out\",\n",
" #\"a nousr robot, trending on Artstation\", # use \"nousr robot\" with the robot diffusion model (see model_checkpoint setting)\n",
" #\"touhou 1girl komeiji_koishi portrait, green hair\", # waifu diffusion prompts can use danbooru tag groups (see model_checkpoint)\n",
" #\"this prompt has weights if prompt weighting enabled:2 can also do negative:-2\", # (see prompt_weighting)\n",
"]\n",
"\n",
"animation_prompts = {\n",
" 0: \"a beautiful apple, trending on Artstation\",\n",
" 20: \"a beautiful banana, trending on Artstation\",\n",
" 30: \"a beautiful coconut, trending on Artstation\",\n",
" 40: \"a beautiful durian, trending on Artstation\",\n",
"}"
],
"outputs": [],
"execution_count": null
},
{
"cell_type": "markdown",
"metadata": {
"id": "s8RAo2zI-vQm"
},
"source": [
"# Run"
]
},
{
"cell_type": "code",
"metadata": {
"cellView": "form",
"id": "qH74gBWDd2oq"
},
"source": [
"#@markdown **Load Settings**\n",
"override_settings_with_file = False #@param {type:\"boolean\"}\n",
"custom_settings_file = \"/content/drive/MyDrive/Settings.txt\"#@param {type:\"string\"}\n",
"\n",
"def Root():\n",
" return locals()\n",
"\n",
"def DeforumArgs():\n",
" #@markdown **Image Settings**\n",
" W = 512 #@param\n",
" H = 512 #@param\n",
" W, H = map(lambda x: x - x % 64, (W, H)) # resize to integer multiple of 64\n",
"\n",
" #@markdown **Sampling Settings**\n",
" seed = -1 #@param\n",
" sampler = 'klms' #@param [\"klms\",\"dpm2\",\"dpm2_ancestral\",\"heun\",\"euler\",\"euler_ancestral\",\"plms\", \"ddim\"]\n",
" steps = 50 #@param\n",
" scale = 7 #@param\n",
" ddim_eta = 0.0 #@param\n",
" dynamic_threshold = None\n",
" static_threshold = None \n",
"\n",
" #@markdown **Save & Display Settings**\n",
" save_samples = True #@param {type:\"boolean\"}\n",
" save_settings = True #@param {type:\"boolean\"}\n",
" display_samples = True #@param {type:\"boolean\"}\n",
" save_sample_per_step = False #@param {type:\"boolean\"}\n",
" show_sample_per_step = False #@param {type:\"boolean\"}\n",
"\n",
" #@markdown **Prompt Settings**\n",
" prompt_weighting = False #@param {type:\"boolean\"}\n",
" normalize_prompt_weights = True #@param {type:\"boolean\"}\n",
" log_weighted_subprompts = False #@param {type:\"boolean\"}\n",
"\n",
" #@markdown **Batch Settings**\n",
" n_batch = 1 #@param\n",
" batch_name = \"StableFun\" #@param {type:\"string\"}\n",
" filename_format = \"{timestring}_{index}_{prompt}.png\" #@param [\"{timestring}_{index}_{seed}.png\",\"{timestring}_{index}_{prompt}.png\"]\n",
" seed_behavior = \"iter\" #@param [\"iter\",\"fixed\",\"random\"]\n",
" make_grid = False #@param {type:\"boolean\"}\n",
" grid_rows = 2 #@param \n",
" outdir = get_output_folder(output_path, batch_name)\n",
"\n",
" #@markdown **Init Settings**\n",
" use_init = False #@param {type:\"boolean\"}\n",
" strength = 0.0 #@param {type:\"number\"}\n",
" strength_0_no_init = True # Set the strength to 0 automatically when no init image is used\n",
" init_image = \"https://cdn.pixabay.com/photo/2022/07/30/13/10/green-longhorn-beetle-7353749_1280.jpg\" #@param {type:\"string\"}\n",
" # Whiter areas of the mask are areas that change more\n",
" use_mask = False #@param {type:\"boolean\"}\n",
