update v1.1

- Added prompt generation modes - Added face swap support using ReActor extension - Added new configuration options for FaceSwapLab extension - Added debug mode and changed various messages to only log when it's enabled - Updated readme to add better installation instructions - Renamed some configuration options to better reflect what they do - Fixed FaceSwapLab settings never being actually used (except for source face) - Fixed various options being defined as integer instead of float such as cfg scale or upscaler scale - Fixed special html characters rendering incorrectly such as apostrophes - Disabled image generation if not used in chat mode
Trojaner · Dec 17, 2023 · 58d2b4b · 58d2b4b
1 parent 8928b9e
commit 58d2b4b
Show file tree

Hide file tree

Showing 11 changed files with 636 additions and 146 deletions.
diff --git a/README.md b/README.md
@@ -19,12 +19,14 @@ Demo:
 
 ## Supported Stable Diffusion WebUI Extensions
 - [FaceSwapLab](https://github.com/glucauze/sd-webui-faceswaplab)
+- [ReActor](https://github.com/Gourieff/sd-webui-reactor)
 
 ## Installation
-- Open a shell inside the text-generation-webui root directory. 
+- Open a shell with cmd_linux.sh/cmd_macos.sh/cmd_windows.bat inside text-generation-webui folder. 
 - Run `git clone https://github.com/Trojaner/text-generation-webui-stable_diffusion extensions/stable_diffusion`.
-- Open the [settings.debug.yaml](https://github.com/Trojaner/text-generation-webui-stable_diffusion/blob/main/settings.debug.yaml) file, copy the extension related settings to your own settings.json and adjust them to your own preferences.
-- Add `stable_diffusion` to the enabled extensions.
+- Run `pip install -r extensions/stable_diffusion/requirements.txt` to install the required dependencies.	
+- Open the [settings.debug.yaml](https://github.com/Trojaner/text-generation-webui-stable_diffusion/blob/main/settings.debug.yaml) file, copy the extension related settings to your own settings.json in the text-generation-webui directory.
+- Add `stable_diffusion` to the enabled extensions in settings.json.
 
 > **Note**
 > If you install this extension manually, make sure the extension directory is called stable_diffusion. 
@@ -54,17 +56,16 @@ text-generation-webui, Visual Studio Code and Python 3.10 are required for devel
 - Last but not least, ensure that you do not accidentally commit changes you might have made to the `settings.debug.yaml` or `launch.json` files unless intentional.
 
 ## Ethical Guidelines
-This extension integrates with the FaceSwapLab extension for stable-diffusion-webui and hence allows to swap faces in the generated images. This extension is not intended to for the creation of non-consensual deepfake content. Please use this extension responsibly and do not use it to create such content. The main purpose of the face swapping functionality is to allow the creation of consistent images of text-generation-webui characters. If you are unsure whether your use case is ethical, please refrain from using this extension.
+This extension integrates with various face swap extensions for stable-diffusion-webui and hence allows to swap faces in the generated images. This extension is not intended to for the creation of non-consensual deepfake content. Please use this extension responsibly and do not use it to create such content. The main purpose of the face swapping functionality is to allow the creation of consistent images of text-generation-webui characters. If you are unsure whether your use case is ethical, please refrain from using this extension.
 
 The maintainers and contributors of this extension cannot be held liable for any misuse of this extension but will try to prevent such misuse by all means.
 
 ## Todo
-- Some basic Gradio UI for fine-tuning the extension parameters at runtime
+- ~~Some basic Gradio UI for fine-tuning the extension parameters at runtime~~
+- ~~Support [ReActor](https://github.com/Gourieff/sd-webui-reactor) as alternative faceswap integration [[api implementation](https://github.com/Gourieff/sd-webui-reactor/blob/main/scripts/reactor_api.py)]~~ 
 - Character specific parameters
 - Custom LogitsProcessor or grammar implementation for generating proper and weighted SD image generation prompts
-- Support [ReActor](https://github.com/Gourieff/sd-webui-reactor) as alternative faceswap integration [[api implementation](https://github.com/Gourieff/sd-webui-reactor/blob/main/scripts/reactor_api.py)] 
 - Test with vanilla AUTOMATIC1111
-- Standalone mode using diffusers and without stable-diffusion-webui
 - Integrate with other SD extensions / scripts?
 
 ## See also

diff --git a/assets/demo1.png b/assets/demo1.png
diff --git a/assets/demo2.png b/assets/demo2.png
diff --git a/ext_modules/image_generator.py b/ext_modules/image_generator.py
@@ -7,34 +7,39 @@
 from webuiapi import WebUIApiResult
 from modules.logging_colors import logger
 from ..context import GenerationContext
-from ..params import TriggerMode
+from ..params import (
+    ContinuousModePromptGenerationMode,
+    InteractiveModePromptGenerationMode,
+    TriggerMode,
+)
 from .vram_manager import VramReallocationTarget, attempt_vram_reallocation
 
 
 def normalize_prompt(prompt: str) -> str:
-    if prompt is None or prompt.replace(",", "").replace("\n", "").strip() == "":
+    if prompt is None:
         return ""
 
