A SillyTavern extension for automatic AI image generation during roleplay. Detects [pic prompt="..."] tags in LLM responses and generates images inline using your configured image generation API.
-
Auto Image Detection - Automatically finds
[pic prompt="..."]tags in AI responses and generates images -
Prompt Presets - Save and switch between different prompt injection presets
-
Prompt Injection Positions - System, User, AI, Append to User Message, or Macro Mode
-
Macro Mode - Use
{{IkarusAutoImage-prompt}}anywhere in your own prompt preset -
Auto-Fix Pic Format - Fixes malformed
pic prompt="..."formats before extraction -
Optional
/sdFiltering - Can also filter native/sd,/imagine, and/imgprompts -
Character Prompts - Per-character prompt slots (5 per card) that append to image generation prompts
-
Tag Replacements - Trigger-based find-and-replace for image prompts with parent/child hierarchy support

-
First Full, Rest Dedupe - First match gets the full replacement, later matches can use a shorter tag
-
Prompt Filters - Conditional actions (remove, append, replace) on generated prompts
-
Global/Character Scope - Replacements and filters can be global or locked to specific character cards

-
Global Replacement Manager - Popup window with folder/category organization, search, and full CRUD

-
Double Cleaner - Removes duplicate tags after processing
-
Stable Character Data - Uses avatar filenames as keys, not numeric IDs, so data survives reordering
- Open SillyTavern
- Go to Extensions -> Install Extension
- Paste this repo URL:
https://github.com/IkarusV/IkarusAutoImage - Click Install
- Clone or download this repo into your SillyTavern
data/default-user/extensions/folder:cd <SillyTavern>/data/default-user/extensions/ git clone https://github.com/IkarusV/IkarusAutoImage.git
- Restart SillyTavern
- Configure your image generation API in SillyTavern (e.g., Stable Diffusion, NovelAI)
- Open the Ikarus Auto Image panel in the Extensions sidebar
- Set up your prompt injection template
- The extension will automatically detect and render images during chat
Create rules that swap character names for danbooru tags:
- Trigger:
Bloom, Dragon fire Fairy, Bloom winx - Replace:
<lora:AnimaBloom:1>Bloom, fairy wings, red hair
Add per-character style instructions, danbooru tag lists, or generation rules that only apply to specific character cards.
Click Manage to open the folder/category organizer for your global replacements. Create categories (e.g., "Cartoon", "Video Games") and folders (e.g., "Winx Club", "Nier Automata") to keep things tidy.
Added a standalone mode that spawn a pop up window that can be opened and closed to auto generate image or even ask your ai to generate more and save by chat ! but also saved in your galery in case.

MIT