Distilling Diffusion Models into Conditional GANs Article Swipe
Minguk Kang
,
Richard Zhang
,
Connelly Barnes
,
Sylvain Paris
,
Suha Kwak
,
Jaesik Park
,
Eli Shechtman
,
Jun-Yan Zhu
,
Taesung Park
·
YOU?
·
· 2024
· Open Access
·
· DOI: https://doi.org/10.48550/arxiv.2405.05967
YOU?
·
· 2024
· Open Access
·
· DOI: https://doi.org/10.48550/arxiv.2405.05967
We propose a method to distill a complex multistep diffusion model into a single-step conditional GAN student model, dramatically accelerating inference, while preserving image quality. Our approach interprets diffusion distillation as a paired image-to-image translation task, using noise-to-image pairs of the diffusion model's ODE trajectory. For efficient regression loss computation, we propose E-LatentLPIPS, a perceptual loss operating directly in diffusion model's latent space, utilizing an ensemble of augmentations. Furthermore, we adapt a diffusion model to construct a multi-scale discriminator with a text alignment loss to build an effective conditional GAN-based formulation. E-LatentLPIPS converges more efficiently than many existing distillation methods, even accounting for dataset construction costs. We demonstrate that our one-step generator outperforms cutting-edge one-step diffusion distillation models -- DMD, SDXL-Turbo, and SDXL-Lightning -- on the zero-shot COCO benchmark.
Related Topics
Concepts
Metadata
- Type
- preprint
- Language
- en
- Landing Page
- http://arxiv.org/abs/2405.05967
- https://arxiv.org/pdf/2405.05967
- OA Status
- green
- Cited By
- 2
- Related Works
- 10
- OpenAlex ID
- https://openalex.org/W4396822614
All OpenAlex metadata
Raw OpenAlex JSON
- OpenAlex ID
-
https://openalex.org/W4396822614Canonical identifier for this work in OpenAlex
- DOI
-
https://doi.org/10.48550/arxiv.2405.05967Digital Object Identifier
- Title
-
Distilling Diffusion Models into Conditional GANsWork title
- Type
-
preprintOpenAlex work type
- Language
-
enPrimary language
- Publication year
-
2024Year of publication
- Publication date
-
2024-05-09Full publication date if available
- Authors
-
Minguk Kang, Richard Zhang, Connelly Barnes, Sylvain Paris, Suha Kwak, Jaesik Park, Eli Shechtman, Jun-Yan Zhu, Taesung ParkList of authors in order
- Landing page
-
https://arxiv.org/abs/2405.05967Publisher landing page
- PDF URL
-
https://arxiv.org/pdf/2405.05967Direct link to full text PDF
- Open access
-
YesWhether a free full text is available
- OA status
-
greenOpen access status per OpenAlex
- OA URL
-
https://arxiv.org/pdf/2405.05967Direct OA link when available
- Concepts
-
Diffusion, Computer science, Econometrics, Statistical physics, Mathematics, Thermodynamics, PhysicsTop concepts (fields/topics) attached by OpenAlex
- Cited by
-
2Total citation count in OpenAlex
- Citations by year (recent)
-
2025: 2Per-year citation counts (last 5 years)
- Related works (count)
-
10Other works algorithmically related by OpenAlex
Full payload
| id | https://openalex.org/W4396822614 |
|---|---|
| doi | https://doi.org/10.48550/arxiv.2405.05967 |
| ids.doi | https://doi.org/10.48550/arxiv.2405.05967 |
| ids.openalex | https://openalex.org/W4396822614 |
| fwci | |
| type | preprint |
| title | Distilling Diffusion Models into Conditional GANs |
| biblio.issue | |
| biblio.volume | |
| biblio.last_page | |
| biblio.first_page | |
| topics[0].id | https://openalex.org/T13702 |
| topics[0].field.id | https://openalex.org/fields/17 |
| topics[0].field.display_name | Computer Science |
| topics[0].score | 0.20309999585151672 |
| topics[0].domain.id | https://openalex.org/domains/3 |
| topics[0].domain.display_name | Physical Sciences |
| topics[0].subfield.id | https://openalex.org/subfields/1702 |
| topics[0].subfield.display_name | Artificial Intelligence |
| topics[0].display_name | Machine Learning in Healthcare |
| topics[1].id | https://openalex.org/T10201 |
| topics[1].field.id | https://openalex.org/fields/17 |
| topics[1].field.display_name | Computer Science |
| topics[1].score | 0.18649999797344208 |
