perf: improve startup time by doing some imports lazily (#233)

just running `aimg --help` or `aimg --version` was very slow due to all the imports being brought in eagerly

Before changes `aimg --help`
`2.24s user 4.05s system 184% cpu 3.416 total`

After changes:
`0.04s user 0.02s system 8% cpu 0.625 total`

Used `PYTHONPROFILEIMPORTTIME=1 aimg --help` to find time consuming imports.

Also switched to using `scripts` instead of `entrypoints` since the scripts are much faster.

Made duplicate SAMPLER_TYPE_OPTIONS that can be loaded without loading all the samplers themselves.

Likely a breaking change - not sure.

.github/workflows/ci.yaml

@@ -33,7 +33,7 @@ jobs:
         python-version: 3.9
     - name: Install dependencies
       run: |
-        python -m pip install --disable-pip-version-check black isort
+        python -m pip install --disable-pip-version-check black==22.12.0 isort==5.11.4
     - name: Autoformatter
       run: |
         black --diff .

README.md

@@ -291,6 +291,8 @@ docker run -it --gpus all -v $HOME/.cache/huggingface:/root/.cache/huggingface -
 
 ## ChangeLog
 
+- perf: cli now has minimal overhead such that `aimg --help` runs in ~650ms instead of ~3400ms
+
 **8.3.1**
 - fix: init-image-strength type
 

imaginairy/__init__.py

@@ -1,19 +1,12 @@
-import os.path
+import os
 
 # tells pytorch to allow MPS usage (for Mac M1 compatibility)
 os.putenv("PYTORCH_ENABLE_MPS_FALLBACK", "1")
-import PIL.Image  # noqa
 
 from .api import imagine, imagine_image_files  # noqa
-from .enhancers.describe_image_blip import generate_caption  # noqa
 from .schema import (  # noqa
     ImaginePrompt,
     ImagineResult,
     LazyLoadingImage,
     WeightedPrompt,
 )
-from .version import __version__  # noqa
-
-# https://stackoverflow.com/questions/71738218/module-pil-has-not-attribute-resampling
-if not hasattr(PIL.Image, "Resampling"):  # Pillow<9.0
-    PIL.Image.Resampling = PIL.Image

imaginairy/api.py

@@ -2,41 +2,7 @@ import logging
 import os
 import re
 
-import numpy as np
-import torch
-import torch.nn
-from einops import rearrange, repeat
-from PIL import Image, ImageDraw, ImageOps
-from pytorch_lightning import seed_everything
-from torch.cuda import OutOfMemoryError
-
-from imaginairy.animations import make_bounce_animation
-from imaginairy.enhancers.clip_masking import get_img_mask
-from imaginairy.enhancers.describe_image_blip import generate_caption
-from imaginairy.enhancers.face_restoration_codeformer import enhance_faces
-from imaginairy.enhancers.upscale_realesrgan import upscale_image
-from imaginairy.img_utils import pillow_fit_image_within, pillow_img_to_torch_image
-from imaginairy.log_utils import (
-    ImageLoggingContext,
-    log_conditioning,
-    log_img,
-    log_latent,
-)
-from imaginairy.model_manager import get_diffusion_model
-from imaginairy.modules.midas.utils import AddMiDaS
-from imaginairy.outpaint import outpaint_arg_str_parse, prepare_image_for_outpaint
-from imaginairy.safety import SafetyMode, create_safety_score
-from imaginairy.samplers import SAMPLER_LOOKUP
-from imaginairy.samplers.base import NoiseSchedule, noise_an_image
-from imaginairy.samplers.editing import CFGEditingDenoiser
-from imaginairy.schema import ImaginePrompt, ImagineResult
-from imaginairy.utils import (
-    fix_torch_group_norm,
-    fix_torch_nn_layer_norm,
-    get_device,
-    platform_appropriate_autocast,
-    randn_seeded,
-)
+from imaginairy.schema import SafetyMode
 
 logger = logging.getLogger(__name__)
 
@@ -64,6 +30,11 @@ def imagine_image_files(
     make_compare_gif=False,
     return_filename_type="generated",
 ):
+    from PIL import ImageDraw
+
+    from imaginairy.animations import make_bounce_animation
+    from imaginairy.img_utils import pillow_fit_image_within
+
     generated_imgs_path = os.path.join(outdir, "generated")
     os.makedirs(generated_imgs_path, exist_ok=True)
 
