PixVerve: Advancing Native UHR Image Generation to 100MP with a Large-Scale High-Quality Dataset
Summary
This paper introduces PixVerve-95K, a large-scale open-source dataset of 95K ultra-high-resolution (100MP) images with annotations, and PixVerve-Bench, a benchmark for evaluating native 100MP text-to-image generation, extending existing T2I models to unprecedented resolutions.
View Cached Full Text
Cached at: 05/20/26, 10:37 AM
Paper page - PixVerve: Advancing Native UHR Image Generation to 100MP with a Large-Scale High-Quality Dataset
Source: https://huggingface.co/papers/2605.20147 Authors:
,
,
,
,
,
,
,
,
,
,
,
,
Abstract
A large-scale UHR image-text dataset and evaluation benchmark are introduced to advance ultra-high-resolution text-to-image generation capabilities.
Text-to-Image (T2I) models have recently seen notable progress around 1K and 2K resolution. With the extreme desire for better visual experience and the rapid development of imaging technology, the demand forUltra-High-Resolution(UHR) image generation has grown significantly. However, UHR image generation poses great challenges due to the scarcity and complexity of high-resolution content. In this paper, we first introducePixVerve-95K, a high-quality, open-source UHR T2I dataset curated with a carefully designed data pipeline, which contains 95K images across diverse scenarios (each image has a minimum pixel-count of 100M) and seven-dimensional annotations. Based on our large-scaleimage-text dataset, we take a pioneering step to extend various T2I foundation models to native 100MP generation with three training schemes. Finally, leveraging both conventional metrics andmultimodal large language model-based assessments, our proposedPixVerve-Benchbenchmark establishes a comprehensive evaluation protocol for UHR images encompassingvisual qualityandsemantic alignment. Extensive experimental results on our benchmark and the constructive exploration of training strategies collaboratively provide valuable insights for future breakthroughs.
View arXiv pageView PDFProject pageGitHub5Add to collection
Get this paper in your agent:
hf papers read 2605\.20147
Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash
Models citing this paper0
No model linking this paper
Cite arxiv.org/abs/2605.20147 in a model README.md to link it from this page.
Datasets citing this paper0
No dataset linking this paper
Cite arxiv.org/abs/2605.20147 in a dataset README.md to link it from this page.
Spaces citing this paper0
No Space linking this paper
Cite arxiv.org/abs/2605.20147 in a Space README.md to link it from this page.
Collections including this paper0
No Collection including this paper
Add this paper to acollectionto link it from this page.
Similar Articles
@drfeifei: I’m very excited by this new benchmark dataset for visual generation that is suitable for the modern era of large scale…
Introducing GPIC (Giant Permissive Image Corpus), a large-scale dataset of 100M VLM-captioned image-text pairs for training and 1M pairs for benchmarking, fully permissive for research and commercial use.
UltraFlux: Data-Model Co-Design for High-quality Native 4K Text-to-Image Generation across Diverse Aspect Ratios
UltraFlux introduces a data-model co-design approach for native 4K text-to-image generation across diverse aspect ratios, addressing positional encoding, VAE compression, and optimization challenges. It outperforms existing open-source baselines and matches proprietary models like Seedream 4.0.
UltraViT: Latency-Optimized On-device Vision Encoder for Large Vision-Language Models
UltraViT is a latency-optimized vision encoder for large vision-language models, designed for on-device deployment with a pyramidal architecture and a two-stage generative pre-training strategy, achieving state-of-the-art performance at 1.7x speed.
HiDream-ai/HiDream-O1-Image
HiDream-ai has open-sourced HiDream-O1-Image (8B), a unified image generative foundation model built on a Pixel-level Unified Transformer (UiT) that natively handles text-to-image, image editing, and subject-driven personalization at up to 2048×2048 resolution without external VAEs or disjoint text encoders. It debuted at #8 in the Artificial Analysis Text to Image Arena and is positioned as a leading open-weights text-to-image model.
HiDream-ai/HiDream-O1-Image-Dev
HiDream-ai has released HiDream-O1-Image-Dev, an 8B parameter open-source image generation model that uses a pixel-level unified transformer without external VAEs. It ranks #8 in the Artificial Analysis Text to Image Arena and supports high-resolution generation up to 2,048x2,048.