| topics[1].domain.id | https://openalex.org/domains/3 |
| topics[1].domain.display_name | Physical Sciences |
| topics[1].subfield.id | https://openalex.org/subfields/1702 |
| topics[1].subfield.display_name | Artificial Intelligence |
| topics[1].display_name | Speech Recognition and Synthesis |
| is_xpac | False |
| apc_list | |
| apc_paid | |
| concepts[0].id | https://openalex.org/C69357855 |
| concepts[0].level | 2 |
| concepts[0].score | 0.5745677351951599 |
| concepts[0].wikidata | https://www.wikidata.org/wiki/Q163214 |
| concepts[0].display_name | Diffusion |
| concepts[1].id | https://openalex.org/C41008148 |
| concepts[1].level | 0 |
| concepts[1].score | 0.46923738718032837 |
| concepts[1].wikidata | https://www.wikidata.org/wiki/Q21198 |
| concepts[1].display_name | Computer science |
| concepts[2].id | https://openalex.org/C149782125 |
| concepts[2].level | 1 |
| concepts[2].score | 0.3661797046661377 |
| concepts[2].wikidata | https://www.wikidata.org/wiki/Q160039 |
| concepts[2].display_name | Econometrics |
| concepts[3].id | https://openalex.org/C121864883 |
| concepts[3].level | 1 |
| concepts[3].score | 0.34478700160980225 |
| concepts[3].wikidata | https://www.wikidata.org/wiki/Q677916 |
| concepts[3].display_name | Statistical physics |
| concepts[4].id | https://openalex.org/C33923547 |
| concepts[4].level | 0 |
| concepts[4].score | 0.19796967506408691 |
| concepts[4].wikidata | https://www.wikidata.org/wiki/Q395 |
| concepts[4].display_name | Mathematics |
| concepts[5].id | https://openalex.org/C97355855 |
| concepts[5].level | 1 |
| concepts[5].score | 0.13396358489990234 |
| concepts[5].wikidata | https://www.wikidata.org/wiki/Q11473 |
| concepts[5].display_name | Thermodynamics |
| concepts[6].id | https://openalex.org/C121332964 |
| concepts[6].level | 0 |
| concepts[6].score | 0.09870824217796326 |
| concepts[6].wikidata | https://www.wikidata.org/wiki/Q413 |
| concepts[6].display_name | Physics |
| keywords[0].id | https://openalex.org/keywords/diffusion |
| keywords[0].score | 0.5745677351951599 |
| keywords[0].display_name | Diffusion |
| keywords[1].id | https://openalex.org/keywords/computer-science |
| keywords[1].score | 0.46923738718032837 |
| keywords[1].display_name | Computer science |
| keywords[2].id | https://openalex.org/keywords/econometrics |
| keywords[2].score | 0.3661797046661377 |
| keywords[2].display_name | Econometrics |
| keywords[3].id | https://openalex.org/keywords/statistical-physics |
| keywords[3].score | 0.34478700160980225 |
| keywords[3].display_name | Statistical physics |
| keywords[4].id | https://openalex.org/keywords/mathematics |
| keywords[4].score | 0.19796967506408691 |
| keywords[4].display_name | Mathematics |
| keywords[5].id | https://openalex.org/keywords/thermodynamics |
| keywords[5].score | 0.13396358489990234 |
| keywords[5].display_name | Thermodynamics |
| keywords[6].id | https://openalex.org/keywords/physics |
| keywords[6].score | 0.09870824217796326 |
| keywords[6].display_name | Physics |
| language | en |
| locations[0].id | pmh:oai:arXiv.org:2405.05967 |
| locations[0].is_oa | True |
| locations[0].source.id | https://openalex.org/S4306400194 |
| locations[0].source.issn | |
| locations[0].source.type | repository |
| locations[0].source.is_oa | True |
| locations[0].source.issn_l | |
| locations[0].source.is_core | False |
| locations[0].source.is_in_doaj | False |
| locations[0].source.display_name | arXiv (Cornell University) |
| locations[0].source.host_organization | https://openalex.org/I205783295 |
| locations[0].source.host_organization_name | Cornell University |
| locations[0].source.host_organization_lineage | https://openalex.org/I205783295 |
| locations[0].license | |
| locations[0].pdf_url | https://arxiv.org/pdf/2405.05967 |
| locations[0].version | submittedVersion |
| locations[0].raw_type | text |
| locations[0].license_id | |
| locations[0].is_accepted | False |
| locations[0].is_published | False |
| locations[0].raw_source_name | |
| locations[0].landing_page_url | http://arxiv.org/abs/2405.05967 |