@@ -167,6 +138,16 @@ def imagine(
     add_caption=False,
     unsafe_retry_count=1,
 ):
+    import torch.nn
+
+    from imaginairy.schema import ImaginePrompt
+    from imaginairy.utils import (
+        fix_torch_group_norm,
+        fix_torch_nn_layer_norm,
+        get_device,
+        platform_appropriate_autocast,
+    )
+
     prompts = [ImaginePrompt(prompts)] if isinstance(prompts, str) else prompts
     prompts = [prompts] if isinstance(prompts, ImaginePrompt) else prompts
 
@@ -214,6 +195,34 @@ def _generate_single_image(
     half_mode=None,
     add_caption=False,
 ):
+    import numpy as np
+    import torch.nn
+    from einops import rearrange, repeat
+    from PIL import Image, ImageOps
+    from pytorch_lightning import seed_everything
+    from torch.cuda import OutOfMemoryError
+
+    from imaginairy.enhancers.clip_masking import get_img_mask
+    from imaginairy.enhancers.describe_image_blip import generate_caption
+    from imaginairy.enhancers.face_restoration_codeformer import enhance_faces
+    from imaginairy.enhancers.upscale_realesrgan import upscale_image
+    from imaginairy.img_utils import pillow_fit_image_within, pillow_img_to_torch_image
+    from imaginairy.log_utils import (
+        ImageLoggingContext,
+        log_conditioning,
+        log_img,
+        log_latent,
+    )
+    from imaginairy.model_manager import get_diffusion_model
+    from imaginairy.modules.midas.utils import AddMiDaS
+    from imaginairy.outpaint import outpaint_arg_str_parse, prepare_image_for_outpaint
+    from imaginairy.safety import create_safety_score
+    from imaginairy.samplers import SAMPLER_LOOKUP
+    from imaginairy.samplers.base import NoiseSchedule, noise_an_image
+    from imaginairy.samplers.editing import CFGEditingDenoiser
+    from imaginairy.schema import ImaginePrompt, ImagineResult
+    from imaginairy.utils import get_device, randn_seeded
+
     latent_channels = 4
     downsampling_factor = 8
     batch_size = 1

imaginairy/bin/aimg

@@ -0,0 +1,6 @@
+#!/bin/env python
+# -*- coding: utf-8 -*-
+
+if __name__ == "__main__":
+    from imaginairy import cmds
+    cmds.aimg()

imaginairy/bin/imagine

@@ -0,0 +1,6 @@
+#!/bin/env python
+# -*- coding: utf-8 -*-
+
+if __name__ == "__main__":
+    from imaginairy import cmds
+    cmds.imagine_cmd()

imaginairy/cmds.py

@@ -1,29 +1,10 @@
 import logging
 import math
-import os.path
 
 import click
 from click_shell import shell
-from tqdm import tqdm
-
-from imaginairy import LazyLoadingImage, __version__, config, generate_caption
-from imaginairy.animations import make_bounce_animation
-from imaginairy.api import imagine_image_files
-from imaginairy.debug_info import get_debug_info
-from imaginairy.enhancers.prompt_expansion import expand_prompts
-from imaginairy.enhancers.upscale_realesrgan import upscale_image
-from imaginairy.log_utils import configure_logging
-from imaginairy.prompt_schedules import parse_schedule_strs, prompt_mutator
-from imaginairy.samplers import SAMPLER_TYPE_OPTIONS
-from imaginairy.schema import ImaginePrompt
-from imaginairy.surprise_me import create_surprise_me_images
-from imaginairy.train import train_diffusion_model
-from imaginairy.training_tools.image_prep import (
-    create_class_images,
-    get_image_filenames,
-    prep_images,
-)
-from imaginairy.training_tools.prune_model import prune_diffusion_ckpt
+
+from imaginairy import config
 
 logger = logging.getLogger(__name__)
 
