image
imagewidth (px)
384
1.47k
text
stringlengths
5
398
doom eternal, game concept art, veins and worms, muscular, crustacean exoskeleton, chiroptera head, chiroptera ears, mecha, ferocious, fierce, hyperrealism, fine details, artstation, cgsociety, zbrush, no background
a beautiful photorealistic painting of cemetery urbex unfinished building building industrial architecture nature abandoned by thomas cole, nature extraterrestial tron forest darkacademia thermal vision futuristic tokyo, archdaily, wallpaper, highly detailed, trending on artstation.
doom eternal, evangelion, game concept art, veins and worms, muscular, crustacean exoskeleton, chiroptera head, chiroptera ears, mecha, ferocious, fierce, hyperrealism, fine details, artstation, cgsociety, zbrush, no background
doom eternal, metro exodus, game concept art, veins and worms, muscular, crustacean exoskeleton, chiroptera head, chiroptera ears, mecha, ferocious, fierce, hyperrealism, fine details, artstation, cgsociety, zbrush, no background
industrial architecture by zack snyder, atlantis at dawn laser meadow wilderness sunlight neon noir architecture island sunset tundra futuristic alien cosmic hyperrealism fisheye at dusk assassin's creed, archdaily, wallpaper, highly detailed, trending on artstation.
a beautiful painting of chernobyl by nekro, pascal blanche, john harris, greg rutkowski, sin jong hun, moebius, simon stalenhag. in style of cg art. ray tracing. cel shading. hyper detailed. realistic. ue 5. maya. octane render.
doom eternal, metro exodus, game concept art, veins and worms, muscular, crustacean exoskeleton, chiroptera head, chiroptera ears, mecha, ferocious, fierce, hyperrealism, fine details, artstation, cgsociety, zbrush, no background
building by denise scott brown, tron tokyo anime at night alien flowers cgsociety nightvision at dusk nightsky sunlight dramatic lighting fisheye studio ghibli cyberpunk at dawn rainforest thermal vision thermal imaging reclaimed by nature darkacademia myst meadow retrowave landscape, archdaily, wallpaper, highly detailed, trending on artstation.
only with red, a red lizardman iattacked by technology, red computers wires, in a red laboratory, dorohedoro vibes, in the style of beksinski, parts by edward hopper, parts by michael sowa, parts by yue minjun, intricate composition, red by caravaggio, highly quality, highly detailed, masterpiece, red light, artstation, 4 k
a beautiful very detailed illustration of abandoned urbex unfinished building city nature industrial architecture architecture building spaceport by caspar david friedrich, scumm bar meadow nature synthwave, archdaily, wallpaper, highly detailed, trending on artstation.
a beautiful very detailed highly detailed building ranch by frank gehry, galactic darkacademia tron dramatic lighting studio ghibli at night alien cyberpunk nature forest sci - fi vaporwave universe retro synthwave nightvision elysian magic realism scumm bar made of glass, archdaily, wallpaper, highly detailed, trending on artstation.
abstract watercolor painting of hedonic scandinavian king house interior, in stone, straw roof, war lord, magical and traditional, winter, cinematic light, national romanticism by anders zorn, by greg rutkowski, by greg manchess
the great divide, digital painting, highly detailed, concept art, trending on artstation, epic composition, sharp focus, flawless render, masterpiece, volumetric lighting
a beautiful detailed painting of city abandoned industrial architecture building tomb nature urbex by jeanne gang, nightsky magic realism vaporwave laser otherworldly fisheye at fall sea universe synthwave landscape nature made of glass sunlight thermal vision infrared dramatic lighting, archdaily, wallpaper, highly detailed, trending on artstation.
divine homonculus, digital painting, highly detailed, concept art, trending on artstation, epic composition, sharp focus, flawless render, masterpiece, volumetric lighting
a beautiful very detailed render of nature unfinished building abandoned architecture building athenaeum urbex industrial architecture by john howe, matte painting san andreas forest otherworldly at night extraterrestial at dawn, archdaily, wallpaper, highly detailed, trending on artstation.