| locations[1].id | doi:10.48550/arxiv.2405.05967 |
| locations[1].is_oa | True |
| locations[1].source.id | https://openalex.org/S4306400194 |
| locations[1].source.issn | |
| locations[1].source.type | repository |
| locations[1].source.is_oa | True |
| locations[1].source.issn_l | |
| locations[1].source.is_core | False |
| locations[1].source.is_in_doaj | False |
| locations[1].source.display_name | arXiv (Cornell University) |
| locations[1].source.host_organization | https://openalex.org/I205783295 |
| locations[1].source.host_organization_name | Cornell University |
| locations[1].source.host_organization_lineage | https://openalex.org/I205783295 |
| locations[1].license | |
| locations[1].pdf_url | |
| locations[1].version | |
| locations[1].raw_type | article |
| locations[1].license_id | |
| locations[1].is_accepted | False |
| locations[1].is_published | |
| locations[1].raw_source_name | |
| locations[1].landing_page_url | https://doi.org/10.48550/arxiv.2405.05967 |
| indexed_in | arxiv, datacite |
| authorships[0].author.id | https://openalex.org/A5032772116 |
| authorships[0].author.orcid | https://orcid.org/0000-0002-6618-5971 |
| authorships[0].author.display_name | Minguk Kang |
| authorships[0].author_position | first |
| authorships[0].raw_author_name | Kang, Minguk |
| authorships[0].is_corresponding | False |
| authorships[1].author.id | https://openalex.org/A5101616527 |
| authorships[1].author.orcid | https://orcid.org/0000-0001-7712-1247 |
| authorships[1].author.display_name | Richard Zhang |
| authorships[1].author_position | middle |
| authorships[1].raw_author_name | Zhang, Richard |
| authorships[1].is_corresponding | False |
| authorships[2].author.id | https://openalex.org/A5083898813 |
| authorships[2].author.orcid | |
| authorships[2].author.display_name | Connelly Barnes |
| authorships[2].author_position | middle |
| authorships[2].raw_author_name | Barnes, Connelly |
| authorships[2].is_corresponding | False |
| authorships[3].author.id | https://openalex.org/A5087205270 |
| authorships[3].author.orcid | |
| authorships[3].author.display_name | Sylvain Paris |
| authorships[3].author_position | middle |
| authorships[3].raw_author_name | Paris, Sylvain |
| authorships[3].is_corresponding | False |
| authorships[4].author.id | https://openalex.org/A5101352002 |
| authorships[4].author.orcid | |
| authorships[4].author.display_name | Suha Kwak |
| authorships[4].author_position | middle |
| authorships[4].raw_author_name | Kwak, Suha |
| authorships[4].is_corresponding | False |
| authorships[5].author.id | https://openalex.org/A5100611457 |
| authorships[5].author.orcid | https://orcid.org/0000-0001-5541-409X |
| authorships[5].author.display_name | Jaesik Park |
| authorships[5].author_position | middle |
| authorships[5].raw_author_name | Park, Jaesik |
| authorships[5].is_corresponding | False |
| authorships[6].author.id | https://openalex.org/A5024777042 |
| authorships[6].author.orcid | https://orcid.org/0000-0002-6783-1795 |
| authorships[6].author.display_name | Eli Shechtman |
| authorships[6].author_position | middle |
| authorships[6].raw_author_name | Shechtman, Eli |
| authorships[6].is_corresponding | False |
| authorships[7].author.id | https://openalex.org/A5102883508 |
| authorships[7].author.orcid | https://orcid.org/0000-0001-8504-3410 |
| authorships[7].author.display_name | Jun-Yan Zhu |
| authorships[7].author_position | middle |
| authorships[7].raw_author_name | Zhu, Jun-Yan |
| authorships[7].is_corresponding | False |
| authorships[8].author.id | https://openalex.org/A5100643662 |
| authorships[8].author.orcid | https://orcid.org/0000-0002-8294-590X |
| authorships[8].author.display_name | Taesung Park |
| authorships[8].author_position | last |
| authorships[8].raw_author_name | Park, Taesung |
| authorships[8].is_corresponding | False |
| has_content.pdf | False |
| has_content.grobid_xml | False |
| is_paratext | False |
| open_access.is_oa | True |
