AI Paper Atlas

Research map / Image generation / Text-to-image diffusion

SplitFlux: Learning to Decouple Content and Style from a Single Image

Yitong Yang, Yinglin Wang, Changshuo Wang, Yongjun Zhang, Ziyang Chen, Shuting He

CVPR 2026 · Image generation › Text-to-image diffusion · approach: Layout-to-image control · 2 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add RCA (rank compression plus amplified scale on blocks 30-31) to LoRA-Flux (ablation)SplitFlux · DisentanglementDINO-C: 0.756 -> 0.784 (Table 3)Improved
add RCA with alpha=2 vs alpha=8 (ablation)SplitFlux · RCA ablationCLIP-C: 0.878 -> 0.879No clear effect
fine-tune blocks 30-31 (vs no fine-tuning of them) (ablation)SplitFlux · RCA ablationDINO-C: 0.709 -> 0.784Improved
extend RCA block range from 30-31 to 30-35 (ablation)SplitFlux · RCA ablationDINO-S: 0.370 -> 0.307Hurt
add Visual-Gated LoRA (VGRA) (ablation)SplitFlux · RecontextualizationqualitativeImproved
increase content-branch rank r_cnt (ablation)VGRA · RecontextualizationqualitativeMixed
SplitFlux vs LoRA-Flux on content preservationSplitFlux · DisentanglementVLM-C: 17.5% -> 67.5%Improved
SplitFlux vs LoRA-Flux on styleSplitFlux · DisentanglementCLIP-S: 0.665 -> 0.666No clear effect
fine-tune Flux single stream blocks (not double stream) for style/content (ablation)Flux · Prompt replacementsemantic injection success: 0% -> 100%Improved
SplitFlux vs UnZipLoRA on merger styleSplitFlux · MergerDINO-S: 0.415 -> 0.368Hurt

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy