CivArchive
    The Thing (1982) - Cinematographic Style - v1.0
    NSFW
    Preview 29049202
    Preview 29049199
    Preview 29049196
    Preview 29049187
    Preview 29049188
    Preview 29049190
    Preview 29049200
    Preview 29049195
    Preview 29049198
    Preview 29049189
    Preview 29049193
    Preview 29081609
    Preview 29081610
    Preview 29084648
    Preview 29084953
    Preview 29085191
    Preview 29085883
    Preview 29086581

    The Thing (1982) is a sci-fi horror film directed by John Carpenter. Set in a remote Antarctic research station, the story follows a group of scientists who encounter a shape-shifting alien that can assimilate and imitate any living organism. As paranoia and mistrust grow, the team must figure out who is still human before the creature destroys them all. Known for its intense atmosphere, groundbreaking practical effects, and chilling isolation, The Thing has become a cult classic in the horror genre.

    ⚠️IMPORTANT: this version is mostly broken and is very hard to use. I intend to iterate on it. But I think it's already usable in some cases, these cases being: dark scenes, dark photography.

    Please also note this is a LoRA about the style of the movie, not the content specifically. For instance, you can't prompt "R.J. MacReady", at least in the current version.

    STRENGTH: around 1

    Prompting: verbose style will work better, take inspiration from the showcase.

    It is possible to find unexpected uses with experimentation if you feel adventurous.

    All the captions used in the training are included if you want to study them (not the images because they are copyrighted).

    Top 20 Nouns and Adjectives by Frequency:

    • man: 305

    • lit: 275

    • person: 274

    • dimly: 272

    • right: 268

    • visible: 244

    • jacket: 240

    • left: 233

    • background: 220

    • room: 218

    • scene: 205

    • dark: 179

    • snow: 168

    • environment: 166

    • light: 165

    • individual: 162

    • hair: 159

    • shirt: 154

    • closeup: 152

    • face: 133

    TRAINING INFO:

    500 movie stills captioned by cogvlm2-llama3

    Trained on ai-toolkit

    Description

    First release, very hard to use, probably bad choice of images and mistakes in captioning.

    FAQ

    LORA
    Flux.1 D

    Details

    Downloads
    187
    Platform
    CivitAI
    Platform Status
    Available
    Created
    9/12/2024
    Updated
    5/12/2026
    Deleted
    -

    Files

    the_thing_captions.zip

    Mirrors

    CivitAI (1 mirrors)

    the_thing_v1.safetensors

    Mirrors

    CivitAI (1 mirrors)

    Available On (1 platform)

    Same model published on other platforms. May have additional downloads or version variants.