Saved in:
Bibliographic Details
Main Authors: Do, Tung, Nguyen, Thuan Hoang, Li, Hao
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2605.15592
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914568818655232
author Do, Tung
Nguyen, Thuan Hoang
Li, Hao
author_facet Do, Tung
Nguyen, Thuan Hoang
Li, Hao
contents Few-step image generation has seen rapid progress, with consistency and meanflow-based methods significantly reducing the number of sampling steps. Despite their low inference cost, these approaches often suffer from training instability and limited scalability. Sphere Encoder is a recent alternative that produces high-quality images in only a few steps; however, it requires repeated transitions between the pixel space and latent space during inference while jointly optimizing reconstruction and generation within a single architecture. This design leads to computational inefficiency and objective conflict between reconstruction and generation. To address these limitations, we decouple the framework into a fixed pretrained image encoder and a separate latent denoising model trained entirely in a spherical latent space. Our approach eliminates repeated pixel-space operations during training and inference, improving efficiency and allowing reconstruction and generation to specialize independently. On Animal-Faces, Oxford-Flowers and ImageNet-1K datasets, our method significantly outperforms Sphere Encoder in both generation quality and inference speed, while achieving competitive results against strong few-step and multi-step baselines.
format Preprint
id arxiv_https___arxiv_org_abs_2605_15592
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Efficient Image Synthesis with Sphere Latent Encoder
Do, Tung
Nguyen, Thuan Hoang
Li, Hao
Computer Vision and Pattern Recognition
Few-step image generation has seen rapid progress, with consistency and meanflow-based methods significantly reducing the number of sampling steps. Despite their low inference cost, these approaches often suffer from training instability and limited scalability. Sphere Encoder is a recent alternative that produces high-quality images in only a few steps; however, it requires repeated transitions between the pixel space and latent space during inference while jointly optimizing reconstruction and generation within a single architecture. This design leads to computational inefficiency and objective conflict between reconstruction and generation. To address these limitations, we decouple the framework into a fixed pretrained image encoder and a separate latent denoising model trained entirely in a spherical latent space. Our approach eliminates repeated pixel-space operations during training and inference, improving efficiency and allowing reconstruction and generation to specialize independently. On Animal-Faces, Oxford-Flowers and ImageNet-1K datasets, our method significantly outperforms Sphere Encoder in both generation quality and inference speed, while achieving competitive results against strong few-step and multi-step baselines.
title Efficient Image Synthesis with Sphere Latent Encoder
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2605.15592