a beautiful detailed illustration of abandoned cottage by edward hopper, apocalyptic morning sun bladerunner 2 0 4 9 cyberpunk retrowave dramatic lighting hyperrealism retro futuristic cgsociety myst flowers, archdaily, wallpaper, highly detailed, trending on artstation.
cyber moai on easter island, digital painting, highly detailed, concept art, trending on artstation, epic composition, sharp focus, flawless render, masterpiece, volumetric lighting
5 5 mm macro photo of a woodpecker inside a house in an oak tree, looking through the window. dof. bokeh. by greg rutkowski. highly detailed 8 k. intricate. lifelike. soft light. cinematic processing
from the coast of north oregon, usa ( oc ) [ r / earthporn ]
hyperrealistic portrait of a philippine chilled character in a scenic environment, flowers, by beksinski zdzislaw, buchholz, quint, yoro sean, christensen james c.,
a beautiful painting of a retro futuristic stockholm, hyperdetailed, by simon stalenhag, artstation, pixiv, 8 k, octane render
a beautiful ultradetailed anime illustration of unknown backroom level nature by bjarke ingels, landscape sunlight laser studio ghibli infrared cyberpunk at night synthwave tokyo vaporwave neon noir magic realism, archdaily, wallpaper, highly detailed, trending on artstation.
the statue david from michelangelo holding a smartphone, scene is stonehenge ecclipse party
a beautiful painting of a retro futuristic stockholm, hyperdetailed, by simon stalenhag, cyberpunk, synthwave, digital illustration, concept art, wallpaper, artstation, pixiv, 8 k, octane render
a beautiful hyperdetailed render of house building architecture unfinished building by alvar aalto, neon noir universe flowers architecture island bladerunner 2 0 4 9 lightpaint junglepunk, archdaily, wallpaper, highly detailed, trending on artstation.
a cat wearing a fake mustache, rainbows, bright colors, psychedelic, fish eye lens, vector art, fantasy poster by helen huang and frank frazetta and salvador dali and norman rockwell,
a beautiful ultradetailed anime illustration of industrial architecture cavern abandoned urbex by april gornik, evil cyberpunk postcyberpunk extraterrestial water at spring bladerunner 2 0 4 9 azeroth landscape retro jupiter universe junglepunk, archdaily, wallpaper, highly detailed, trending on artstation.
portrait painting of a lost boy by cedric peyravernay and greg ruthkowski, in the style of dishonored concept art, concept design, trending on artstation
a screenshot of daniel radcliffe in the video game simpsons hit and run. 3 d rendering. unreal engine. amazing likeness. very detailed. cartoon caricature
charcoal painting of a girl sculpture, facing sideways, distorted, abstract, creepy, fine details, intricate, elegant, highly detailed, ray tracing, digital painting, artstation, concept art, art by artgerm and greg rutkowski and alphonse mucha, 8 k
a beautiful very detailed anime illustration of urbex abandoned city unfinished building hallway by camille pissarro, vice city san andreas assassin's creed retro crystal wilderness morning sun matte painting cyberpunk junglepunk liberty city, archdaily, wallpaper, highly detailed, trending on artstation.
a charming illustration of two young high school sweethearts set in a school setting, concept art, digital painting, trending on artstation hq, deviantart, pinterest, unreal engine 5,
hyperrealistic portrait of a philippine baby character in a scenic environment, flowers, by beksinski zdzislaw, buchholz, quint, yoro sean, christensen james c.,
symmetrical centered medium head - on portrait of maya ali as a divinity sorcerer goddess, matte painting concept art, pastel colours, official fanart behance hd artstation by jesper ejsing, by rhads and makoto shinkai and lois van baarle and ilya kuvshinov and rossdraws
mount epic nether - gorged greatwyrm screenshot world of warcraft mounts worldboss comics that looks like it is from borderlands by jesper ejsing, by rhads, makoto shinkai alena aenami artworks in 4 k beeple, by thomas kinkade heartstone league of legends dofus overwatch
