Skip to content

Dataset · SilentAntagonist

450k vintage 20th-century photographs with captions

SilentAntagonist/vintage-photography-450k-high-quality-captions

A high-resolution collection of 20th-century analog photos, with detailed descriptions already generated for 20k of them.

What's inside

The dataset contains 450,000 film-shot 20th-century photographs, many in high resolution. Currently, English text descriptions generated by the InternVL2 26B model are ready for 20,000 images, while annotation of the rest is ongoing.

What you can do with it

  • LoRA training for generating realistic retro and film photographs
  • Fine-tuning text-to-image models on 20th-century aesthetics
  • Training classification and feature extraction models on archival photos
Tasks
Image classification, Image feature extraction, Text → image
Languages
English
Size
100K–1M rows
Downloads this month
138
Likes
36

How to load it

from datasets import load_dataset

ds = load_dataset("SilentAntagonist/vintage-photography-450k-high-quality-captions")