datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
3dfront_render_viewsRenderedTextThis dataset has been created by Stability AI and LAION.
This dataset contains 12 million 1024x1024 images of handwritten text written on a digital 3D sheet of paper generated using Blender geometry nodes and rendered using Blender Cycles. The text has varying font size, color, and rotation, and the paper was rendered under random lighting conditions.
Note that, the first 10 million examples are in the root folder of this dataset repository and the remaining 2 million are in ./remaining (due… See the full description on the dataset page: https://huggingface.co/datasets/wendlerc/RenderedText.cmp-v6-base108-renderolmocr-pre-rendered
olmOCR-bench Pre-Rendered
Pre-rendered PNG images of the olmOCR-bench benchmark dataset, ready for zero-setup evaluation of any OCR / vision model.
What This Is
The official olmOCR benchmark requires downloading 1,403 PDFs locally and rendering each page to a PNG image before sending it to a model. Every benchmark runner in the official repo does this same rendering step internally — see olmocr/data/renderpdf.py::render_pdf_to_base64png().
This dataset eliminates that… See the full description on the dataset page: https://huggingface.co/datasets/shhdwi/olmocr-pre-rendered.3dfront-render-viewsdifferentiable-render-camouflage-data
DRC CARLA Multi-Vehicle Camouflage Dataset
This repository releases the audited synthetic data and geometry assets used to
study whether one differentiable vehicle-camouflage generator transfers across
vehicle shapes. The release preserves the original split manifests, collection
protocols, audit records, and SHA-256 checksums. It is intended for reproducible
adversarial-robustness research, including the analysis of negative results.
Release contents… See the full description on the dataset page: https://huggingface.co/datasets/bailuyucha/differentiable-render-camouflage-data.rendered-sst2
Rendered SST-2
The Rendered SST-2 Dataset from Open AI.
Rendered SST2 is an image classification dataset used to evaluate the models capability on optical character recognition. This dataset was generated by rendering sentences in the Standford Sentiment Treebank v2 dataset.
This dataset contains two classes (positive and negative) and is divided in three splits: a train split containing 6920 images (3610 positive and 3310 negative), a validation split containing 872 images (444… See the full description on the dataset page: https://huggingface.co/datasets/nateraw/rendered-sst2.3dfront-render-diffuse3d-front-indoor-renders
Indoor Scene Renders from 3D-FRONT / 3D-FUTURE
20,240 photo-realistic indoor scene renders with instance segmentation, 6DoF
object poses, camera intrinsics and depth — rendered from the 3D-FRONT scene
layouts and 3D-FUTURE furniture models.
This is not a copy of the original 3D-FUTURE render set. It is a separate
render set built from the same assets. See Differences from the original below.
Why this exists
The 3D-FUTURE technical report describes 20,240 rendered… See the full description on the dataset page: https://huggingface.co/datasets/Spatial1ntelligence/3d-front-indoor-renders.IconStack-48M-Rendered-TrainObjaverse-XL-Rigged-Animated-Renders
Objaverse-XL Rigged & Animated — Renders
Visual companion to
Linzhan/Objaverse-XL-Rigged-Animated,
which holds the 7,373 rigged-and-animated GLB assets themselves. This repository holds only what
was rendered from them: a four-view video of every animation clip, and a rest-pose grid per asset.
They live apart from the assets because they are bulky and numerous — 10,355 clip folders — while
the asset repo stays a compact 7,373 GLBs plus two tables. Nothing here is needed to use… See the full description on the dataset page: https://huggingface.co/datasets/Linzhan/Objaverse-XL-Rigged-Animated-Renders.3dfront_rendereligible-scroll-atlas-renders
Get one mesh in about twenty seconds
curl -sO https://raw.githubusercontent.com/rodriguescarson/eligible-scroll-atlas/main/scripts/atlas.py
python atlas.py list --ink-pass # the 5 meshes that pass the pre-registered screen
python atlas.py ink PHerc0125 z10544_w020 --preview # a downsampled ink map, about 12 KB
python atlas.py get PHerc0125 z10544_w020 # the surface volume, 31 planes, plane 15 is the surface
from atlas import meshes… See the full description on the dataset page: https://huggingface.co/datasets/rodriguescarson/eligible-scroll-atlas-renders.objaverse_orbit_rendersText-Render-2M
Text Render 2M Dataset
A large-scale dataset containing 2 million text rendering image-text pairs for training generative models to improve text rendering performance.
Dataset Structure
image: Rendered text image in PNG format
text: Corresponding text content
file_name: Original filename
folder_id: Folder identifier
Usage
This dataset is designed for fine-tuning generative models to improve text rendering capabilities.
from datasets import load_dataset… See the full description on the dataset page: https://huggingface.co/datasets/PosterCraft/Text-Render-2M.rendered-wikipedia-english
Dataset Card for Team-PIXEL/rendered-wikipedia-english
Dataset Summary
This dataset contains the full English Wikipedia from February 1, 2018, rendered into images of 16x8464 resolution.
