Instructions to use peculiar/sbgrid with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Diffusers
How to use peculiar/sbgrid with Diffusers:
pip install -U diffusers transformers accelerate
import torch from diffusers import DiffusionPipeline # switch to "mps" for apple devices pipe = DiffusionPipeline.from_pretrained("krea/Krea-2-Raw", dtype=torch.bfloat16, device_map="cuda") pipe.load_lora_weights("peculiar/sbgrid") prompt = "-" image = pipe(prompt).images[0] - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- Draw Things
- DiffusionBee
Krea2_sbgrid

- Prompt
- -

- Prompt
- -

- Prompt
- -

- Prompt
- -
Model description
Cinematic Storyboard Generator LoRA
A LoRA trained to generate coherent 8-panel cinematic storyboards from character, environment, and visual references.
The model is designed to understand sequential visual storytelling rather than simply reproducing a static storyboard layout. It generates panels with deliberate changes in camera framing, perspective, composition, character performance, action, and environmental context to create a visual progression from beginning to end.
Key Features
- 8-panel cinematic storyboard generation
- Sequential visual storytelling
- Strong shot-to-shot continuity
- Character consistency across panels
- Environment and spatial continuity
- Cinematic camera language
- Wide, medium, close-up, POV, over-the-shoulder, low-angle, high-angle and other varied shot types
- Action progression and visual pacing
- Character interaction and performance
- Establishing shots and environmental storytelling
- Visual escalation, climax and payoff
- Professional production-storyboard composition
Style-Aware Generation
This LoRA is not restricted to a single artistic style.
It is designed to work alongside a supplied visual reference and preserve the reference's fundamental visual language.
For example:
- Photorealistic reference β photorealistic storyboard
- Anime reference β anime storyboard
- Manga reference β manga-style storyboard
- 3D/CGI reference β 3D/CGI storyboard
- Painterly reference β painterly storyboard
- Stylized concept art β matching stylized storyboard
- Other distinctive visual styles β preserve the supplied visual aesthetic
The objective is to make the storyboard structure and cinematic visual storytelling consistent while allowing the source reference to determine the visual style.
Training Concept
The dataset consists of diverse cinematic storyboard sequences covering different characters, environments, genres, actions, compositions and visual styles.
The training focuses on learning:
story progression + cinematic shot design + panel composition + continuity
rather than memorizing specific characters, locations or stories.
Recommended Trigger
`STORYBOARD8`
Example concept:
`STORYBOARD8, eight-panel cinematic storyboard of a woman entering an abandoned hotel and discovering a mysterious figure`
The LoRA is intended to be used with a compatible image-generation model capable of following reference images and maintaining visual consistency.
Intended Use
Ideal for:
- Film pre-visualization
- Music video storyboards
- Animation planning
- Short-film development
- Video-generation planning
- Character action sequences
- Cinematic concept development
- Visual storytelling
- Shot planning
- AI video prompt development
This LoRA is intended to turn a character + environment + story concept into a visually directed sequence of cinematic shots, rather than simply producing eight variations of the same image.
Trigger words
You should use sbgrid to trigger the image generation.
You should use eight panel storyboard to trigger the image generation.
You should use 6 panel storyboard to trigger the image generation.
Download model
Download them in the Files & versions tab.
- Downloads last month
- -
Model tree for peculiar/sbgrid
Base model
krea/Krea-2-Raw