| open_access.oa_url | https://arxiv.org/pdf/2405.05967 |
| open_access.oa_status | green |
| open_access.any_repository_has_fulltext | False |
| created_date | 2025-10-10T00:00:00 |
| display_name | Distilling Diffusion Models into Conditional GANs |
| has_fulltext | False |
| is_retracted | False |
| updated_date | 2025-11-06T06:51:31.235846 |
| primary_topic.id | https://openalex.org/T13702 |
| primary_topic.field.id | https://openalex.org/fields/17 |
| primary_topic.field.display_name | Computer Science |
| primary_topic.score | 0.20309999585151672 |
| primary_topic.domain.id | https://openalex.org/domains/3 |
| primary_topic.domain.display_name | Physical Sciences |
| primary_topic.subfield.id | https://openalex.org/subfields/1702 |
| primary_topic.subfield.display_name | Artificial Intelligence |
| primary_topic.display_name | Machine Learning in Healthcare |
| related_works | https://openalex.org/W4391375266, https://openalex.org/W2748952813, https://openalex.org/W2390279801, https://openalex.org/W2358668433, https://openalex.org/W4396701345, https://openalex.org/W2376932109, https://openalex.org/W2001405890, https://openalex.org/W4396696052, https://openalex.org/W2382290278, https://openalex.org/W4395014643 |
| cited_by_count | 2 |
| counts_by_year[0].year | 2025 |
| counts_by_year[0].cited_by_count | 2 |
| locations_count | 2 |
| best_oa_location.id | pmh:oai:arXiv.org:2405.05967 |
| best_oa_location.is_oa | True |
| best_oa_location.source.id | https://openalex.org/S4306400194 |
| best_oa_location.source.issn | |
| best_oa_location.source.type | repository |
| best_oa_location.source.is_oa | True |
| best_oa_location.source.issn_l | |
| best_oa_location.source.is_core | False |
| best_oa_location.source.is_in_doaj | False |
| best_oa_location.source.display_name | arXiv (Cornell University) |
| best_oa_location.source.host_organization | https://openalex.org/I205783295 |
| best_oa_location.source.host_organization_name | Cornell University |
| best_oa_location.source.host_organization_lineage | https://openalex.org/I205783295 |
| best_oa_location.license | |
| best_oa_location.pdf_url | https://arxiv.org/pdf/2405.05967 |
| best_oa_location.version | submittedVersion |
| best_oa_location.raw_type | text |
| best_oa_location.license_id | |
| best_oa_location.is_accepted | False |
| best_oa_location.is_published | False |
| best_oa_location.raw_source_name | |
| best_oa_location.landing_page_url | http://arxiv.org/abs/2405.05967 |
| primary_location.id | pmh:oai:arXiv.org:2405.05967 |
| primary_location.is_oa | True |
| primary_location.source.id | https://openalex.org/S4306400194 |
| primary_location.source.issn | |
| primary_location.source.type | repository |
| primary_location.source.is_oa | True |
| primary_location.source.issn_l | |
| primary_location.source.is_core | False |
| primary_location.source.is_in_doaj | False |
| primary_location.source.display_name | arXiv (Cornell University) |
| primary_location.source.host_organization | https://openalex.org/I205783295 |
| primary_location.source.host_organization_name | Cornell University |
| primary_location.source.host_organization_lineage | https://openalex.org/I205783295 |
| primary_location.license | |
| primary_location.pdf_url | https://arxiv.org/pdf/2405.05967 |
| primary_location.version | submittedVersion |
| primary_location.raw_type | text |
| primary_location.license_id | |
| primary_location.is_accepted | False |
| primary_location.is_published | False |
| primary_location.raw_source_name | |
| primary_location.landing_page_url | http://arxiv.org/abs/2405.05967 |
| publication_date | 2024-05-09 |
| publication_year | 2024 |
| referenced_works_count | 0 |
| abstract_inverted_index.a | 2, 6, 12, 31, 53, 71, 76, 80 |
| abstract_inverted_index.-- | 118, 123 |
| abstract_inverted_index.We | 0, 106 |
| abstract_inverted_index.an | 64, 86 |
| abstract_inverted_index.as | 30 |
| abstract_inverted_index.in | 58 |
| abstract_inverted_index.of | 39, 66 |
| abstract_inverted_index.on | 124 |
| abstract_inverted_index.to | 4, 74, 84 |
