LlamaCaption2 is a GUI-based image captioning tool that uses local AI models to generate customized captions for images.
#Originally : LlamaScribe by LamEmy |
Revised & Upgraded by AngryHamster
Originally developed by https://civarchive.com/user/LamEmy
revised and upgraded by AngryHamster
Github Page for the LlamaCaption2 :
git clone github.com/robertinosro/LlamaCaption2_AngryHamster
## Features
- Process multiple images in a batch
- Generate AI-powered image descriptions
- Apply custom formatting to captions
- User-friendly interface
- Works with local Ollama models
-Process at once all images in the input folder!
Quick Readme :Setup Requirements
Install Ollama on your system (no Python installation required)
[Ollama](https://ollama.ai/) installed and running locally
(https://ollama.com/library/llama3.2-vision)
- A compatible vision model (like llama3.2-visioon) for image analysis
- A compatible text model (like qwen) for caption refinement
Find the LlamaScribe.exe executable inside the main folder. Start the file. Enjoy captioning!
Getting Started
Launch LlamaScribe.exe
The application will open with the title "LlmamaScribe by LamEmy | Revised & Upgraded by AngryHamster (Civitai)"
Basic Usage
Main Tab:
Select the folder containing your images
Choose your preferred Ollama vision model
Click "Start Processing" to begin captioning your images
Advanced Tab:
Configure system prompts
Adjust additional settings for the AI models
Formatting Tab:
Enable/disable the caption prefix and suffix
Customize the prefix text: "A photo of a woman, bloobikkx1, curvy blonde with (a well-defined neck:1.3) and (natural proportions:1.2), "
Customize the suffix text: " (masterpiece, ultra-realistic, high-definition, 8K, cinematic lighting),(professional photography:1.4), (sharp focus:1.2), (studio lighting:1.2), (clear details:1.3), (professional atmosphere:1.3)"
Output
The application will process all images in the selected folder
Each image will receive a caption following the format:
[Prefix] + [AI-generated description] + [Suffix]
Captions will be saved alongside your images
A more advanced Readme file is included in the main folder!
Description
V 2.0 Upgraded! Adding Custom Prefix/Suffix Text to tailor captions. (trigger words for loras)
Comments (4)
what does this do over something like JoyCaption?
First of all , uyou don't need to install any requirements. i've been strugling with JoyCaption for a long time,and everytime i ran into a black hole. Install this and that , then this is not compatible or missing..all that. The GUI itself is simple and intuitive, and i must say it works really great ( and fast too) The purpose of this is mostly for generating captions for Lora training.
It runs locally with ollama and any vision model of your choice.
Besides the initial image captioning i added the option for Prefix and Suffix. You can customize both the prefix and suffix directly in the application without modifying the source code.
Also there is an Advanced Tab (optional and for advanced users)where you can enter a Custom Prompt for the refiner model.
The repo contains all the info needed to run this Image captioner.
It does not need any requirements to be installed ( unless you want to re-build it and customize on your own)
The changes you made look good, thank you for uploading it. :)
Thanks for looking into it!