Illustration of the proposed DiffDesign.
<div><p>Interior design is a complex and creative discipline involving aesthetics, functionality, ergonomics, and materials science. Effective solutions must meet diverse requirements, typically producing multiple deliverables such as renderings and design drawings from various perspecti...
محفوظ في:
| المؤلف الرئيسي: | |
|---|---|
| مؤلفون آخرون: | |
| منشور في: |
2025
|
| الموضوعات: | |
| الوسوم: |
إضافة وسم
لا توجد وسوم, كن أول من يضع وسما على هذه التسجيلة!
|
| _version_ | 1852016949559558144 |
|---|---|
| author | Tao Geng (32392) |
| author2 | Yuxuan Yang (8088098) |
| author2_role | author |
| author_facet | Tao Geng (32392) Yuxuan Yang (8088098) |
| author_role | author |
| dc.creator.none.fl_str_mv | Tao Geng (32392) Yuxuan Yang (8088098) |
| dc.date.none.fl_str_mv | 2025-09-04T17:55:48Z |
| dc.identifier.none.fl_str_mv | 10.1371/journal.pone.0331240.g002 |
| dc.relation.none.fl_str_mv | https://figshare.com/articles/figure/Illustration_of_the_proposed_DiffDesign_/30057111 |
| dc.rights.none.fl_str_mv | CC BY 4.0 info:eu-repo/semantics/openAccess |
| dc.subject.none.fl_str_mv | Biochemistry Microbiology Genetics Biotechnology Evolutionary Biology Science Policy Environmental Sciences not elsewhere classified Biological Sciences not elsewhere classified Mathematical Sciences not elsewhere classified Information Systems not elsewhere classified extensive experiments conducted enforce view consistency demand significant creativity based alignment module 400 solutions across large image dataset dataset helps fine controllable generation quality 15 spatial types 15 design styles generative works focus controllable diffusion model interior design processes controllable diffusion interior design specific dataset spatial scope design drawings design attributes generative priors generative models various perspectives text descriptions substantial discrepancies rendering backbone promising means practical needs optimal transfer often inefficient meta priors meta prior materials science machine learning improving efficiency disentangling cross denoising process creating designs attention control |
| dc.title.none.fl_str_mv | Illustration of the proposed DiffDesign. |
| dc.type.none.fl_str_mv | Image Figure info:eu-repo/semantics/publishedVersion image |
| description | <div><p>Interior design is a complex and creative discipline involving aesthetics, functionality, ergonomics, and materials science. Effective solutions must meet diverse requirements, typically producing multiple deliverables such as renderings and design drawings from various perspectives. Consequently, interior design processes are often inefficient and demand significant creativity. With advances in machine learning, generative models have emerged as a promising means of improving efficiency by creating designs from text descriptions or sketches. However, few generative works focus on interior design, leading to substantial discrepancies between outputs and practical needs, such as differences in size, spatial scope, and the lack of controllable generation quality. To address these challenges, we propose DiffDesign, a controllable diffusion model with meta priors for efficient interior design generation. Specifically, we utilize the generative priors of a 2D diffusion model pre-trained on a large image dataset as our rendering backbone. We further guide the denoising process by disentangling cross-attention control over design attributes, such as appearance, pose, and size, and introduce an optimal transfer-based alignment module to enforce view consistency. Simultaneously, we construct an interior design-specific dataset, DesignHelper, consisting of over 400 solutions across more than 15 spatial types and 15 design styles. This dataset helps fine-tune DiffDesign. Extensive experiments conducted on various benchmark datasets demonstrate the effectiveness and robustness of DiffDesign.</p></div> |
| eu_rights_str_mv | openAccess |
| id | Manara_2fe00efad5573cbaddce1ea8283cd07c |
| identifier_str_mv | 10.1371/journal.pone.0331240.g002 |
| network_acronym_str | Manara |
| network_name_str | ManaraRepo |
| oai_identifier_str | oai:figshare.com:article/30057111 |
| publishDate | 2025 |
| repository.mail.fl_str_mv | |
| repository.name.fl_str_mv | |
| repository_id_str | |
| rights_invalid_str_mv | CC BY 4.0 |