a beautiful photorealistic illustration of greenhouse industrial architecture unfinished building by rosa bonheur, meadow forest at dusk vaporwave, archdaily, wallpaper, highly detailed, trending on artstation.
dark enlightenment, alchemy, albedo, volumetric dark natural light, oil on canvas, by jack kirby, zdzislaw beksinski, tom lovell, alex malveda, greg staples
a matte painting of a moon exploding in the sky, viewed from a european town street. giant bright explosion, at nighttime with a starry sky. concept art by jorge jacinto, annibale carracci, john constable. sparkly, dramatic, 4 k, angelic, divine. insanely detailed and intricate, hypermaximalist, elegant, ornate, super detailed.
predator from the movie predator doing entertainment for kids at a birthday party
a void beast with shadowy tendrils and piercing red eyes creeping through a white background, grunge, artstation, high quality, hyper detailed
a beautiful painting of chernobyl in the autumn by nekro and pascal blanche and syd mead and greg rutkowski and sin jong hun and moebius and simon stalenhag. in style of cg art. ray tracing. cel shading. hyper detailed. realistic. ue 5. maya. octane render.
a void colossus with shadowy tendrils and piercing red eyes creeping through a white background, grunge, artstation, high quality, hyper detailed
two time pilots wearing flight masks and red rick owens flight suits inside the glowing geometric rainbow portal to the sixth dimension by frank frazetta
unfinished building by santiago calatrava, cgsociety vaporwave flowers tron fisheye rainforest galactic elysian neon signs junglepunk cyberpunk tundra desert sunset morning sun tokyo, archdaily, wallpaper, highly detailed, trending on artstation.
a beautiful painting of chernobyl in the autumn by nekro and pascal blanche and syd mead and greg rutkowski and sin jong hun and moebius and simon stalenhag. in style of cg art. ray tracing. cel shading. hyper detailed. realistic. ue 5. maya. octane render.
a void colossus with shadowy tendrils and piercing red eyes creeping through a white background, grunge, artstation, high quality, hyper detailed
mount epic flying seats screenshot world of warcraft mounts worldboss fantasy comics that looks like it is from borderlands by jesper ejsing, by rhads, makoto shinkai alena aenami artworks in 4 k beeple, thomas kinkade heartstone league of legends dofus overwatch
symmetry!! a manufactured mystical aqua green , canal blue , bronze red of medieval epic random box cube levitate, floral!, elegant, highly detailed, magic surge, intricate, unreal engine 5, concept art , by greg rutkowski, loish, highly rendered, digital painting, artstation hq
a beautiful painting of chernobyl by nekro and pascal blanche and syd mead and greg rutkowski and sin jong hun and moebius and simon stalenhag. in style of cg art. ray tracing. cel shading. hyper detailed. realistic. ue 5. maya. octane render.
landscape, cyberpunk alien tree of life, cyberpunk, leds, bionics, augments, sci fi, horror, monstrous, highly detailed, complex, intricate, matte painting, cinematic, by rhads and mohrbacher and zdzislaw beksinski and akira
sketchbook pages copic markers, magnificent, close up, details, sharp focus, beauty girl asian, sexy, hyper detailed, insane details, intricate, elite, elegant, luxury, by ismail inceoglu dragan bibin hans thoma greg rutkowski alexandros pyromallis rene maritte illustrated, perfect face, fine details, realistic shaded, fine - face, pretty face
urbex by tom wright, at dusk steampunk cosmic morning sun atlantis elysian sunlight universe neon noir extraterrestial tron bioshock tundra myst at dawn matte painting retrowave poppy landscape assassin's creed rainforest fantasy evil, archdaily, wallpaper, highly detailed, trending on artstation.
landscape, cyberpunk alien tree of life, cyberpunk, leds, bionics, augments, sci fi, horror, monstrous, highly detailed, complex, intricate, matte painting, cinematic, by rhads and mohrbacher and zdzislaw beksinski and akira,
anime girl with muscles with boxing gloves, highly detailed, muscular, very beautiful face, digital art, pixiv fanbox, artstation, by greg rutkowski, wlop, miyazaki hayao
