Remove LoRA support

.gitignore

@@ -1,4 +1,3 @@
 __pycache__/
 .venv/
-loras/
 app.log

DOCS.md

@@ -32,15 +32,6 @@ Each model checkpoint has a different aesthetic:
 * [SG161222/Realistic_Vision_V5](https://huggingface.co/SG161222/Realistic_Vision_V5.1_noVAE): realistic
 * [XpucT/Deliberate_v6](https://huggingface.co/XpucT/Deliberate): general purpose stylized
 
-### LoRA
-
-Apply up to 2 LoRA (low-rank adaptation) adapters with adjustable strength:
-
-* [Perfection Style](https://civitai.com/models/411088?modelVersionId=486099): attempts to improve aesthetics, use high strength
-* [Detailed Style](https://civitai.com/models/421162?modelVersionId=486110): attempts to improve details, use low strength
-
-> NB: The trigger words are automatically appended to the positive prompt for you.
-
 ### Styles
 
 [Styles](https://huggingface.co/spaces/adamelliotfields/diffusion/blob/main/data/styles.json) are prompt templates that wrap your positive and negative prompts. They were originally derived from the [twri/sdxl_prompt_styler](https://github.com/twri/sdxl_prompt_styler) Comfy node, but have since been entirely rewritten.
@@ -55,7 +46,7 @@ The `Anime: *` styles work the best with Dreamshaper. When using the anime-speci
 * Guidance: `10`
 * Steps: `50`
 
-You subject should be a few simple tokens like `girl, brunette, blue eyes, armor, nebula, celestial`. Experiment with `Clip Skip` and `Karras`. Finish with the `Perfection Style` LoRA on a moderate setting and upscale. 
+You subject should be a few simple tokens like `girl, brunette, blue eyes, armor, nebula, celestial`. Experiment with `Clip Skip` and `Karras`. 
 
 ### Scale
 

README.md

@@ -1,7 +1,7 @@
 ---
 # https://huggingface.co/docs/hub/en/spaces-config-reference
-title: Diffusion Zero
-short_description: Image generation studio on ZeroGPU
+title: Diffusion
+short_description: Image generation studio for SD 1.5
 emoji: 🧨
 colorFrom: purple
 colorTo: blue
@@ -62,7 +62,6 @@ preload_from_hub:  # up to 10
 Gradio app for Stable Diffusion 1.5 featuring:
 * txt2img and img2img pipelines with ControlNet and IP-Adapter
 * Canny edge detection (more preprocessors coming soon)
-* Curated models, LoRAs, and TI embeddings
 * Compel prompt weighting
 * Hand-written style templates
 * Multiple samplers with Karras scheduling

app.py

@@ -1,6 +1,5 @@
 import argparse
 import json
-import os
 import random
 
 import gradio as gr
@@ -9,7 +8,6 @@ from lib import (
     Config,
     async_call,
     disable_progress_bars,
-    download_civit_file,
     download_repo_files,
     generate,
     read_file,
@@ -215,41 +213,6 @@ with gr.Blocks(
                     label="Scheduler",
                     filterable=False,
                 )
-            with gr.Row():
-                with gr.Group(elem_classes=["gap-0"]):
-                    lora_1 = gr.Dropdown(
-                        min_width=240,
-                        label="LoRA #1",
-                        value="none",
-                        choices=[("None", "none")]
-                        + [
-                            (lora["name"], lora_id) for lora_id, lora in Config.CIVIT_LORAS.items()
-                        ],
-                    )
-                    lora_1_weight = gr.Slider(
-                        value=0.0,
-                        minimum=0.0,
-                        maximum=1.0,
-                        step=0.1,
-                        show_label=False,
-                    )
-                with gr.Group(elem_classes=["gap-0"]):
-                    lora_2 = gr.Dropdown(
-                        min_width=240,
-                        label="LoRA #2",
-                        value="none",
-                        choices=[("None", "none")]
-                        + [
-                            (lora["name"], lora_id) for lora_id, lora in Config.CIVIT_LORAS.items()
-                        ],
-                    )
-                    lora_2_weight = gr.Slider(
-                        value=0.0,
-                        minimum=0.0,
-                        maximum=1.0,
-                        step=0.1,
-                        show_label=False,
-                    )
 
             # Generation settings
             gr.HTML("<h3>Generation</h3>")
@@ -479,10 +442,6 @@ with gr.Blocks(
             image_prompt,
             control_image_prompt,
             ip_image_prompt,
-            lora_1,
-            lora_1_weight,
-            lora_2,
-            lora_2_weight,
             style,
             seed,
             model,
@@ -518,16 +477,6 @@ if __name__ == "__main__":
     for repo_id, allow_patterns in Config.HF_MODELS.items():
         download_repo_files(repo_id, allow_patterns, token=Config.HF_TOKEN)
 
