Make it Home: Automatic Optimization of Furniture Arrangement

Lap-Fai Yu

Sai-Kit Yeung

Chi-Keung Tang

Demetri Terzopoulos

Tony F. Chan

Stanley Osher

Abstract This paper presents a fully-automatic system for generating an optimized indoor scene populated by a variety of furniture objects. Given positive examples of furnished indoor scenes, our system extracts hierarchical and spatial relationships for different types of furniture objects. This step is done once, in advance. The extracted relationships are encoded into priors which are integrated into a cost function that optimizes ergonomic factors, such as visibility and accessibility. To deal with the prohibitively large search space, the cost function is optimized by simulated annealing with a Metropolis Hastings state-search step. We demonstrate that different furniture layouts can be automatically synthesized to decorate an indoor scene. A perceptual study is performed to validate that there is no significant difference in preference on functionality between our synthesized results and those produced by human designers.

Keywords: interior generation, interior modeling, spatial allocation, virtual reality

Publications:

Acknowledgements:

We are grateful to anonymous reviewers for their constructive comments. We also thank Google 3D Warehouse for providing a rich source of 3D models which help tremendously in speeding up the modeling process. Special thanks to Shawn Singh for his professional advice and efforts on video-editing, scene-modeling and rendering; Yu-Wing Tai for his helpful suggestion on the draft of this paper and experiments; Michael S. Brown for narrating the video; Howard Alexander Greene for advice on video-editing; Jan Adamec for providing a modified version of his Room Arranger software which allows our early testing; Wenze Hu for technical advice on stochastic optimization; Lap-Fai Lee for advice on data analysis of the perceptual study. This research was partially supported by the Hong Kong Research Grant Council under grant number 620309, RICE/MURI Award 443948-SN-80050, NSF 443948- TH-22487 and ONR N00014-09-1-0105. Lap-Fai Yu is supported by the Sir Edward Youde Memorial Fellowship.

Results:

While in recent years numerous publications have appeared demonstrating the automatic modeling of building exteriors and facades, the automatic generation of realistic indoor configurations has not yet received the attention that it deserves. With the growing popularity of social virtual worlds and massively-multiplayer online games that feature a large amount of realistic environmental content, automatic methods for optimizing indoor environments are needed, as it would be too tedious and impractical to model every indoor scene manually. Currently, such indoor modeling is usually simplified or even ignored, which severely limits the realism of many virtual environments.

This paper introduces a framework for the automatic generation of furniture layouts, avoiding manual or semi-automated layout approaches, which are impractical in many graphics applications requiring full automation. We believe that our work is the first to comprehensively consider human factors such as accessibility, visibility, pathway constraints, and so forth. We have demonstrated the effectiveness of our approach in generating arrangements for a variety of scenarios. In addition, our results have been deemed by human observers to be perceptually valid in functionality when compared with real arrangements generated by human designers.

Living Room
Bedroom
Factory
Flower Store
Gallery
Resort
Restaurant
Synthesis 1
Synthesis 2
Synthesis 3