Thinking in Boxes: 3D Editing in Real Images Made Easy
Paper • 2606.20556 • Published
import torch
from diffusers import DiffusionPipeline
from diffusers.utils import load_image
# switch to "mps" for apple devices
pipe = DiffusionPipeline.from_pretrained("pradhaansbhat/Thinking-In-Boxes", dtype=torch.bfloat16, device_map="cuda")
prompt = "Turn this cat into a dog"
input_image = load_image("https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/cat.png")
image = pipe(image=input_image, prompt=prompt).images[0][NeurIPS-2026] Thinking in Boxes: 3D Editing in Real Images Made Easy
Pradhaan S Bhat1∗ · Naveen Chandra R1∗ · Rishubh Parihar1 · Vaibhav Vavilala2 · R. Venkatesh Babu1 · D.A. Forsyth · Anand Bhattad4
1 Indian Institute of Science 2 Apple 3 UIUC 4 Johns Hopkins University
∗ Equal Contribution
This is the trained LoRA used in the paper Thinking In Boxes: 3D Editing in Real Images Made Easy.
Thinking-In-Boxes is an Image-to-Image Generative LoRA for black-forest-labs/FLUX.1-Kontext-dev trained for the task of Geometric Image Editing.
This LoRA is trained on the Thinking-In-Boxes dataset available here. Details of training are available in the supplementary section of the paper.
Usage with Diffusers 🧨
Please refer to the GitHub Repository on setup, inference and training.
Citation
If you find our work useful, please consider citing:
@misc{bhat2026thinkingboxes3dediting,
title = {Thinking in Boxes: 3D Editing in Real Images Made Easy},
author = {Pradhaan S Bhat and Naveen Chandra R and Rishubh Parihar and Vaibhav Vavilala and R. Venkatesh Babu and D. A. Forsyth and Anand Bhattad},
year = {2026},
eprint = {2606.20556},
archivePrefix = {arXiv},
primaryClass = {cs.CV},
url = {https://arxiv.org/abs/2606.20556}
}
- Downloads last month
- -
Model tree for pradhaansbhat/Thinking-In-Boxes
Base model
black-forest-labs/FLUX.1-Kontext-dev