symmetry!! a manufactured mystical aqua silvery , canal purple , gilded colorful of medieval epic cube levitate, floral!, elegant, highly detailed, magic surge, intricate, unreal engine 5, concept art , by greg rutkowski, loish, highly rendered, digital painting, artstation hq
a void colossus with shadowy tendrils and piercing red eyes creeping through a white background, grunge, artstation, high quality, hyper detailed
a beautiful very detailed rendering of architecture building kitchen abandoned urbex unfinished building nature city industrial architecture by marina tabassum, sunlight wilderness cosmic bladerunner 2 0 4 9 nature morning sun landscape at night extraterrestial meadow universe, archdaily, wallpaper, highly detailed, trending on artstation.
symmetry!! a manufactured mystical aqua silver canal purple gilded colorful of medieval epic cube levitate design, floral!, elegant, highly detailed, magic surge, intricate, unreal engine 5, concept art , by greg rutkowski, loish, highly rendered, digital painting, artstation hq
a void colossus with shadowy tendrils and piercing red eyes creeping through a white background, grunge, artstation, high quality, hyper detailed
mount epic azure skitterfly screenshot world of warcraft mounts worldboss fantasy comics that looks like it is from borderlands by jesper ejsing, by rhads, makoto shinkai alena aenami artworks in 4 k beeple, thomas kinkade heartstone league of legends dofus overwatch
symmetry!! a manufactured mystical aqua silver canal purple gilded colorful of medieval epic cube levitate in the center display design, floral!, elegant, highly detailed, magic surge, intricate, unreal engine 5, concept art , by greg rutkowski, loish, highly rendered, digital painting, artstation hq
a beautiful very detailed render of tiny home nature architecture unfinished building city by william pereira, neon noir darkacademia reclaimed by nature uv light nightvision assassin's creed fisheye thermal imaging cosmic cyberpunk, archdaily, wallpaper, highly detailed, trending on artstation.
female demon in a medieval dungeon, glowing red eyes, stunning, detailed, by artgerm, by greg rutkowski, by luis royo, by pixar, by myazaki, gothic, final fantasy, fantasy, medieval
symmetry!! a manufactured mystical aqua silver canal purple gilded colorful of medieval epic cube levitate display design, floral!, elegant, highly detailed, magic surge, intricate, unreal engine 5, concept art , by greg rutkowski, loish, highly rendered, digital painting, display, artstation hq
mount epic battle gargon vrednic screenshot world of warcraft mounts worldboss fantasy comics that looks like it is from borderlands by jesper ejsing, by rhads, makoto shinkai alena aenami artworks in 4 k beeple, thomas kinkade heartstone league of legends dofus overwatch
spaceport on a vast desert planet with ships taking off, by ralph mcquarrie, digital painting, highly detailed
portrait of a cloaked succubus, evil, stunning, detailed, by artgerm, by greg rutkowski, by luis royo, by pixar, by myazaki, gothic, final fantasy, fantasy, medieval
a beautiful hyperdetailed anime illustration of arsenal building city nature architecture industrial architecture unfinished building urbex by claude lorrain, at night water at dusk cosmic crystal made of glass thermal vision mercury sunlight, archdaily, wallpaper, highly detailed, trending on artstation.
chained in neon chains, digital collage, decoupage, assemblage, photomontage, canvas texture, hight contrast, minimalist, contemporary art, punk art, photorealistic, portrait, expressionism, masterpiece, dynamic composition, spectacular quality, intricate oil details, broken glass
industrial spaceport on a vast desert planet with ships taking off, by ralph mcquarrie, digital painting, highly detailed
portrait of a cloaked female devil, evil, stunning, detailed, by artgerm, by greg rutkowski, by luis royo, by pixar, by myazaki, gothic, final fantasy, fantasy, medieval
medieval fantasy oil painting, realism, lifelike, portrait of a european dragon that is perched atop a snowy mountain peak, as a dnd characer, surrounded by brick castle fortresses, trending on artstation
