negaga53

negaga53 / neme-anima

Public

3-step all-in-one LoRA builder for Anima (extract -> tag -> train)

43
6
100% credibility
Found May 05, 2026 at 43 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

Neme-Anima extracts character images from anime videos using reference portraits, auto-tags them, and trains custom LoRA models for AI image generation.

How It Works

1
🔍 Discover Neme-Anima

You find a helpful tool that turns your favorite anime videos into custom images for training a personal AI artist.

2
📦 Easy one-click setup

Run a simple script that installs everything you need and opens a friendly web screen.

3
✨ Create your character project

Name your project and pick a folder to store your custom images.

4
📹 Add videos and portraits

Drag in anime episodes and reference pictures of your character to guide the magic.

5
🎯 Extract perfect crops

Watch as it automatically finds, crops, and organizes hundreds of clear images of your character from the videos.

6
🏷️ Review and label images

Browse the images, tweak labels, and delete any you don't like with simple clicks.

7
🚀 Train your custom model

Hit start to teach an AI your character's unique look using ready-to-go settings.

🎉 Your AI character is ready

Enjoy generating endless new images of your character in any pose or scene!

Sign up to see the full architecture

6 more

Sign Up Free

Star Growth

See how this repo grew from 43 to 43 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is neme-anima?

Neme-anima is a 3-step all-in-one LoRA builder that extracts character crops from anime videos using reference images, auto-tags them with WD14 danbooru tags plus optional LLM captions, and trains character-specific LoRAs on the Anima model. Built in Python with a Svelte web UI and FastAPI backend, it splits videos into shots, detects and tracks people, matches to refs via CCIP embeddings, and outputs 1024px crops ready for kohya-ss or OneTrainer on SDXL anime bases like Pony or Illustrious. A one-click bash script handles install, frontend build, and 14GB Anima weights download.

Why is it gaining traction?

It collapses the tedious extract-tag-train pipeline into a single UI with smart caching—reruns skip slow detection/tracking—and multi-character support, letting you route frames across chars or duplicate for shared poses. Bulk regex tag edits, per-frame recrops, and core-tag pruning (dropping constant traits like hair color) cut noise without CLI hassle. The queue pauses before tagging for curation, and training integrates diffusion-pipe with resume/checkpoints.

Who should use this?

Anime AI enthusiasts training character LoRAs from episodic videos, like ripping K-On or Persona casts for Pony/Illustrious fine-tunes. Stable Diffusion tinkerers who hate manual frame-picking and tagging, especially with multiple chars per source.

Verdict

Worth a spin for anime LoRA workflows—polished UI, CLI fallback, and one-click setup punch above 43 stars and 1.0% credibility. Still early: light tests, niche scope. Fork-friendly if you hack Anima trainers.

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.