@@ -111,7 +92,7 @@ logger = logging.getLogger(__name__)
     "--sampler",
     default=config.DEFAULT_SAMPLER,
     show_default=True,
-    type=click.Choice(SAMPLER_TYPE_OPTIONS),
+    type=click.Choice(config.SAMPLER_TYPE_OPTIONS),
     help="What sampling strategy to use.",
 )
 @click.option(
@@ -413,7 +394,7 @@ def imagine_cmd(
     "--sampler",
     default=config.DEFAULT_SAMPLER,
     show_default=True,
-    type=click.Choice(SAMPLER_TYPE_OPTIONS),
+    type=click.Choice(config.SAMPLER_TYPE_OPTIONS),
     help="What sampling strategy to use.",
 )
 @click.option(
@@ -603,6 +584,9 @@ def edit_image(  # noqa
     arg_schedules,
     make_compilation_animation,
 ):
+    from imaginairy.log_utils import configure_logging
+    from imaginairy.surprise_me import create_surprise_me_images
+
     init_image_strength = 1
     if surprise_me and prompt_texts:
         raise ValueError("Cannot use surprise_me and prompt_texts together")
@@ -698,11 +682,23 @@ def _imagine_cmd(
     make_compilation_animation=False,
 ):
     """Have the AI generate images. alias:imagine."""
+    import os.path
+
+    from imaginairy import LazyLoadingImage
+    from imaginairy.animations import make_bounce_animation
+    from imaginairy.api import imagine_image_files
+    from imaginairy.enhancers.prompt_expansion import expand_prompts
+    from imaginairy.log_utils import configure_logging
+    from imaginairy.prompt_schedules import parse_schedule_strs, prompt_mutator
+    from imaginairy.schema import ImaginePrompt
+
     if ctx.invoked_subcommand is not None:
         return
 
     if version:
-        print(__version__)
+        from imaginairy.version import get_version
+
+        print(get_version())
         return
 
     if quiet:
@@ -816,13 +812,14 @@ def aimg():
     ✨ Run `aimg` to start a persistent shell session.  This makes generation and editing much
     quicker since the model can stay loaded in memory.
     """
-    configure_logging()
 
 
 @aimg.command()
 def version():
     """Print the version."""
-    print(__version__)
+    from imaginairy.version import get_version
+
+    print(get_version())
 
 
 @click.argument("image_filepaths", nargs=-1)
@@ -838,6 +835,13 @@ def upscale_cmd(image_filepaths, outdir):
     """
     Upscale an image 4x using AI.
     """
+    import os.path
+
+    from tqdm import tqdm
+
+    from imaginairy import LazyLoadingImage
+    from imaginairy.enhancers.upscale_realesrgan import upscale_image
+
     os.makedirs(outdir, exist_ok=True)
 
     for p in tqdm(image_filepaths):
@@ -847,7 +851,7 @@ def upscale_cmd(image_filepaths, outdir):
         else:
             img = LazyLoadingImage(filepath=p)
         logger.info(
-            f"Upscaling {p} from {img.width}x{img.height } to {img.width * 4}x{img.height*4} and saving it to {savepath}"
+            f"Upscaling {p} from {img.width}x{img.height} to {img.width * 4}x{img.height * 4} and saving it to {savepath}"
         )
 
         img = upscale_image(img)
@@ -859,6 +863,10 @@ def upscale_cmd(image_filepaths, outdir):
 @aimg.command()
 def describe(image_filepaths):
     """Generate text descriptions of images."""
+
+    from imaginairy import LazyLoadingImage
+    from imaginairy.enhancers.describe_image_blip import generate_caption
+
     imgs = []
     for p in image_filepaths:
         if p.startswith("http"):
@@ -991,6 +999,15 @@ def train_concept(
 
     You can find a lot of relevant instructions here: https://github.com/JoePenna/Dreambooth-Stable-Diffusion
     """
+    import os.path
+
+    from imaginairy.train import train_diffusion_model
+    from imaginairy.training_tools.image_prep import (
+        create_class_images,
+        get_image_filenames,
+        prep_images,
+    )
+
     target_size = 512
     # Step 1. Crop and enhance the training images
     prepped_images_path = os.path.join(concept_images_dir, "prepped-images")
@@ -1081,7 +1098,9 @@ def prepare_images(images_dir, is_person, target_size):
         aimg prep-images --person ./images/selfies
         aimg prep-images ./images/toy-train
     """
-    configure_logging()
+
+    from imaginairy.training_tools.image_prep import prep_images
+
     prep_images(images_dir=images_dir, is_person=is_person, target_size=target_size)
 
 
@@ -1097,8 +1116,9 @@ def prune_ckpt(ckpt_paths):
     Example:
         aimg prune-ckpt ./path/to/checkpoint.ckpt
     """
+    from imaginairy.training_tools.prune_model import prune_diffusion_ckpt
+
     click.secho("Pruning checkpoint files...")
-    configure_logging()
     for p in ckpt_paths:
         prune_diffusion_ckpt(p)
 
@@ -1108,6 +1128,8 @@ def system_info():
     """
     Display system information. Submit this when reporting bugs.
     """
+    from imaginairy.debug_info import get_debug_info
+
     for k, v in get_debug_info().items():
         k += ":"
         click.secho(f"{k: <30} {v}")

imaginairy/config.py

@@ -142,3 +142,18 @@ MODEL_CONFIG_SHORTCUTS["openjourney"] = MODEL_CONFIG_SHORTCUTS["openjourney-v2"]
 MODEL_CONFIG_SHORTCUTS["oj"] = MODEL_CONFIG_SHORTCUTS["openjourney-v2"]
 