industrial spaceport on a vast desert planet with ships taking off, by ralph mcquarrie, digital painting, highly detailed
dark enlightenment, full length body portrait of a sci - fi pilot, painted by zdzislaw beksinski, jack kirby, tom lovell, alex malveda, greg staples
two time generals wearing flight masks and red rick owens flight suits with their hands behind their backs inside the glowing geometric rainbow portal to the sixth dimension by frank frazetta
portrait of a cloaked female devil, evil, stunning, detailed, by artgerm, by greg rutkowski, by luis royo, by pixar, by myazaki, gothic, final fantasy, fantasy, medieval
a beautiful very detailed render of nature train station urbex architecture by camille pissarro, futuristic hyperrealism retro sunset otherworldly nightvision anime nature dieselpunk azeroth darkacademia synthwave thermal vision, archdaily, wallpaper, highly detailed, trending on artstation.
industrial spaceport on a vast desert planet with ships taking off, by ralph mcquarrie, digital painting, highly detailed
portrait of a cloaked female devil, evil, smiling, pointy teeth, stunning, detailed, by artgerm, by greg rutkowski, by luis royo, by pixar, by myazaki, gothic, final fantasy, fantasy, medieval
a beautiful hyperdetailed rendering of canyon urbex by stanley spencer, tron mercury flowers sunlight neon signs landscape infrared lightpaint elysian nature postcyberpunk, archdaily, wallpaper, highly detailed, trending on artstation.
portrait of a cloaked female devil, evil, laughing, pointy teeth, stunning, detailed, by artgerm, by greg rutkowski, by luis royo, by pixar, by myazaki, gothic, final fantasy, fantasy, medieval
mount epic proudmoore sea scout screenshot world of warcraft mounts worldboss fantasy comics that looks like it is from borderlands by jesper ejsing, by rhads, makoto shinkai alena aenami artworks in 4 k beeple, thomas kinkade heartstone league of legends dofus overwatch
voxel art bison
a dog doing weights. epic oil painting.
industrial spaceport on a vast city planet with ships taking off, by ralph mcquarrie, digital painting, highly detailed
portrait of a cloaked female devil, evil, ominous, luscious, pointy teeth, stunning, detailed, by artgerm, by greg rutkowski, by luis royo, by pixar, by myazaki, gothic, final fantasy, fantasy, medieval
a dog doing weights on fire. epic oil painting.
she chained in neon chains, digital collage, decoupage, assemblage, photomontage, canvas texture, hight contrast, minimalist, contemporary art, punk art, photorealistic, portrait, expressionism, masterpiece, dynamic composition, spectacular quality, intricate oil details, broken glass
industrial spaceport on coruscant with ships taking off, by ralph mcquarrie, digital painting, highly detailed
a beautiful ultradetailed rendering of architecture industrial architecture urbex housebarn nature by norman foster, at dawn sunlight bladerunner 2 0 4 9 myst magic realism laser cgsociety crystal tundra sunset steampunk san andreas vice city nightvision cyberpunk sci - fi at dusk, archdaily, wallpaper, highly detailed, trending on artstation.
portrait of a cloaked female devil, evil, ominous, luscious, stunning, detailed, by artgerm, by greg rutkowski, by luis royo, by pixar, by myazaki, gothic, final fantasy, fantasy, medieval
monsters playing board games, oil painting by justin gerard, deviantart, pixar, bright colors
c 3 po looks like nick nolte with spaghetti in his mouth
portrait of a cloaked female devil, evil, ominous, luscious, ginger hair, stunning, detailed, by artgerm, by greg rutkowski, by luis royo, by pixar, by myazaki, gothic, final fantasy, fantasy, medieval
monsters playing board games, oil painting by justin gerard, deviantart, pixar, bright colors
concept art of chewie by ralph mcquarrie, digital painting, highly detailed
overcook videogame in the style of hr giger
monsters playing board games, oil painting by justin gerard, deviantart, pixar, bright colors
she chained in neon chains, digital collage, decoupage, assemblage, photomontage, canvas texture, minimalist, contemporary art, punk art, photorealistic, portrait, expressionism, masterpiece, dynamic composition, spectacular quality, intricate oil details, broken glass

