{"slug": "high-fidelity-two-step-image-generation-via-teacher-aligned-end-to-end", "title": "High-Fidelity Two-Step Image Generation via Teacher-Aligned End-to-End Distillation", "summary": "Researchers have developed Z-Image Turbo++, a two-step image generation model distilled from an eight-step teacher model, achieving high-fidelity results through three targeted design choices. The method uses teacher-generated images for adversarial training, assigns independent parameters to each denoising step, and employs end-to-end training with iterative regularization to narrow the quality gap between two-step and eight-step generation. This advancement demonstrates that carefully tailored distillation strategies can significantly improve the quality-efficiency trade-off in few-step image generation.", "body_md": "arXiv:2606.12575v1 Announce Type: new\nAbstract: Few-step diffusion distillation has become increasingly mature for 4-8-step generation, yet pushing further to 2 steps remains challenging. In this work, we introduce Z-Image Turbo++, a high-quality 2-step image generation model distilled from the 8-step Z-Image Turbo teacher. Our method addresses the central bottlenecks of increased task difficulty and limited model capacity in 2-step generation through three simple but effective design choices tailored to this regime. First, we propose Distribution-Aligned Adversarial Learning, which uses teacher-generated images rather than external real images as real samples for GAN training, providing a more attainable and informative adversarial target. Second, we adopt Step-Decoupled Parameterization, assigning independent model parameters to the two denoising steps to better match their distinct capacity demands. Third, we perform End-to-End Training with Iterative Regularization, allowing the first step to receive gradients from final image quality while preserving a meaningful intermediate generation through an explicit step-1 loss. Together, these designs substantially narrow the quality gap between 2-step and 8-step generation in both qualitative and quantitative evaluations, highlighting the potential of carefully tailored distillation strategies for improving the quality-efficiency trade-off in few-step generation.", "url": "https://wpnews.pro/news/high-fidelity-two-step-image-generation-via-teacher-aligned-end-to-end", "canonical_source": "https://arxiv.org/abs/2606.12575", "published_at": "2026-06-12 04:00:00+00:00", "updated_at": "2026-06-12 04:49:04.414678+00:00", "lang": "en", "topics": ["generative-ai", "machine-learning", "computer-vision", "artificial-intelligence", "neural-networks"], "entities": ["Z-Image Turbo++", "Z-Image Turbo", "Distribution-Aligned Adversarial Learning", "Step-Decoupled Parameterization", "End-to-End Training with Iterative Regularization"], "alternates": {"html": "https://wpnews.pro/news/high-fidelity-two-step-image-generation-via-teacher-aligned-end-to-end", "markdown": "https://wpnews.pro/news/high-fidelity-two-step-image-generation-via-teacher-aligned-end-to-end.md", "text": "https://wpnews.pro/news/high-fidelity-two-step-image-generation-via-teacher-aligned-end-to-end.txt", "jsonld": "https://wpnews.pro/news/high-fidelity-two-step-image-generation-via-teacher-aligned-end-to-end.jsonld"}}