| spelling | Illustration of the proposed DiffDesign.Tao Geng (32392)Yuxuan Yang (8088098)BiochemistryMicrobiologyGeneticsBiotechnologyEvolutionary BiologyScience PolicyEnvironmental Sciences not elsewhere classifiedBiological Sciences not elsewhere classifiedMathematical Sciences not elsewhere classifiedInformation Systems not elsewhere classifiedextensive experiments conductedenforce view consistencydemand significant creativitybased alignment module400 solutions acrosslarge image datasetdataset helps finecontrollable generation quality15 spatial types15 design stylesgenerative works focuscontrollable diffusion modelinterior design processescontrollable diffusioninterior designspecific datasetspatial scopedesign drawingsdesign attributesgenerative priorsgenerative modelsvarious perspectivestext descriptionssubstantial discrepanciesrendering backbonepromising meanspractical needsoptimal transferoften inefficientmeta priorsmeta priormaterials sciencemachine learningimproving efficiencydisentangling crossdenoising processcreating designsattention control<div><p>Interior design is a complex and creative discipline involving aesthetics, functionality, ergonomics, and materials science. Effective solutions must meet diverse requirements, typically producing multiple deliverables such as renderings and design drawings from various perspectives. Consequently, interior design processes are often inefficient and demand significant creativity. With advances in machine learning, generative models have emerged as a promising means of improving efficiency by creating designs from text descriptions or sketches. However, few generative works focus on interior design, leading to substantial discrepancies between outputs and practical needs, such as differences in size, spatial scope, and the lack of controllable generation quality. To address these challenges, we propose DiffDesign, a controllable diffusion model with meta priors for efficient interior design generation. Specifically, we utilize the generative priors of a 2D diffusion model pre-trained on a large image dataset as our rendering backbone. We further guide the denoising process by disentangling cross-attention control over design attributes, such as appearance, pose, and size, and introduce an optimal transfer-based alignment module to enforce view consistency. Simultaneously, we construct an interior design-specific dataset, DesignHelper, consisting of over 400 solutions across more than 15 spatial types and 15 design styles. This dataset helps fine-tune DiffDesign. Extensive experiments conducted on various benchmark datasets demonstrate the effectiveness and robustness of DiffDesign.</p></div>2025-09-04T17:55:48ZImageFigureinfo:eu-repo/semantics/publishedVersionimage10.1371/journal.pone.0331240.g002https://figshare.com/articles/figure/Illustration_of_the_proposed_DiffDesign_/30057111CC BY 4.0info:eu-repo/semantics/openAccessoai:figshare.com:article/300571112025-09-04T17:55:48Z |
| spellingShingle | Illustration of the proposed DiffDesign. Tao Geng (32392) Biochemistry Microbiology Genetics Biotechnology Evolutionary Biology Science Policy Environmental Sciences not elsewhere classified Biological Sciences not elsewhere classified Mathematical Sciences not elsewhere classified Information Systems not elsewhere classified extensive experiments conducted enforce view consistency demand significant creativity based alignment module 400 solutions across large image dataset dataset helps fine controllable generation quality 15 spatial types 15 design styles generative works focus controllable diffusion model interior design processes controllable diffusion interior design specific dataset spatial scope design drawings design attributes generative priors generative models various perspectives text descriptions substantial discrepancies rendering backbone promising means practical needs optimal transfer often inefficient meta priors meta prior materials science machine learning improving efficiency disentangling cross denoising process creating designs attention control |
| status_str | publishedVersion |
| title | Illustration of the proposed DiffDesign. |
| title_full | Illustration of the proposed DiffDesign. |
| title_fullStr | Illustration of the proposed DiffDesign. |
| title_full_unstemmed | Illustration of the proposed DiffDesign. |
| title_short | Illustration of the proposed DiffDesign. |
| title_sort | Illustration of the proposed DiffDesign. |
| topic | Biochemistry Microbiology Genetics Biotechnology Evolutionary Biology Science Policy Environmental Sciences not elsewhere classified Biological Sciences not elsewhere classified Mathematical Sciences not elsewhere classified Information Systems not elsewhere classified extensive experiments conducted enforce view consistency demand significant creativity based alignment module 400 solutions across large image dataset dataset helps fine controllable generation quality 15 spatial types 15 design styles generative works focus controllable diffusion model interior design processes controllable diffusion interior design specific dataset spatial scope design drawings design attributes generative priors generative models various perspectives text descriptions substantial discrepancies rendering backbone promising means practical needs optimal transfer often inefficient meta priors meta prior materials science machine learning improving efficiency disentangling cross denoising process creating designs attention control |