DiffusionDB-Pixelart

Dataset Summary

This is a subset of the DiffusionDB 2M dataset which has been turned into pixel-style art.

DiffusionDB is the first large-scale text-to-image prompt dataset. It contains 14 million images generated by Stable Diffusion using prompts and hyperparameters specified by real users.

DiffusionDB is publicly available at 🤗 Hugging Face Dataset.

Supported Tasks and Leaderboards

The unprecedented scale and diversity of this human-actuated dataset provide exciting research opportunities in understanding the interplay between prompts and generative models, detecting deepfakes, and designing human-AI interaction tools to help users more easily use these models.

Languages

The text in the dataset is mostly English. It also contains other languages such as Spanish, Chinese, and Russian.

Subset

DiffusionDB provides two subsets (DiffusionDB 2M and DiffusionDB Large) to support different needs. The pixelated version of the data was taken from the DiffusionDB 2M and has 2000 examples only.

Subset Num of Images Num of Unique Prompts Size Image Directory Metadata Table
DiffusionDB-pixelart 2k ~1.5k ~1.6GB images/ metadata.parquet

Images in DiffusionDB-pixelart are stored in png format.

Dataset Structure

We use a modularized file structure to distribute DiffusionDB. The 2k images in DiffusionDB-pixelart are split into folders, where each folder contains 1,000 images and a JSON file that links these 1,000 images to their prompts and hyperparameters.

# DiffusionDB 2k
./
├── images
│   ├── part-000001
│   │   ├── 3bfcd9cf-26ea-4303-bbe1-b095853f5360.png
│   │   ├── 5f47c66c-51d4-4f2c-a872-a68518f44adb.png
│   │   ├── 66b428b9-55dc-4907-b116-55aaa887de30.png
│   │   ├── [...]
│   │   └── part-000001.json
│   ├── part-000002
│   ├── part-000003
│   ├── [...]
│   └── part-002000
└── metadata.parquet

These sub-folders have names part-0xxxxx, and each image has a unique name generated by UUID Version 4. The JSON file in a sub-folder has the same name as the sub-folder. Each image is a PNG file (DiffusionDB-pixelart). The JSON file contains key-value pairs mapping image filenames to their prompts and hyperparameters.

Data Instances

For example, below is the image of ec9b5e2c-028e-48ac-8857-a52814fd2a06.png and its key-value pair in part-000001.json.

{
  "ec9b5e2c-028e-48ac-8857-a52814fd2a06.png": {
    "p": "doom eternal, game concept art, veins and worms, muscular, crustacean exoskeleton, chiroptera head, chiroptera ears, mecha, ferocious, fierce, hyperrealism, fine details, artstation, cgsociety, zbrush, no background ",
    "se": 3312523387,
    "c": 7.0,
    "st": 50,
    "sa": "k_euler"
    },
}

Data Fields

  • key: Unique image name
  • p: Text

Dataset Metadata

To help you easily access prompts and other attributes of images without downloading all the Zip files, we include a metadata table metadata.parquet for DiffusionDB-pixelart.