-    return prompt.strip().strip(",").strip().strip("\n")
-
-
-def description_to_prompt(description: str) -> str:
-    return normalize_prompt(
-        description.replace("*", "")
+    return (
+        prompt.split("\n")[0]
+        .replace("*", "")
         .replace('"', "")
-        .replace(".", "")
-        .replace("!", "")
-        .replace("?", "")
-        .replace("#", "")
-        .replace(":", "")
-        .replace(";", "")
-        .split("\n")[0]
+        .replace(".", ",")
+        .replace("!", ",")
+        .replace("?", ",")
+        .replace(":", ",")
+        .replace(";", ",")
+        .strip()
         .strip("(")
         .strip(")")
+        .strip(",")
+        .strip()
+        .lower()
     )
 
 
-def generate_html_images_for_context(context: GenerationContext) -> str | None:
+def generate_html_images_for_context(
+    context: GenerationContext,
+) -> tuple[str | None, str, str]:
     """
     Generates images for the given context using Stable Diffusion
     and returns the result as HTML output
@@ -43,27 +48,41 @@ def generate_html_images_for_context(context: GenerationContext) -> str | None:
     attempt_vram_reallocation(VramReallocationTarget.STABLE_DIFFUSION, context)
     sd_client = context.sd_client
 
-    if context.params.trigger_mode == TriggerMode.INTERACTIVE:
-        full_prompt = (
-            description_to_prompt(context.output_text) + ","
-            if context.output_text
-            else ""
-        ) + normalize_prompt(context.params.base_prompt_suffix)
-    else:
-        full_prompt = (
-            normalize_prompt(context.params.default_prompt)
-            + ","
-            + normalize_prompt(context.params.base_prompt_suffix)
-        )
+    base_prompt = context.params.default_prompt
+
+    if context.params.trigger_mode == TriggerMode.INTERACTIVE and (
+        context.params.interactive_mode_prompt_generation_mode
+        == InteractiveModePromptGenerationMode.GENERATED_TEXT
+        or InteractiveModePromptGenerationMode.DYNAMIC
+    ):
+        base_prompt = context.output_text or ""
+
+    if context.params.trigger_mode == TriggerMode.CONTINUOUS and (
+        context.params.continuous_mode_prompt_generation_mode
+        == ContinuousModePromptGenerationMode.GENERATED_TEXT
+    ):
+        base_prompt = context.output_text or ""
+
+    base_prompt = normalize_prompt(base_prompt)
+
+    full_prompt = (
+        base_prompt
+        + (", " if base_prompt and base_prompt != "" else "")
+        + normalize_prompt(context.params.base_prompt_suffix)
+    )
 
     full_negative_prompt = normalize_prompt(context.params.base_negative_prompt)
 
-    logger.info(
-        (
-            "[SD WebUI Integration] Using stable-diffusion-webui to generate images.\n"
+    logger.info
+    (
+        "[SD WebUI Integration] Using stable-diffusion-webui to generate images."
+        + (
+            f"\n"
             f"  Prompt: {full_prompt}\n"
             f"  Negative Prompt: {full_negative_prompt}"
         )
+        if context.params.debug_mode_enabled
+        else ""
     )
 
     try:
@@ -89,23 +108,44 @@ def generate_html_images_for_context(context: GenerationContext) -> str | None:
 
         if len(response.images) == 0:
             logger.error("[SD WebUI Integration] Failed to generate any images.")
-            return None
+            return None, full_prompt, full_negative_prompt
 
         formatted_result = ""
         style = 'style="width: 100%; max-height: 100vh;"'
         for image in response.images:
             if context.params.faceswaplab_enabled:
-                logger.info("[SD WebUI Integration] Using FaceSwapLab to swap faces.")
+                if context.params.debug_mode_enabled:
+                    logger.info(
+                        "[SD WebUI Integration] Using FaceSwapLab to swap faces."
+                    )
 
                 try:
                     response = sd_client.faceswaplab_swap_face(
                         image,
-                        face=context.params.faceswaplab_source_face,
+                        params=context.params,
                         use_async=False,
                     )
                     image = response.image
                 except Exception as e:
-                    logger.error(f"[SD WebUI Integration] Failed to swap faces: {e}")
+                    logger.error(
+                        f"[SD WebUI Integration] FaceSwapLab failed to swap faces: {e}"
+                    )
+
+            if context.params.reactor_enabled:
+                if context.params.debug_mode_enabled:
+                    logger.info("[SD WebUI Integration] Using Reactor to swap faces.")
+
+                try:
+                    response = sd_client.reactor_swap_face(
+                        image,
+                        params=context.params,
+                        use_async=False,
+                    )
+                    image = response.image
+                except Exception as e:
+                    logger.error(
+                        f"[SD WebUI Integration] Reactor failed to swap faces: {e}"
+                    )
 
             if context.params.save_images:
                 character = (
@@ -140,4 +180,4 @@ def generate_html_images_for_context(context: GenerationContext) -> str | None:
     finally:
         attempt_vram_reallocation(VramReallocationTarget.LLM, context)
 
-    return formatted_result.rstrip("\n")
+    return formatted_result.rstrip("\n"), full_prompt, full_negative_prompt
diff --git a/ext_modules/text_analyzer.py b/ext_modules/text_analyzer.py
@@ -9,7 +9,7 @@ def try_get_description_prompt(
     Checks if the given message contains any triggers and returns the prompt if it does.
     """
 
-    trigger_regex = params.interactive_mode_trigger_regex
+    trigger_regex = params.interactive_mode_input_trigger_regex
     subject_regex = params.interactive_mode_subject_regex
     default_subject = params.interactive_mode_default_subject
     default_description_prompt = params.interactive_mode_description_prompt