model manager defaults to consistent values of device and precision

invokeai/backend/__init__.py

@@ -6,7 +6,10 @@ from .generator import (
     InvokeAIGeneratorBasicParams,
     InvokeAIGeneratorFactory,
     InvokeAIGenerator,
-    InvokeAIGeneratorOutput
+    InvokeAIGeneratorOutput,
+    Txt2Img,
+    Img2Img,
+    Inpaint
 )
 from .model_management import ModelManager
 from .args import Args

invokeai/backend/generator/base.py

@@ -5,6 +5,7 @@ including img2img, txt2img, and inpaint
 from __future__ import annotations
 
 import importlib
+import itertools
 import dataclasses
 import diffusers
 import os
@@ -20,7 +21,7 @@ from PIL import Image, ImageChops, ImageFilter
 from accelerate.utils import set_seed
 from diffusers import DiffusionPipeline
 from tqdm import trange
-from typing import List, Type
+from typing import List, Type, Iterator
 from dataclasses import dataclass, field
 from diffusers.schedulers import SchedulerMixin as Scheduler
 
@@ -77,7 +78,7 @@ class InvokeAIGeneratorOutput:
 class InvokeAIGeneratorFactory(object):
     def __init__(self,
                  model_manager: ModelManager,
-                 params: InvokeAIGeneratorBasicParams
+                 params: InvokeAIGeneratorBasicParams=InvokeAIGeneratorBasicParams(),
                  ):
         self.model_manager = model_manager
         self.params = params
@@ -115,7 +116,7 @@ class InvokeAIGenerator(metaclass=ABCMeta):
 
     def __init__(self,
                  model_manager: ModelManager,
-                 params: InvokeAIGeneratorBasicParams=InvokeAIGeneratorBasicParams(),
+                 params: InvokeAIGeneratorBasicParams,
                  ):
         self.model_manager=model_manager
         self.params=params
@@ -124,9 +125,30 @@ class InvokeAIGenerator(metaclass=ABCMeta):
                  prompt: str='',
                  callback: callable=None,
                  step_callback: callable=None,
+                 iterations: int=1,
                  **keyword_args,
-                 )->List[InvokeAIGeneratorOutput]:
+                 )->Iterator[InvokeAIGeneratorOutput]:
+        '''
+        Return an iterator across the indicated number of generations.
+        Each time the iterator is called it will return an InvokeAIGeneratorOutput
+        object. Use like this:
 
+           outputs = txt2img.generate(prompt='banana sushi', iterations=5)
+           for result in outputs:
+               print(result.image, result.seed)
+
+        In the typical case of wanting to get just a single image, iterations
+        defaults to 1 and do:
+
+           output = next(txt2img.generate(prompt='banana sushi')
+
+        Pass None to get an infinite iterator.
+
+           outputs = txt2img.generate(prompt='banana sushi', iterations=None)
+           for o in outputs:
+               print(o.image, o.seed)
+        
+        '''
         model_name = self.params.model_name or self.model_manager.current_model
         model_info: dict = self.model_manager.get_model(model_name)
         model:StableDiffusionGeneratorPipeline = model_info['model']
@@ -149,8 +171,9 @@ class InvokeAIGenerator(metaclass=ABCMeta):
 
         generator_args = dataclasses.asdict(self.params)
         generator_args.update(keyword_args)
-        
+
-        while True:
+        iteration_count = range(iterations) if iterations else itertools.count(start=0, step=1)
+        for i in iteration_count:
             results = generator.generate(prompt,
                                          conditioning=(uc, c, extra_conditioning_info),
                                          sampler=scheduler,

invokeai/backend/model_management/model_manager.py

@@ -34,8 +34,7 @@ from picklescan.scanner import scan_file_path
 from invokeai.backend.globals import Globals, global_cache_dir
 
 from ..stable_diffusion import StableDiffusionGeneratorPipeline
-from ..util import CPU_DEVICE, ask_user, download_with_resume
+from ..util import CUDA_DEVICE, ask_user, download_with_resume
-
 
 class SDLegacyType(Enum):
     V1 = 1
@@ -51,23 +50,28 @@ VAE_TO_REPO_ID = {  # hack, see note in convert_and_import()
 }
 
 class ModelManager(object):
+    '''
+    Model manager handles loading, caching, importing, deleting, converting, and editing models.
+    '''
     def __init__(
         self,
-        config: OmegaConf,
+        config: OmegaConf|Path,
-        device_type: torch.device = CPU_DEVICE,
+        device_type: torch.device = CUDA_DEVICE,
         precision: str = "float16",
         max_loaded_models=DEFAULT_MAX_MODELS,
         sequential_offload=False,
     ):
         """
-        Initialize with the path to the models.yaml config file,
+        Initialize with the path to the models.yaml config file or
-        the torch device type, and precision. The optional
+        an initialized OmegaConf dictionary. Optional parameters
-        min_avail_mem argument specifies how much unused system
+        are the torch device type, precision, max_loaded_models,
-        (CPU) memory to preserve. The cache of models in RAM will
+        and sequential_offload boolean. Note that the default device 
-        grow until this value is approached. Default is 2G.
+        type and precision are set up for a CUDA system running at half precision.
         """
         # prevent nasty-looking CLIP log message
         transformers.logging.set_verbosity_error()
+        if not isinstance(config, DictConfig):
+            config = OmegaConf.load(config)
         self.config = config
         self.precision = precision
         self.device = torch.device(device_type)
@@ -557,7 +561,7 @@ class ModelManager(object):
         """
         model_name = model_name or Path(repo_or_path).stem
         model_description = (
-            model_description or f"Imported diffusers model {model_name}"
+            description or f"Imported diffusers model {model_name}"
         )
         new_config = dict(
             description=model_description,