Two tables share the same schema, and each row represents an image. We store these tables in the Parquet format because Parquet is column-based: you can efficiently query individual columns (e.g., prompts) without reading the entire table.

Below are three random rows from metadata.parquet.

image_name prompt part_id seed step cfg sampler width height user_name timestamp image_nsfw prompt_nsfw
0c46f719-1679-4c64-9ba9-f181e0eae811.png a small liquid sculpture, corvette, viscous, reflective, digital art 1050 2026845913 50 7 8 512 512 c2f288a2ba9df65c38386ffaaf7749106fed29311835b63d578405db9dbcafdb 2022-08-11 09:05:00+00:00 0.0845108 0.00383462
a00bdeaa-14eb-4f6c-a303-97732177eae9.png human sculpture of lanky tall alien on a romantic date at italian restaurant with smiling woman, nice restaurant, photography, bokeh 905 1183522603 50 10 8 512 768 df778e253e6d32168eb22279a9776b3cde107cc82da05517dd6d114724918651 2022-08-19 17:55:00+00:00 0.692934 0.109437
6e5024ce-65ed-47f3-b296-edb2813e3c5b.png portrait of barbaric spanish conquistador, symmetrical, by yoichi hatakenaka, studio ghibli and dan mumford 286 1713292358 50 7 8 512 640 1c2e93cfb1430adbd956be9c690705fe295cbee7d9ac12de1953ce5e76d89906 2022-08-12 03:26:00+00:00 0.0773138 0.0249675

Metadata Schema

metadata.parquet schema:

Column Type Description
image_name string Image UUID filename.
text string The text prompt used to generate this image.

Warning Although the Stable Diffusion model has an NSFW filter that automatically blurs user-generated NSFW images, this NSFW filter is not perfect—DiffusionDB still contains some NSFW images. Therefore, we compute and provide the NSFW scores for images and prompts using the state-of-the-art models. The distribution of these scores is shown below. Please decide an appropriate NSFW score threshold to filter out NSFW images before using DiffusionDB in your projects.

Data Splits

For DiffusionDB-pixelart, we split 2k images into folders where each folder contains 1,000 images and a JSON file.

Loading Data Subsets

DiffusionDB is large! However, with our modularized file structure, you can easily load a desirable number of images and their prompts and hyperparameters. In the example-loading.ipynb notebook, we demonstrate three methods to load a subset of DiffusionDB. Below is a short summary.

Method 1: Using Hugging Face Datasets Loader

You can use the Hugging Face Datasets library to easily load prompts and images from DiffusionDB. We pre-defined 16 DiffusionDB subsets (configurations) based on the number of instances. You can see all subsets in the Dataset Preview.

import numpy as np
from datasets import load_dataset

# Load the dataset with the `2k_random_1k` subset
dataset = load_dataset('jainr3/diffusiondb-pixelart', '2k_random_1k')

Dataset Creation

Curation Rationale

Recent diffusion models have gained immense popularity by enabling high-quality and controllable image generation based on text prompts written in natural language. Since the release of these models, people from different domains have quickly applied them to create award-winning artworks, synthetic radiology images, and even hyper-realistic videos.

However, generating images with desired details is difficult, as it requires users to write proper prompts specifying the exact expected results. Developing such prompts requires trial and error, and can often feel random and unprincipled. Simon Willison analogizes writing prompts to wizards learning “magical spells”: users do not understand why some prompts work, but they will add these prompts to their “spell book.” For example, to generate highly-detailed images, it has become a common practice to add special keywords such as “trending on artstation” and “unreal engine” in the prompt.

Prompt engineering has become a field of study in the context of text-to-text generation, where researchers systematically investigate how to construct prompts to effectively solve different down-stream tasks. As large text-to-image models are relatively new, there is a pressing need to understand how these models react to prompts, how to write effective prompts, and how to design tools to help users generate images. To help researchers tackle these critical challenges, we create DiffusionDB, the first large-scale prompt dataset with 14 million real prompt-image pairs.

