Research map / Image generation / Text-to-image diffusion
Zero-Shot Contrastive Loss for Text-Guided Diffusion Image Style Transfer
Serin Yang, Hyunmin Hwang, Jong Chul Ye
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 changed | Where | Measured | Effect |
|---|---|---|---|
| add ZeCon patch-wise contrastive content loss to MSE and VGG content losses (ablation) | ZeCon · user study (content) | user study content score: 3.29 -> 4.81 | Improved |
| use ZeCon loss alone instead of MSE and VGG combination (ablation) | ZeCon · user study (content) | user study content score: 3.29 -> 4.29 | Improved |
| add directional CLIP loss to global CLIP loss (ablation) | ZeCon style guidance · qualitative styles | stylization strength | Improved |
| use patch-based CLIP guidance instead of whole-image guidance (ablation) | ZeCon style guidance · qualitative styles | stylization coverage | Improved |
| add augmentation and use N=32 patches (N<32 insufficient) (ablation) | patch CLIP loss · Golden style | stylization: inference time 24 seconds with N=32 | Improved |
| increase t0 for the forward diffusion (ablation) | ZeCon · style transfer | CLIP score vs content preservation | Mixed |
| use ZeCon instead of DiffusionCLIP for inference | DiffusionCLIP · computational cost | inference time (sec): 96 -> 38 | Improved |
| apply method to unseen-domain images | DiffusionCLIP · user study content score | content score: DiffusionCLIP 3.71 -> 3.29 (photo -> unseen); ZeCon 4.76 -> 4.62 | Improved |
| use ZeCon instead of StyleCLIP on face identity loss | StyleCLIP · Face ID | Face ID (lower better): 0.3750 -> 0.3881 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 3 of 3 on most shared results.
- DiffusionCLIP — DiffusionCLIP: Text-Guided Diffusion Models for Robust Image Manipulation (CVPR 2022) · better on 4 of 4 shared results
- CLIPstyler — CLIPstyler: Image Style Transfer With a Single Text Condition (CVPR 2022) · better on 1 of 1 shared result
- VQGAN-CLIP — VQGAN-CLIP: Open Domain Image Generation and Editing with Natural Language Guidance (ECCV 2022) · better on 1 of 1 shared result
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
- Z*: Zero-shot Style Transfer via Attention Reweighting (CVPR 2024 · similarity 0.91)
- Less is More: Masking Elements in Image Condition Features Avoids Content Leakages in Style Transfer Diffusion Models (ICLR 2025 · similarity 0.90)
- Style Injection in Diffusion: A Training-free Approach for Adapting Large-scale Diffusion Models for Style Transfer (CVPR 2024 · similarity 0.89)
- Diffusion Self-Distillation for Zero-Shot Customized Image Generation (CVPR 2025 · similarity 0.89)
- Style Nursing with Spatial and Semantic Guidance for Zero-Shot Traffic Scene Style Transfer (AAAI 2025 · similarity 0.89)
- R&B: Region and Boundary Aware Zero-shot Grounded Text-to-image Generation (ICLR 2024 · similarity 0.88)
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.