 MODEL_SHORT_NAMES = sorted(MODEL_CONFIG_SHORTCUTS.keys())
+
+SAMPLER_TYPE_OPTIONS = [
+    "plms",
+    "ddim",
+    "k_dpm_fast",
+    "k_dpm_adaptive",
+    "k_lms",
+    "k_dpm_2",
+    "k_dpm_2_a",
+    "k_dpmpp_2m",
+    "k_dpmpp_2s_a",
+    "k_euler",
+    "k_euler_a",
+    "k_heun",
+]

imaginairy/debug_info.py

@@ -3,13 +3,14 @@ import sys
 
 import torch
 
-from imaginairy import __version__
 from imaginairy.utils import get_device, get_hardware_description
+from imaginairy.version import get_version
 
 
 def get_debug_info():
+
     data = {
-        "imaginairy_version": __version__,
+        "imaginairy_version": get_version(),
         "imaginairy_path": os.path.dirname(__file__),
         "python_version": f"{sys.version_info.major}.{sys.version_info.minor}.{sys.version_info.micro}",
         "python_installation_path": sys.executable,

imaginairy/log_utils.py

@@ -4,12 +4,6 @@ import re
 import time
 import warnings
 
-import torch
-from pytorch_lightning import _logger as pytorch_logger
-from torchvision.transforms import ToPILImage
-from transformers.modeling_utils import logger as modeling_logger
-from transformers.utils.logging import _configure_library_root_logger
-
 _CURRENT_LOGGING_CONTEXT = None
 
 logger = logging.getLogger(__name__)
@@ -159,6 +153,9 @@ class ImageLoggingContext:
     def log_img(self, img, description):
         if not self.debug_img_callback:
             return
+        import torch
+        from torchvision.transforms import ToPILImage
+
         self.image_count += 1
         if isinstance(img, torch.Tensor):
             img = ToPILImage()(img.squeeze().cpu().detach())
@@ -200,6 +197,8 @@ def filesafe_text(t):
 
 
 def conditioning_to_img(conditioning):
+    from torchvision.transforms import ToPILImage
+
     return ToPILImage()(conditioning)
 
 
@@ -252,6 +251,9 @@ def configure_logging(level="INFO"):
 
 
 def disable_transformers_custom_logging():
+    from transformers.modeling_utils import logger as modeling_logger
+    from transformers.utils.logging import _configure_library_root_logger
+
     _configure_library_root_logger()
     _logger = modeling_logger.parent
     _logger.handlers = []
@@ -263,6 +265,8 @@ def disable_transformers_custom_logging():
 
 
 def disable_pytorch_lighting_custom_logging():
+    from pytorch_lightning import _logger as pytorch_logger
+
     try:
         from pytorch_lightning.utilities.seed import log  # noqa
 

imaginairy/safety.py

@@ -6,15 +6,11 @@ from diffusers.pipelines.stable_diffusion import safety_checker as safety_checke
 from transformers import AutoFeatureExtractor
 
 from imaginairy.enhancers.blur_detect import is_blurry
+from imaginairy.schema import SafetyMode
 
 logger = logging.getLogger(__name__)
 
 
-class SafetyMode:
-    STRICT = "strict"
-    RELAXED = "relaxed"
-
-
 class SafetyResult:
     # increase this value to create a stronger `nfsw` filter
     # at the cost of increasing the possibility of filtering benign images

imaginairy/schema.py

@@ -4,17 +4,8 @@ import logging
 import os.path
 import random
 from datetime import datetime, timezone
-from functools import lru_cache
-
-import requests
-from PIL import Image, ImageOps
-from urllib3.exceptions import LocationParseError
-from urllib3.util import parse_url
 
 from imaginairy import config
-from imaginairy.model_manager import get_model_default_image_size
-from imaginairy.samplers import SAMPLER_LOOKUP, SamplerName
-from imaginairy.utils import get_device, get_hardware_description
 
 logger = logging.getLogger(__name__)
 
@@ -36,6 +27,9 @@ class LazyLoadingImage:
 