Source Data

Initial Data Collection and Normalization

We construct DiffusionDB by scraping user-generated images on the official Stable Diffusion Discord server. We choose Stable Diffusion because it is currently the only open-source large text-to-image generative model, and all generated images have a CC0 1.0 Universal Public Domain Dedication license that waives all copyright and allows uses for any purpose. We choose the official Stable Diffusion Discord server because it is public, and it has strict rules against generating and sharing illegal, hateful, or NSFW (not suitable for work, such as sexual and violent content) images. The server also disallows users to write or share prompts with personal information.

Who are the source language producers?

The language producers are users of the official Stable Diffusion Discord server.

Annotations

The dataset does not contain any additional annotations.

Annotation process

[N/A]

Who are the annotators?

[N/A]

Personal and Sensitive Information

The authors removed the discord usernames from the dataset. We decide to anonymize the dataset because some prompts might include sensitive information: explicitly linking them to their creators can cause harm to creators.

Considerations for Using the Data

Social Impact of Dataset

The purpose of this dataset is to help develop better understanding of large text-to-image generative models. The unprecedented scale and diversity of this human-actuated dataset provide exciting research opportunities in understanding the interplay between prompts and generative models, detecting deepfakes, and designing human-AI interaction tools to help users more easily use these models.

It should note that we collect images and their prompts from the Stable Diffusion Discord server. The Discord server has rules against users generating or sharing harmful or NSFW (not suitable for work, such as sexual and violent content) images. The Stable Diffusion model used in the server also has an NSFW filter that blurs the generated images if it detects NSFW content. However, it is still possible that some users had generated harmful images that were not detected by the NSFW filter or removed by the server moderators. Therefore, DiffusionDB can potentially contain these images. To mitigate the potential harm, we provide a Google Form on the DiffusionDB website where users can report harmful or inappropriate images and prompts. We will closely monitor this form and remove reported images and prompts from DiffusionDB.

Discussion of Biases

The 14 million images in DiffusionDB have diverse styles and categories. However, Discord can be a biased data source. Our images come from channels where early users could use a bot to use Stable Diffusion before release. As these users had started using Stable Diffusion before the model was public, we hypothesize that they are AI art enthusiasts and are likely to have experience with other text-to-image generative models. Therefore, the prompting style in DiffusionDB might not represent novice users. Similarly, the prompts in DiffusionDB might not generalize to domains that require specific knowledge, such as medical images.

Other Known Limitations

Generalizability. Previous research has shown a prompt that works well on one generative model might not give the optimal result when used in other models. Therefore, different models can need users to write different prompts. For example, many Stable Diffusion prompts use commas to separate keywords, while this pattern is less seen in prompts for DALL-E 2 or Midjourney. Thus, we caution researchers that some research findings from DiffusionDB might not be generalizable to other text-to-image generative models.

Additional Information

Dataset Curators

DiffusionDB is created by Jay Wang, Evan Montoya, David Munechika, Alex Yang, Ben Hoover, Polo Chau.

Licensing Information

The DiffusionDB dataset is available under the CC0 1.0 License. The Python code in this repository is available under the MIT License.

Citation Information

@article{wangDiffusionDBLargescalePrompt2022,
  title = {{{DiffusionDB}}: {{A}} Large-Scale Prompt Gallery Dataset for Text-to-Image Generative Models},
  author = {Wang, Zijie J. and Montoya, Evan and Munechika, David and Yang, Haoyang and Hoover, Benjamin and Chau, Duen Horng},
  year = {2022},
  journal = {arXiv:2210.14896 [cs]},
  url = {https://arxiv.org/abs/2210.14896}
}

Contributions

If you have any questions, feel free to open an issue or contact the original author Jay Wang.

Downloads last month
165
Edit dataset card

Models trained or fine-tuned on jainr3/diffusiondb-pixelart