| abstract_inverted_index.we | 50, 69 |
| abstract_inverted_index.For | 45 |
| abstract_inverted_index.GAN | 15 |
| abstract_inverted_index.ODE | 43 |
| abstract_inverted_index.Our | 25 |
| abstract_inverted_index.and | 121 |
| abstract_inverted_index.for | 102 |
| abstract_inverted_index.our | 109 |
| abstract_inverted_index.the | 40, 125 |
| abstract_inverted_index.COCO | 127 |
| abstract_inverted_index.DMD, | 119 |
| abstract_inverted_index.even | 100 |
| abstract_inverted_index.into | 11 |
| abstract_inverted_index.loss | 48, 55, 83 |
| abstract_inverted_index.many | 96 |
| abstract_inverted_index.more | 93 |
| abstract_inverted_index.text | 81 |
| abstract_inverted_index.than | 95 |
| abstract_inverted_index.that | 108 |
| abstract_inverted_index.with | 79 |
| abstract_inverted_index.adapt | 70 |
| abstract_inverted_index.build | 85 |
| abstract_inverted_index.image | 23 |
| abstract_inverted_index.model | 10, 73 |
| abstract_inverted_index.pairs | 38 |
| abstract_inverted_index.task, | 35 |
| abstract_inverted_index.using | 36 |
| abstract_inverted_index.while | 21 |
| abstract_inverted_index.costs. | 105 |
| abstract_inverted_index.latent | 61 |
| abstract_inverted_index.method | 3 |
| abstract_inverted_index.model, | 17 |
| abstract_inverted_index.models | 117 |
| abstract_inverted_index.paired | 32 |
| abstract_inverted_index.space, | 62 |
| abstract_inverted_index.complex | 7 |
| abstract_inverted_index.dataset | 103 |
| abstract_inverted_index.distill | 5 |
| abstract_inverted_index.model's | 42, 60 |
| abstract_inverted_index.propose | 1, 51 |
| abstract_inverted_index.student | 16 |
| abstract_inverted_index.approach | 26 |
| abstract_inverted_index.directly | 57 |
| abstract_inverted_index.ensemble | 65 |
| abstract_inverted_index.existing | 97 |
| abstract_inverted_index.methods, | 99 |
| abstract_inverted_index.one-step | 110, 114 |
| abstract_inverted_index.quality. | 24 |
| abstract_inverted_index.GAN-based | 89 |
| abstract_inverted_index.alignment | 82 |
| abstract_inverted_index.construct | 75 |
| abstract_inverted_index.converges | 92 |
| abstract_inverted_index.diffusion | 9, 28, 41, 59, 72, 115 |
| abstract_inverted_index.effective | 87 |
| abstract_inverted_index.efficient | 46 |
| abstract_inverted_index.generator | 111 |
| abstract_inverted_index.multistep | 8 |
| abstract_inverted_index.operating | 56 |
| abstract_inverted_index.utilizing | 63 |
| abstract_inverted_index.zero-shot | 126 |
| abstract_inverted_index.accounting | 101 |
| abstract_inverted_index.benchmark. | 128 |
| abstract_inverted_index.inference, | 20 |
| abstract_inverted_index.interprets | 27 |
| abstract_inverted_index.perceptual | 54 |
| abstract_inverted_index.preserving | 22 |
| abstract_inverted_index.regression | 47 |
| abstract_inverted_index.SDXL-Turbo, | 120 |
| abstract_inverted_index.conditional | 14, 88 |
| abstract_inverted_index.demonstrate | 107 |
| abstract_inverted_index.efficiently | 94 |
| abstract_inverted_index.multi-scale | 77 |
| abstract_inverted_index.outperforms | 112 |
| abstract_inverted_index.single-step | 13 |
| abstract_inverted_index.trajectory. | 44 |
| abstract_inverted_index.translation | 34 |
| abstract_inverted_index.Furthermore, | 68 |
| abstract_inverted_index.accelerating | 19 |
| abstract_inverted_index.computation, | 49 |
| abstract_inverted_index.construction | 104 |
| abstract_inverted_index.cutting-edge | 113 |
| abstract_inverted_index.distillation | 29, 98, 116 |
| abstract_inverted_index.dramatically | 18 |
| abstract_inverted_index.formulation. | 90 |
| abstract_inverted_index.E-LatentLPIPS | 91 |
| abstract_inverted_index.discriminator | 78 |
| abstract_inverted_index.E-LatentLPIPS, | 52 |
| abstract_inverted_index.SDXL-Lightning | 122 |
| abstract_inverted_index.augmentations. | 67 |
| abstract_inverted_index.image-to-image | 33 |
| abstract_inverted_index.noise-to-image | 37 |
| cited_by_percentile_year | |
| countries_distinct_count | 0 |
| institutions_distinct_count | 9 |
| citation_normalized_percentile |