-    # download civit loras
-    for lora_id, lora in Config.CIVIT_LORAS.items():
-        file_path = os.path.join(os.path.dirname(__file__), "loras")
-        download_civit_file(
-            lora_id,
-            lora["model_version_id"],
-            file_path=file_path,
-            token=Config.CIVIT_TOKEN,
-        )
-
     # https://www.gradio.app/docs/gradio/interface#interface-queue
     demo.queue(default_concurrency_limit=1).launch(
         server_name=args.server,

lib/__init__.py

@@ -3,7 +3,6 @@ from .inference import generate
 from .utils import (
     async_call,
     disable_progress_bars,
-    download_civit_file,
     download_repo_files,
     read_file,
 )
@@ -12,7 +11,6 @@ __all__ = [
     "Config",
     "async_call",
     "disable_progress_bars",
-    "download_civit_file",
     "download_repo_files",
     "generate",
     "read_file",

lib/config.py

@@ -57,7 +57,6 @@ sd_files = [
 # Using namespace instead of dataclass for simplicity
 Config = SimpleNamespace(
     HF_TOKEN=os.environ.get("HF_TOKEN", None),
-    CIVIT_TOKEN=os.environ.get("CIVIT_TOKEN", None),
     ZERO_GPU=import_module("spaces").config.Config.zero_gpu,
     # TODO: fix model config redundancy
     HF_MODELS={
@@ -74,22 +73,6 @@ Config = SimpleNamespace(
         "SG161222/Realistic_Vision_V5.1_noVAE": ["Realistic_Vision_V5.1_fp16-no-ema.safetensors"],
         "XpucT/Deliberate": ["Deliberate_v6.safetensors"],
     },
-    CIVIT_LORAS={
-        # https://civitai.com/models/411088?modelVersionId=486099
-        "perfection_style": {
-            "model_id": "411088",
-            "model_version_id": "486099",
-            "name": "Perfection Style",
-            "trigger": "perfection style",
-        },
-        # https://civitai.com/models/421162?modelVersionId=486110
-        "detailed_style": {
-            "model_id": "421162",
-            "model_version_id": "486110",
-            "name": "Detailed Style",
-            "trigger": "detailed style",
-        },
-    },
     MONO_FONTS=["monospace"],
     SANS_FONTS=[
         "sans-serif",

lib/inference.py

@@ -66,10 +66,6 @@ def generate(
     image_prompt=None,
     control_image_prompt=None,
     ip_image_prompt=None,
-    lora_1=None,
-    lora_1_weight=0.0,
-    lora_2=None,
-    lora_2_weight=0.0,
     style=None,
     seed=None,
     model="Lykon/dreamshaper-8",
@@ -159,40 +155,6 @@ def generate(
     pipe = loader.pipe
     upscaler = loader.upscaler
 
-    # load loras
-    loras = []
-    weights = []
-    loras_and_weights = [(lora_1, lora_1_weight), (lora_2, lora_2_weight)]
-    loras_dir = os.path.abspath(os.path.join(os.path.dirname(__file__), "..", "loras"))
-    total_loras = sum(1 for lora, _ in loras_and_weights if lora and lora.lower() != "none")
-    desc_loras = "Loading LoRAs"
-    if total_loras > 0:
-        with timer(f"Loading {total_loras} LoRA{'s' if total_loras > 1 else ''}"):
-            safe_progress(progress, 0, total_loras, desc_loras)
-            for i, (lora, weight) in enumerate(loras_and_weights):
-                if lora and lora.lower() != "none" and lora not in loras:
-                    config = Config.CIVIT_LORAS.get(lora)
-                    if config:
-                        try:
-                            pipe.load_lora_weights(
-                                loras_dir,
-                                adapter_name=lora,
-                                weight_name=f"{lora}.{config['model_version_id']}.safetensors",
-                            )
-                            weights.append(weight)
-                            loras.append(lora)
-                            safe_progress(progress, i + 1, total_loras, desc_loras)
-                        except Exception:
-                            raise Error(f"Error loading {config['name']} LoRA")
-
-    # unload after generating or if there was an error
-    try:
-        if loras:
-            pipe.set_adapters(loras, adapter_weights=weights)
-    except Exception:
-        pipe.unload_lora_weights()
-        raise Error("Error setting LoRA weights")
-
     # Load negative embedding if requested
     if negative_embedding:
         embeddings_dir = os.path.abspath(
@@ -234,9 +196,6 @@ def generate(
             if negative_embedding:
                 negative_styled += f", <{Config.NEGATIVE_EMBEDDING}>"
 
