Pose Catalog

Taxonomy

27 controlled-vocabulary dimensions across 7 facets. Every value in the open dataset draws from this vocabulary.

Pose structure

7 dimensions · v1.0.0 · CC-BY-4.0

Structural classification of erotic poses and sex positions as they appear in adult video and print. Covers the position family (solo / one-on-one / group), the primary act, the partner arrangement, penetration type, facing, dynamics, and the surface or support used. The arrangement sub-groups follow the widely-used structural taxonomy of named positions (penetrating-partner-on-top, from-behind, receiving-partner-on-top, seated/kneeling, standing, side-lying, inverted) documented in reference works such as the Wikipedia 'List of sex positions' and Wikidata's 'sex position' (Q8394) class; all bucket definitions here are original wording. Reference-grade descriptors intended for content tagging, search facets, and text-to-image prompt conditioning. Adults only.

Dimensions

  • Position family

    categorical

    Highest-level structural grouping by number of participants.

    Top-level partition by participant count, aligned with the group-vs-pair distinction used across sexology reference works.

    Values (3)

    • soloSoloA single participant; self-stimulation or display posing.
    • one_on_oneOne-on-oneTwo participants.
    • groupGroupThree or more participants.
  • Primary act

    categorical

    The dominant sexual act that defines the position. A position may involve more than one act; this records the defining one.

    Aligned with the exclusively-penetrative / non-exclusively-penetrative / non-penetrative partition used in named-position reference taxonomies.

    Values (7)

    • vaginal_penetrationVaginal penetrationPenile or object vaginal penetration is the defining act.
    • anal_penetrationAnal penetrationPenile or object anal penetration is the defining act.
    • oralOralOral-genital or oral-anal contact is the defining act.
    • manualManualManual stimulation (hand/finger) is the defining act.
    • non_penetrativeNon-penetrativeBody-to-body contact without penetration (e.g. frottage, intercrural, tribadism).
    • display_soloSolo displaySolo posing or self-stimulation for the camera; no partner contact.
    • mixedMixedMultiple simultaneous acts of comparable prominence (common in group positions).
  • Partner arrangement

    categorical

    The spatial arrangement of the participants relative to one another.

    Structural sub-groups mirror the named-position organisation in the Wikipedia 'List of sex positions' (penetrating-on-top, from-behind, receiving-on-top, sitting/kneeling, standing, less-common/inverted); wording original.

    Values (11)

    • penetrating_on_topPenetrating partner on topPenetrating partner above the receiving partner (e.g. missionary family).
    • receiving_on_topReceiving partner on topReceiving partner astride or above (e.g. cowgirl / woman-on-top family).
    • from_behindFrom behindPenetration or contact from the rear (e.g. doggy style, prone bone, spoons).
    • seatedSeatedOne or both partners seated (e.g. lotus, lap position).
    • kneelingKneelingOne or both partners kneeling.
    • standingStandingOne or both partners standing, possibly with support.
    • side_lyingSide-lyingBoth partners on their sides (e.g. spoons, lateral).
    • invertedInvertedOne partner partly or fully inverted (e.g. piledriver, suspended).
    • head_to_toeHead-to-toe (mutual)Reciprocal, oriented head-to-toe (e.g. 69).
    • over_faceOver the faceOne partner positioned above the other's face (e.g. facesitting).
    • solo_arrangementSoloSingle participant; no partner arrangement.
  • Penetration type

    categorical

    Which orifices are penetrated and by how many partners, where applicable.

    Descriptive enumeration of penetration configurations, including the multi-penetration terms catalogued on Wikidata.

    Values (9)

    • noneNoneNo penetration.
    • vaginalVaginalSingle vaginal penetration.
    • analAnalSingle anal penetration.
    • oralOralOral penetration.
    • manual_insertionManual insertionHand or finger insertion.
    • double_vaginalDouble vaginalTwo simultaneous vaginal penetrations.
    • double_analDouble analTwo simultaneous anal penetrations.
    • double_penetrationDouble penetration (vaginal + anal)Simultaneous vaginal and anal penetration by different partners.
    • multipleMultiple / otherThree or more simultaneous penetrations, or a combination not otherwise listed.
  • Facing

    categorical

    Orientation of participants relative to each other.

    Relative orientation descriptor used in position taxonomies.

