story-shorts-character-lora

Story Shorts Character LoRA

A collection of LoRA models for generating consistent flat-style characters for AI-narrated story short-form videos.

Characters are designed with:

• Simple, clean linework — minimal detail, easy to animate or sequence
• Flat vector-like illustration style — bold colors, no complex shading
• High consistency across expressions and poses — suitable for multi-scene storytelling

Intended Use

These models are built to support short-form video production where a narrator tells a story over a series of generated images. Each LoRA trains a specific character so that it remains visually coherent across different scenes, emotions, and backgrounds.

Workflow:

1. Generate training images via Runware AI
2. Train LoRA on character-specific image set
3. Use LoRA at inference time to keep character consistent across scenes

Models

Usage

# Example using diffusers
from diffusers import StableDiffusionPipeline
import torch

pipe = StableDiffusionPipeline.from_pretrained("base-model-id", torch_dtype=torch.float16)
pipe.load_lora_weights("baleenme/story-shorts-character-lora", weight_name="character_name.safetensors")

prompt = "TRIGGER_WORD, flat illustration, simple linework, white background, smiling"
image = pipe(prompt).images[0]

Training Details

• Image style: Flat vector illustration, minimal linework
• Training tool: Runware AI
• Image count per character: ~20–50 curated images
• Base model: (varies per character, listed per model)

License

Apache 2.0 — free to use for personal and commercial projects.