-            for lora in loras:
-                positive_styled += f", {Config.CIVIT_LORAS[lora]['trigger']}"
-
             positive_embeds, negative_embeds = compel.pad_conditioning_tensors_to_same_length(
                 [compel(positive_styled), compel(negative_styled)]
             )
@@ -278,8 +237,6 @@ def generate(
         finally:
             if negative_embedding:
                 pipe.unload_textual_inversion()
-            if loras:
-                pipe.unload_lora_weights()
             CURRENT_STEP = 0
             CURRENT_IMAGE += 1
 

lib/utils.py

@@ -7,7 +7,6 @@ from contextlib import contextmanager
 from typing import Callable, Tuple, TypeVar
 
 import anyio
-import httpx
 import numpy as np
 import torch
 from anyio import Semaphore
@@ -19,7 +18,6 @@ from transformers import logging as transformers_logging
 from typing_extensions import ParamSpec
 
 from .annotators import CannyAnnotator
-from .logger import Logger
 
 T = TypeVar("T")
 P = ParamSpec("P")
@@ -91,34 +89,6 @@ def download_repo_files(repo_id, allow_patterns, token=None):
     return snapshot_path
 
 
-def download_civit_file(lora_id, version_id, file_path=".", token=None):
-    base_url = "https://civitai.com/api/download/models"
-    file = f"{file_path}/{lora_id}.{version_id}.safetensors"
-    log = Logger("download_civit_file")
-
-    if os.path.exists(file):
-        return
-
-    try:
-        params = {"token": token}
-        response = httpx.get(
-            f"{base_url}/{version_id}",
-            timeout=None,
-            params=params,
-            follow_redirects=True,
-        )
-
-        response.raise_for_status()
-        os.makedirs(file_path, exist_ok=True)
-
-        with open(file, "wb") as f:
-            f.write(response.content)
-    except httpx.HTTPStatusError as e:
-        log.error(f"{e.response.status_code} {e.response.text}")
-    except httpx.RequestError as e:
-        log.error(f"RequestError: {e}")
-
-
 def image_to_pil(image: Image.Image):
     """Converts various image inputs to RGB PIL Image."""
     if isinstance(image, str) and os.path.isfile(image):

partials/intro.html

@@ -1,13 +1,13 @@
 <div id="intro">
   <div>
     <h1>
-      Diffusion&nbsp;<span>ZERO</span>
+      Diffusion
     </h1>
     <svg xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 15 15">
       <path d="M7.48877 6.75C7.29015 6.75 7.09967 6.82902 6.95923 6.96967C6.81879 7.11032 6.73989 7.30109 6.73989 7.5C6.73989 7.69891 6.81879 7.88968 6.95923 8.03033C7.09967 8.17098 7.29015 8.25 7.48877 8.25C7.68738 8.25 7.87786 8.17098 8.0183 8.03033C8.15874 7.88968 8.23764 7.69891 8.23764 7.5C8.23764 7.30109 8.15874 7.11032 8.0183 6.96967C7.87786 6.82902 7.68738 6.75 7.48877 6.75ZM7.8632 0C11.2331 0 11.3155 2.6775 9.54818 3.5625C8.80679 3.93 8.47728 4.7175 8.335 5.415C8.69446 5.565 9.00899 5.7975 9.24863 6.0975C12.0195 4.5975 15 5.19 15 7.875C15 11.25 12.3265 11.325 11.4428 9.5475C11.0684 8.805 10.2746 8.475 9.57813 8.3325C9.42836 8.6925 9.19621 9 8.89665 9.255C10.3869 12.0225 9.79531 15 7.11433 15C3.74438 15 3.67698 12.315 5.44433 11.43C6.17823 11.0625 6.50774 10.2825 6.65751 9.5925C6.29056 9.4425 5.96855 9.2025 5.72891 8.9025C2.96555 10.3875 0 9.8025 0 7.125C0 3.75 2.666 3.6675 3.54967 5.445C3.92411 6.1875 4.71043 6.51 5.40689 6.6525C5.54918 6.2925 5.78882 5.9775 6.09586 5.7375C4.60559 2.97 5.1972 0 7.8632 0Z"></path>
     </svg>
   </div>
   <p>
-    Stable Diffusion on ZeroGPU.
+    Image generation studio for Stable Diffusion.
   </p>
 </div>