tina390/SceneBloom-AssetLoRAs

SceneBloom Asset LoRAs

Paper | Code

The asset LoRA adapters of SceneBloom, an indoor scene generator for embodied learning. Each is a LoRA adapter of FLUX.1 [dev] for one mesh in SceneBloom-Assets. The asset library is extensible, and we will release more adapters here.

Contents

Path	Description
adapters/	One adapter per mesh, covering different furniture categories. Each is in a folder named <mesh>__<trigger>, for example bed__yte/.
manifest.tsv	Category, mesh and trigger word of every adapter.

Trigger words

Each adapter has its own trigger word, given in the trigger column of manifest.tsv. Prompt with a photo of <trigger> <category>, for example a photo of yte bed.

Usage

The SceneBloom code downloads this repository with scripts/download/download_assets.sh and loads the adapters of each zone together. To try one adapter on its own:

import torch
from diffusers import FluxPipeline

pipe = FluxPipeline.from_pretrained("black-forest-labs/FLUX.1-dev",
                                    torch_dtype=torch.bfloat16).to("cuda")
pipe.load_lora_weights("tina390/SceneBloom-AssetLoRAs", revision="v1.0",
                       weight_name="adapters/bed__yte/pytorch_lora_weights.safetensors")
image = pipe("a photo of yte bed", num_inference_steps=28).images[0]

License

The adapters are Derivatives of the FLUX.1 [dev] Model and are distributed under the FLUX.1 [dev] Non-Commercial License v1.1.1 (LICENSE.md). Any rights to use the FLUX.1 [dev] Model and these adapters are granted to you directly by Black Forest Labs Inc. under that license, which permits non-commercial use only and requires content filtering or review of generated images. The adapters are not an official product of Black Forest Labs Inc. and are not endorsed, approved or validated by it.

The adapters adapters/*_art*__*/ were trained on renders of ArtVIP meshes, licensed under Apache-2.0 (licenses/ArtVIP/).

Attribution notice:

The FLUX.1 [dev] Model is licensed by Black Forest Labs Inc. under the FLUX.1 [dev] Non-Commercial License. Copyright Black Forest Labs Inc.

IN NO EVENT SHALL BLACK FOREST LABS INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH USE OF THIS MODEL.

We modified the FLUX.1 [dev] Model by training the LoRA adapters in this repository.

Citation

@article{liu2026scenebloom,
  title   = {SceneBloom: Agentic Indoor Environment Generation with Realistic and Diverse Layouts for Embodied Learning},
  author  = {Liu, Yuhan and Qin, Lianhui and Wan, Shenji},
  journal = {arXiv preprint arXiv:2610.32978},
  year    = {2026}
}