         # validate url is valid url
         if url:
+            from urllib3.exceptions import LocationParseError
+            from urllib3.util import parse_url
+
             try:
                 parsed_url = parse_url(url)
             except LocationParseError:
@@ -53,6 +47,7 @@ class LazyLoadingImage:
             raise AttributeError()
         if self._img:
             return getattr(self._img, key)
+        from PIL import Image, ImageOps
 
         if self._lazy_filepath:
             self._img = Image.open(self._lazy_filepath)
@@ -60,6 +55,8 @@ class LazyLoadingImage:
                 f"Loaded input 🖼  of size {self._img.size} from {self._lazy_filepath}"
             )
         elif self._lazy_url:
+            import requests
+
             self._img = Image.open(
                 requests.get(self._lazy_url, stream=True, timeout=60).raw
             )
@@ -148,6 +145,8 @@ class ImaginePrompt:
         self.validate()
 
     def validate(self):
+        from imaginairy.samplers import SAMPLER_LOOKUP, SamplerName
+
         self.prompts = self.process_prompt_input(self.prompts)
 
         if self.tile_mode is True:
@@ -182,6 +181,8 @@ class ImaginePrompt:
         )
 
         if self.height is None or self.width is None or self.steps is None:
+            from imaginairy.model_manager import get_model_default_image_size
+
             SamplerCls = SAMPLER_LOOKUP[self.sampler_type]
             self.steps = self.steps or SamplerCls.default_steps
             self.width = self.width or get_model_default_image_size(self.model)
@@ -281,6 +282,8 @@ class ImagineResult:
         timings=None,
         progress_latents=None,
     ):
+        from imaginairy.utils import get_device, get_hardware_description
+
         self.prompt = prompt
 
         self.images = {"generated": img}
@@ -327,6 +330,8 @@ class ImagineResult:
         return " ".join(f"{k}:{v:.2f}s" for k, v in self.timings.items())
 
     def _exif(self):
+        from PIL import Image
+
         exif = Image.Exif()
         exif[ExifCodes.ImageDescription] = self.prompt.prompt_description()
         exif[ExifCodes.UserComment] = json.dumps(self.metadata_dict())
@@ -349,6 +354,6 @@ class ImagineResult:
         img.convert("RGB").save(save_path, exif=self._exif())
 
 
-@lru_cache(maxsize=2)
-def _get_briefly_cached_url(url):
-    return requests.get(url, timeout=60)
+class SafetyMode:
+    STRICT = "strict"
+    RELAXED = "relaxed"

imaginairy/version.py

@@ -1,6 +1,7 @@
-from importlib.metadata import PackageNotFoundError, version
+def get_version():
+    from importlib.metadata import PackageNotFoundError, version
 
-try:
-    __version__ = version("imaginairy")
-except PackageNotFoundError:
-    __version__ = None
+    try:
+        return version("imaginairy")
+    except PackageNotFoundError:
+        return None

setup.py

@@ -16,16 +16,12 @@ setup(
         "Source": "https://github.com/brycedrennan/imaginAIry",
     },
     packages=find_packages(include=("imaginairy", "imaginairy.*")),
-    entry_points={
-        "console_scripts": [
-            "imagine=imaginairy.cmds:imagine_cmd",
-            "aimg=imaginairy.cmds:aimg",
-        ],
-    },
+    scripts=["imaginairy/bin/aimg", "imaginairy/bin/imagine"],
     package_data={
         "imaginairy": [
             "configs/*.yaml",
             "data/*.*",
+            "bin/*.*",
             "enhancers/phraselists/*.txt",
             "vendored/clip/*.txt.gz",
             "vendored/clipseg/*.pth",

tests/test_config.py

@@ -0,0 +1,6 @@
+from imaginairy import config
+from imaginairy.samplers import SAMPLER_TYPE_OPTIONS
+
+
+def test_sampler_options():
+    assert set(config.SAMPLER_TYPE_OPTIONS) == set(SAMPLER_TYPE_OPTIONS)

tox.ini

@@ -11,7 +11,7 @@ format = pylint
 skip = */.tox/*,*/.env/*,build/*,*/downloads/*,other/*,prolly_delete/*,downloads/*,imaginairy/vendored/*,testing_support/vastai_cli_official.py,.eggs/*
 linters = pylint,pycodestyle,pyflakes,mypy
 ignore =
-    Z999,C0103,C0301,C0302,C0114,C0115,C0116,
+    Z999,C0103,C0301,C0302,C0114,C0115,C0116,C0415,
     Z999,D100,D101,D102,D103,D105,D106,D107,D200,D202,D203,D205,D212,D400,D401,D406,D407,D413,D415,D417,
     Z999,E203,E501,E1101,E1131,E1135,E1136,
     Z999,R0901,R0902,R0903,R0904,R0193,R0912,R0913,R0914,R0915,R1702,