" use_alpha_as_mask = False # use the alpha channel of the init image as the mask\n",
" mask_file = \"https://www.filterforge.com/wiki/images/archive/b/b7/20080927223728%21Polygonal_gradient_thumb.jpg\" #@param {type:\"string\"}\n",
" invert_mask = False #@param {type:\"boolean\"}\n",
" # Adjust mask image, 1.0 is no adjustment. Should be positive numbers.\n",
" mask_brightness_adjust = 1.0 #@param {type:\"number\"}\n",
" mask_contrast_adjust = 1.0 #@param {type:\"number\"}\n",
" # Overlay the masked image at the end of the generation so it does not get degraded by encoding and decoding\n",
" overlay_mask = True # {type:\"boolean\"}\n",
" # Blur edges of final overlay mask, if used. Minimum = 0 (no blur)\n",
" mask_overlay_blur = 5 # {type:\"number\"}\n",
"\n",
" n_samples = 1 # doesnt do anything\n",
" precision = 'autocast' \n",
" C = 4\n",
" f = 8\n",
"\n",
" prompt = \"\"\n",
" timestring = \"\"\n",
" init_latent = None\n",
" init_sample = None\n",
" init_c = None\n",
"\n",
" return locals()\n",
"\n",
"root = Root()\n",
"args_dict = DeforumArgs()\n",
"anim_args_dict = DeforumAnimArgs()\n",
"\n",
"if override_settings_with_file:\n",
" load_args(args_dict,anim_args_dict,custom_settings_file)\n",
"\n",
"root = SimpleNamespace(**root)\n",
"args = SimpleNamespace(**args_dict)\n",
"anim_args = SimpleNamespace(**anim_args_dict)\n",
"\n",
"args.timestring = time.strftime('%Y%m%d%H%M%S')\n",
"args.strength = max(0.0, min(1.0, args.strength))\n",
"\n",
"root.model = model\n",
"root.device = device\n",
"root.models_path = models_path\n",
"root.output_path = output_path\n",
"root.half_precision = True\n",
"\n",
"if args.seed == -1:\n",
" args.seed = random.randint(0, 2**32 - 1)\n",
"if not args.use_init:\n",
" args.init_image = None\n",
"if args.sampler == 'plms' and (args.use_init or anim_args.animation_mode != 'None'):\n",
" print(f\"Init images aren't supported with PLMS yet, switching to KLMS\")\n",
" args.sampler = 'klms'\n",
"if args.sampler != 'ddim':\n",
" args.ddim_eta = 0\n",
"\n",
"if anim_args.animation_mode == 'None':\n",
" anim_args.max_frames = 1\n",
"elif anim_args.animation_mode == 'Video Input':\n",
" args.use_init = True\n",
"\n",
"# clean up unused memory\n",
"gc.collect()\n",
"torch.cuda.empty_cache()\n",
"\n",
"# dispatch to appropriate renderer\n",
"if anim_args.animation_mode == '2D' or anim_args.animation_mode == '3D':\n",
" render_animation(args, anim_args, animation_prompts, root)\n",
"elif anim_args.animation_mode == 'Video Input':\n",
" render_input_video(args, anim_args, animation_prompts, root)\n",
"elif anim_args.animation_mode == 'Interpolation':\n",
" render_interpolation(args, anim_args, animation_prompts, root)\n",
"else:\n",
" render_image_batch(args, prompts, root) "
],
"outputs": [],
"execution_count": null
},
{
"cell_type": "markdown",
"metadata": {
"id": "4zV0J_YbMCTx"
},
"source": [
"# Create video from frames"
]
},
{
"cell_type": "code",
"metadata": {
"cellView": "form",
"id": "no2jP8HTMBM0"
},
"source": [
"skip_video_for_run_all = True #@param {type: 'boolean'}\n",
"fps = 12 #@param {type:\"number\"}\n",
"#@markdown **Manual Settings**\n",
"use_manual_settings = False #@param {type:\"boolean\"}\n",
