mlsanityData to 14 Oct 2026

TeleStyle V2: Beyond Content-Preserving Style Transfer with Self-Distillation and Distribution-Matching-Distillation

arXiv 2606.20709 ↗First posted 16 Jun 2026cs.CVcited by 0

On the map: Knowledge distillation → Multimodal distillation

Abstract (arXiv)

Given a content reference and a style reference, content-preserving style transfer requires the model to generate stylized outputs with content and style consistency. We introduced TeleStyle V1 to tackle this problem. However, TeleStyle V1 is trained with photorealistic content reference and artistic style reference, which makes it incapable to cope with artistic content reference and realistic style reference in most cases. In this paper, we designed a Self-Distillation data synthesis strategy to construct such triplets from TeleStyle V1. Trained with such self-distilled triplets, our TeleStyle V2 supports Content-Style references in the forms of Realistic-and-Realistic (RnR), Realistic-and-Stylized (RnS), Stylized-and-Realistic (SnR), Stylized-and-Stylized (SnS). In addition, we found Distribution Matching Distillation could preserve the general text-guided image editing capability of the foundation model and fix the content consistency degradation caused by SFT process. Through quantitative evaluations, our TeleStyleV2-QIE-2509-DMD performs at least on par with Qwen-Image-Edit-2509-DMD, demonstrating strong general image editing skills beyond content-preserving style transfer. We observed the content/style reference order confusion problem in TeleStyle V1 and further introduced prompt enhancer to solve it. TeleStyle V2 uses Qwen-Image-Edit's VLM encoder, Qwen2.5-VL-7B, to generate content prompt and style prompt for free. TeleStyle V2 could achieve comparable style transfer performance with state-of-the-art commercial model, gemini-3-pro-image-preview.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Multimodal distillation, one of 12 directions in Knowledge distillation. Builds on 24 of its 36 references that are in the map. First posted on arXiv 16 Jun 2026.

Builds on

24 of 36 references.

References · 24

  1. 2020Denoising Diffusion Probabilistic Modelscited by 21,076
  2. 2021LoRA: Low-Rank Adaptation of Large Language Modelscited by 14,380
  3. 2023Direct Preference Optimization: Your Language Model is Secretly a Reward Modelcited by 8,065
  4. 2022Scalable Diffusion Models with Transformerscited by 6,054
  5. 2025Qwen2.5-VL Technical Reportcited by 5,587
  6. 2021RoFormer: Enhanced Transformer with Rotary Position Embeddingcited by 4,703
  7. 2024GPT-4o System Cardcited by 4,588
  8. 2023SDXL: Improving Latent Diffusion Models for High-Resolution Image Synthesiscited by 4,356
  9. 2024Scaling Rectified Flow Transformers for High-Resolution Image Synthesiscited by 4,181
  10. 2022Imagic: Text-Based Real Image Editing with Diffusion Modelscited by 1,178
  11. 2025Qwen-Image Technical Reportcited by 1,044
  12. 2024Improved Distribution Matching Distillation for Fast Image Synthesiscited by 684
12 more
  1. 2025Step1X-Edit: A Practical Framework for General Image Editingcited by 429
  2. 2020V4D:4D Convolutional Neural Networks for Video-level Representation Learningcited by 33
  3. 2024StyleTokenizer: Defining Image Style by a Single Instance for Controlling Diffusion Modelscited by 32
  4. 2025OmniConsistency: Learning Style-Agnostic Consistency from Paired Stylization Datacited by 32
  5. 2023Forgedit: Text Guided Image Editing via Learning and Forgettingcited by 26
  6. 2025Phased DMD: Few-step Distribution Matching Distillation via Score Matching within Subintervalscited by 24
  7. 2022TFCNet: Temporal Fully Connected Networks for Static Unbiased Temporal Reasoningcited by 12
  8. 2020Knowledge Integration Networks for Action Recognitioncited by 8
  9. 2025CDST: Color Disentangled Style Transfer for Universal Style Reference Customizationcited by 6
  10. 2026TeleStyle: Content-Preserving Style Transfer in Images and Videoscited by 5
  11. 2026QwenStyle: Content-Preserving Style Transfer with Qwen-Image-Editcited by 4
  12. 2026Style-CCL: Content-Preserving Style Transfer via Curriculum Continual Learningcited by 1

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2022PCA-Based Knowledge Distillation Towards Lightweight and Content-Style Balanced Photorealistic Style Transfer Models
  2. 2025FantasyStyle: Controllable Stylized Distillation for 3D Gaussian Splatting
  3. 2025OmniStyle: Filtering High Quality Style Transfer Data at Scale
  4. 2024InstantStyle-Plus: Style Transfer with Content-Preserving in Text-to-Image Generation
  5. 2025SCAdapter: Content-Style Disentanglement for Diffusion Style Transfer
  6. 2025Training-Free Multi-Style Fusion Through Reference-Based Adaptive Modulation
  7. 2026Staying True to the Origin: Continuous Image Stylization with Smooth Transitions
  8. 2026Scheduled Style Injection: Expanding the Style-Content Pareto Frontier in Training-Free Diffusion-based Style Transfer