I will post celebrity photo Datasets here that were collected by me and captioned by GPT-4 Vision. The cover images are not in the dataset but are images I generated with Loras I have trained. Currently there are 8572 photos on 52 different celebrities with captions in total. I will update these numbers when I post the newest version :
Adriana Lima : 150 HQ Images, uncropped and captioned by GPT-4 Vision
Alba Baptista : 110 HQ Images, uncropped and captioned by GPT-4 Vision
Alexandra Daddario : 91 HQ Images, uncropped and captioned by GPT-4 Vision
Alexandra Saint Mleux : 40 Images, uncropped and captioned by GPT-4 Vision
Ana de Armas : 254 HQ Images, uncropped and captioned by GPT-4 Vision
Barbara Palvin : 276 HQ Images, uncropped and captioned by GPT-4 Vision
Barbara Palvin : 52 Images, uncropped and captioned by GPT-4 Vision
Bryce Dallas Howard : 165 HQ Images, uncropped and captioned by GPT-4 Vision
Cailee Spaeny : 100 HQ Images, uncropped and captioned by GPT-4 Vision
Courtney Eaton : 178 HQ Images, uncropped and captioned by GPT-4 Vision
Dua Lipa : 134 HQ Images, uncropped and captioned by GPT-4 Vision
Elizabeth Lail : 216 HQ Images, uncropped and captioned by GPT-4 Vision
Elizabeth Olsen : 199 HQ Images, uncropped and captioned by GPT-4 Vision
Elizabeth Olsen (2012) : 220 HQ Images, uncropped and captioned by GPT-4 Vision
Elle Fanning : 182 HQ Images, uncropped and captioned by GPT-4 Vision
Emilia Clarke : 309 HQ Images, uncropped and captioned by GPT-4 Vision
Emma Watson : 175 HQ Images, uncropped and captioned by GPT-4 Vision
Florence Pugh : 302 HQ Images, uncropped and captioned by GPT-4 Vision
Freida Pinto : 118 HQ Images, uncropped and captioned by GPT-4 Vision
Gal Gadot : 183 HQ Images, uncropped and captioned by GPT-4 Vision
Gigi Hadid : 239 HQ Images, uncropped and captioned by GPT-4 Vision
Hailee Steinfeld : 92 HQ Images, uncropped and captioned by GPT-4 Vision
Isabela Merced : 181 HQ Images, uncropped and captioned by GPT-4 Vision
Jenna Coleman : 202 HQ Images, uncropped and captioned by GPT-4 Vision
Jenna Ortega : 223 HQ Images, uncropped and captioned by GPT-4 Vision
Jennifer Connelly (1990s) : 132 Images, uncropped and captioned by GPT-4 Vision
Kaya Scodelario : 111 HQ Images, uncropped and captioned by GPT-4 Vision
Kaya Scodelario : 88 HQ Images, uncropped and captioned by GPT-4 Vision
Kristine Froseth : 93 Images, uncropped and captioned by GPT-4 Vision
Kurodahana : 200 Images, uncropped and captioned by GPT-4 Vision
Leighton Meester : 240 HQ Images, uncropped and captioned by GPT-4 Vision
Liza Soberano : 251 Images, uncropped and captioned by GPT-4 Vision
Madelaine Petsch : 92 Images, uncropped and captioned by GPT-4 Vision
Madison Beer : 123 HQ Images, uncropped and captioned by GPT-4 Vision
Margot Robbie : 272 HQ Images, uncropped and captioned by GPT-4 Vision
Miranda Kerr : 160 HQ Images, uncropped and captioned by GPT-4 Vision
Natalie Dormer : 196 HQ Images, uncropped and captioned by GPT-4 Vision
Natalie Portman : 137 HQ Images, uncropped and captioned by GPT-4 Vision
Olivia Cooke : 163 HQ Images, uncropped and captioned by GPT-4 Vision
Olivia Wilde : 116 HQ Images, uncropped and captioned by GPT-4o
Rachel McAdams : 96 HQ Images, uncropped and captioned by GPT-4 Vision
Sadie Sink : 212 HQ Images, uncropped and captioned by GPT-4 Vision
Saoirse Ronan : 160 HQ Images, uncropped and captioned by GPT-4 Vision
Sydney Sweeney : 165 HQ Images, uncropped and captioned by GPT-4 Vision
TinaKitten : 63 Images, uncropped and captioned by GPT-4 Vision
Tingting Lai : 90 Images, uncropped and captioned by GPT-4 Vision
Valkyrae : 210 Images, uncropped and captioned by GPT-4 Vision
Vanessa Merrell : 268 Images, uncropped and captioned by GPT-4 Vision
Vanessa Merrell : 40 Images, uncropped and captioned by GPT-4 Vision
Yuval Gonen : 64 Images, uncropped and captioned by GPT-4 Vision
Zendaya Coleman : 80 HQ Images, uncropped and captioned by GPT-4 Vision
Zoë Kravitz : 129 HQ Images, uncropped and captioned by GPT-4 Vision
The captioning prompt was : "For the given image caption in full sentences objectively in moderate detail (but do not describe the intrinsic characteristics of a person like age, ethnicity or their features) 'A [closeup portrait photo, full body photo, half body photo, etc.] of a woman.[Describe her hair length and style in detail. Mention if she has makeup on and describe the makeup and also describe earrings or jewlery she is wearing. Describe in detail the direction she is looking i.e. away from camera, to the side etc .Describe her clothing in detail. Lastly describe the background.' Remember to use an objective tone."
Description
200 Photos of @kurodahana with GPT-4 Vision captions
FAQ
Comments (18)
Fuck yeah another dataset and even with captions, I'll 5 star rate this even if I won't use it because WE NEED MORE DATASETS ON CIVITAI DARNIT
Pls do Gemma Atkinson and Katie McGrath :D
Maisie williams please !
lol ya'll really want Civitai taken down...
That is amazing. Hope it will be extremely helpfull when the next model after SDXL is released. Or even for SDXL and different base models.
Any plans on making the same for some big pornostars like Sasha Grey?
Thanks a lot for these! Definitely agree that it'd be great to see more HQ datasets so those w/ good resources (or $ for runpod etc) can just plug & play to enrich their trained models.
Curious if not including the name in the caption was intentional, or just to stay true to 100% GPT4v's outputs? There are caption apps/repos that will quickly add or swap tags like the the celebrities name. Thx.
may I ask how do u caption images with gpt 4 vision? It comes with the 20 dollar monthly subscription right?
Jennifer Connelly <3
Eiza Gonzales? =)
a one question, gpt4 vision or api, apply filter nsfw? i want caption some poses
Nice work
When you have the chance, could you create one of Mia Goth?
is it possible to request Monica Bellucci and natalie dormer?
Could you add Brie Larson, Daisy Ridley and Lauren Mayberry?
Thank you for these. Sourcing images is always a huge hassle when experimenting with training concepts. Having them pre-captioned is a bonus!
Appreciate you for sharing! I'm trying to set up a workflow for curating training data and just having a data set to train against is so helpful and convenient.
This is pretty nice, but I think it's against Civitai rules, since those are not AI generated, and you do not own those photos to upload them here...
You can use them from training, sure, but I don't think you can upload them here.
Can we expect an article soon?
I need u version of Belle Delphine😅 please😍
Looks like we don't have an active mirror for this file right now.
CivArchive is a community-maintained index — we catalog mirrors that volunteers upload to HuggingFace, torrents, and other public hosts. Looks like no one has uploaded a copy of this file yet.
Some files do get recovered over time through contributions. If you're looking for this one, feel free to ask in Discord, or help preserve it if you have a copy.