"image_path = \"/content/drive/MyDrive/AI/StableDiffusion/2022-09/20220903000939_%05d.png\" #@param {type:\"string\"}\n",
"mp4_path = \"/content/drive/MyDrive/AI/StableDiffu'/content/drive/MyDrive/AI/StableDiffusion/2022-09/sion/2022-09/20220903000939.mp4\" #@param {type:\"string\"}\n",
"render_steps = False #@param {type: 'boolean'}\n",
"path_name_modifier = \"x0_pred\" #@param [\"x0_pred\",\"x\"]\n",
"\n",
"\n",
"if skip_video_for_run_all == True:\n",
" print('Skipping video creation, uncheck skip_video_for_run_all if you want to run it')\n",
"else:\n",
" import os\n",
" import subprocess\n",
" from base64 import b64encode\n",
"\n",
" print(f\"{image_path} -> {mp4_path}\")\n",
"\n",
" if use_manual_settings:\n",
" max_frames = \"200\" #@param {type:\"string\"}\n",
" else:\n",
" if render_steps: # render steps from a single image\n",
" fname = f\"{path_name_modifier}_%05d.png\"\n",
" all_step_dirs = [os.path.join(args.outdir, d) for d in os.listdir(args.outdir) if os.path.isdir(os.path.join(args.outdir,d))]\n",
" newest_dir = max(all_step_dirs, key=os.path.getmtime)\n",
" image_path = os.path.join(newest_dir, fname)\n",
" print(f\"Reading images from {image_path}\")\n",
" mp4_path = os.path.join(newest_dir, f\"{args.timestring}_{path_name_modifier}.mp4\")\n",
" max_frames = str(args.steps)\n",
" else: # render images for a video\n",
" image_path = os.path.join(args.outdir, f\"{args.timestring}_%05d.png\")\n",
" mp4_path = os.path.join(args.outdir, f\"{args.timestring}.mp4\")\n",
" max_frames = str(anim_args.max_frames)\n",
"\n",
" # make video\n",
" cmd = [\n",
" 'ffmpeg',\n",
" '-y',\n",
" '-vcodec', 'png',\n",
" '-r', str(fps),\n",
" '-start_number', str(0),\n",
" '-i', image_path,\n",
" '-frames:v', max_frames,\n",
" '-c:v', 'libx264',\n",
" '-vf',\n",
" f'fps={fps}',\n",
" '-pix_fmt', 'yuv420p',\n",
" '-crf', '17',\n",
" '-preset', 'veryfast',\n",
" '-pattern_type', 'sequence',\n",
" mp4_path\n",
" ]\n",
" process = subprocess.Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE)\n",
" stdout, stderr = process.communicate()\n",
" if process.returncode != 0:\n",
" print(stderr)\n",
" raise RuntimeError(stderr)\n",
"\n",
" mp4 = open(mp4_path,'rb').read()\n",
" data_url = \"data:video/mp4;base64,\" + b64encode(mp4).decode()\n",
" display.display( display.HTML(f'<video controls loop><source src=\"{data_url}\" type=\"video/mp4\"></video>') )"
],
"outputs": [],
"execution_count": null
},
{
"cell_type": "markdown",
"source": [
"# Disconnect when finished\n"
],
"metadata": {
"id": "XccAk0RoRme0"
}
},
{
"cell_type": "code",
"source": [
"skip_disconnect_for_run_all = True #@param {type: 'boolean'}\n",
"\n",
"if skip_disconnect_for_run_all == True:\n",
" print('Skipping disconnect, uncheck skip_disconnect_for_run_all if you want to run it')\n",
"else:\n",
" from google.colab import runtime\n",
" runtime.unassign()"
],
"metadata": {
"cellView": "form",
"id": "_x6obwPURfSm"
},
"execution_count": null,
"outputs": []
}
],
"metadata": {
"accelerator": "GPU",
"colab": {
"collapsed_sections": [],
"private_outputs": true,
"provenance": []
},
"gpuClass": "standard",
"kernelspec": {
"display_name": "Python 3 (ipykernel)",
"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": "3.9.13"
}
},
"nbformat": 4,
"nbformat_minor": 0
}