The original text dataset was built from a Wikipedia dump. Each example in the original text dataset contained the content of one full Wikipedia article with cleaning to strip markdown and unwanted sections (references, etc.). Each rendered example contains a subset of one full article.… See the full description on the dataset page: https://huggingface.co/datasets/Team-PIXEL/rendered-wikipedia-english.i1-rendered_text-tfrecordi1: A Simple and Fully Open Recipe for Strong Text-to-Image Models
Boya Zeng, Tianze Luo, Shu Pu, Jucheng Shen, Taiming Lu, Gabriel Sarch, Zhuang Liu
Princeton University
[arXiv][code][model][project page]
Overview
To prepare the dataset for training, we store the image-caption pairs as TFRecords.
This HuggingFace dataset contains the TFRecords corresponding to the rendered_text dataset at 256×256 resolution.
It also serves as an example of what a dataset processed using… See the full description on the dataset page: https://huggingface.co/datasets/i1-datasets/i1-rendered_text-tfrecord.dna_rendering_processed
DNA-Rendering-Processed Dataset
Project Page | Paper | Code | Model
To enable Diffuman4D model training, we meticulously process the DNA-Rendering dataset by recalibrating camera parameters, optimizing image color correction matrices (CCMs), predicting foreground masks, and estimating human skeletons.
To promote future research in the field of human-centric 3D/4D generation, we have open-sourced our re-annotated labels for the DNA-Rendering dataset in this repo, which includes… See the full description on the dataset page: https://huggingface.co/datasets/krahets/dna_rendering_processed.i1-rendered_text-512-resolution-1m-tfrecordi1: A Simple and Fully Open Recipe for Strong Text-to-Image Models
Boya Zeng, Tianze Luo, Shu Pu, Jucheng Shen, Taiming Lu, Gabriel Sarch, Zhuang Liu
Princeton University
[arXiv][code][model][project page]
Overview
To prepare the dataset for training, we store the image-caption pairs as TFRecords.
This HuggingFace dataset contains the TFRecords corresponding to the rendered_text dataset at 512×512 resolution. Concretely, we only retain raw images with a shorter edge of… See the full description on the dataset page: https://huggingface.co/datasets/i1-datasets/i1-rendered_text-512-resolution-1m-tfrecord.Objaverse_render_randomomnidocbench-render-compare
OmniDocBench Render-and-Compare
This dataset contains the rendered HTML reconstructions and comparison images produced
by a render-and-compare pipeline — a reference-free visual similarity evaluation
framework for OCR systems.
Overview
The pipeline processes each page of OmniDocBench through
a Qwen3.5-122B-A10B OCR model, renders the structured output back to a PNG via HTML
(reconstructed.png), and compares it against the original page scan (masked_original.png)
using… See the full description on the dataset page: https://huggingface.co/datasets/gt-free-ocr-metrics/omnidocbench-render-compare.rendered-bookcorpus
Dataset Card for Team-PIXEL/rendered-bookcorpus
Dataset Summary
This dataset is a version of the BookCorpus available at https://huggingface.co/datasets/bookcorpusopen with examples rendered as images with resolution 16x8464 pixels.
The original BookCorpus was introduced by Zhu et al. (2015) in Aligning Books and Movies: Towards Story-Like Visual Explanations by Watching Movies and Reading Books and contains 17868 books of various genres. The rendered BookCorpus was used… See the full description on the dataset page: https://huggingface.co/datasets/Team-PIXEL/rendered-bookcorpus.rendered-bookcorpus-bigramsprocgen-renderformer
Procgen RenderFormer Dataset
Procedurally generated indoor scenes with ground-truth path-traced renders and
precomputed 3-slat VAE latents, built for training RenderFormer-style
neural renderers. Each sample is one scene observed from 14 camera poses along
an orbit.
Configs
Config
Scenes
Samples (scene x frame)
Notes
main
~307,000
~4.3 M
primary training set
zoom
~84,000
~1.2 M
tighter framing variant
validation
~1,000
~14 K
held-out assets, not… See the full description on the dataset page: https://huggingface.co/datasets/eternity304/procgen-renderformer.rendered-bookcorpus-8x8BEAT_Rendered_Videosobjaverse_rendering_setrenderObjaverse_2d_renders
2D Image/Depth Rendering of Objaverse Dataset
In total, the rendered split contains 167,857 objects. The object ids are in the completed_renders.txt file. After unzipping, the image/depth renders are in the following folder strunture:
# e.g.,
000-000/000074a334c541878360457c672b6c2e
├── depth.zip
├── image.zip
├── metadata.json
└── transforms_train.json
Camera Intrinsics
72 views per-object, uniformly sampled on the upper hemisphere
Image dimensions: 400×400… See the full description on the dataset page: https://huggingface.co/datasets/ShapeSplats/Objaverse_2d_renders.RenderMatte-dataset
RenderMatte Dataset
This dataset is released as independent tar archives under data/, following the same layout style as Renz-7/new_matting.
Files
data/dataset_part_01.tar: 89254 files, 17251840000 bytes
data/dataset_part_02.tar: 26590 files, 17199759360 bytes
data/dataset_part_03.tar: 14974 files, 17190840320 bytes
data/dataset_part_04.tar: 17392 files, 17192970240 bytes
data/dataset_part_05.tar: 13178 files, 17189857280 bytes
data/dataset_part_06.tar: 9539… See the full description on the dataset page: https://huggingface.co/datasets/Renz-7/RenderMatte-dataset.