    Values (5)

    • face_to_faceFace to facePartners face each other.
    • front_to_backFront to backOne partner's front to the other's back.
    • head_to_toeHead to toeReversed, reciprocal orientation.
    • side_by_sideSide by sidePartners aligned laterally.
    • variableVariableOrientation varies or is not fixed.
  • Dynamics

    categorical

    The characteristic motion of the position, relevant to video vs. still-image depiction.

    Motion descriptor for distinguishing static hold poses from active positions in moving-image media.

    Values (5)

    • static_holdStatic holdA held pose with little movement; typical of print stills.
    • thrustingThrustingReciprocating penetrative motion.
    • grindingGrindingCircular or rocking contact motion.
    • ridingRidingReceiving partner drives vertical motion from above.
    • bobbingBobbingRhythmic oral motion.
  • Support / surface

    categorical

    The surface or means of support the position relies on.

    Aligned with the 'furniture and apparatus' distinctions drawn in named-position references.

    Values (8)

    • bed_flatFlat surface / bedHorizontal surface such as a bed or floor.
    • edge_of_surfaceEdge of surfaceOne partner at the edge of a bed or table.
    • against_wallAgainst a wallWall used for bracing, usually standing.
    • standing_unsupportedStanding, unsupportedStanding without external bracing.
    • seated_furnitureSeated on furnitureChair, couch, or similar.
    • partner_supportedPartner-supportedOne partner bears the other's weight (e.g. carried).
    • suspendedSuspendedOne partner suspended or largely off any surface.
    • apparatusApparatusPurpose-built furniture or apparatus is central to the position.
References (4)
  1. Wikidata. Sex position (Q8394). https://www.wikidata.org/wiki/Q8394 (structured data CC0).
  2. Wikipedia contributors. List of sex positions. https://en.wikipedia.org/wiki/List_of_sex_positions (structure referenced; wording original).
  3. Vatsyayana. Kama Sutra (public domain) — historical named-position source.
  4. Danbooru tag group: posture and sex_acts — folksonomy structure for position and participant-composition axes.

Participant composition

4 dimensions · v1.0.0 · CC-BY-4.0

The number and apparent-gender make-up of the participants in a pose, and their roles. Adult media conventionally encodes multi-person composition with letter codes (M = apparent male, F = apparent female), e.g. MMF or FFM; that convention is reproduced here for interoperability with existing adult-industry and booru-style tagging. Codes describe apparent on-camera presentation only and imply nothing about identity. This axis is orthogonal to pose structure: the same structural position can occur across several compositions. Adults only.

Dimensions

  • Participant count

    ordinal

    Number of participants in the depiction.

    Direct count of participants depicted.

    Values (4)

    • oneOneSolo.
    • twoTwoA pair.
    • threeThreeThreesome.
    • four_plusFour or moreFour or more participants.
  • Composition code

    categorical

    Apparent-gender make-up of the participants, in the conventional letter-code form.

    Adult-industry / booru letter-code convention (M = apparent male, F = apparent female); describes apparent presentation only.

    Values (11)

    • solo_fSolo (F)One apparent-female participant.
    • solo_mSolo (M)One apparent-male participant.
    • MFM/FOne apparent-male and one apparent-female participant.
    • FFF/FTwo apparent-female participants.
    • MMM/MTwo apparent-male participants.
    • MMFMMFTwo apparent-male, one apparent-female.
    • FFMFFMTwo apparent-female, one apparent-male.
    • MFMMFMMale-female-male arrangement (positional variant of MMF).
    • FFFFFFThree apparent-female participants.
    • MMMMMMThree apparent-male participants.
    • group_mixedGroup (mixed)Four or more, mixed or unspecified make-up.
  • Receiving partners

    ordinal

    How many participants are in the receiving role.

    Count of partners in the receiving role in the depicted act.

    Values (4)

    • r00
    • r11
    • r22
    • r3_plus3+
  • Penetrating partners

    ordinal

    How many participants are in the penetrating role.

    Count of partners in the penetrating role in the depicted act.

    Values (4)

    • p00
    • p11
    • p22
    • p3_plus3+
References (2)
  1. Danbooru tag group: sex_acts — participant-composition letter codes (MMF/FFM/…) and solo/duo/group tags.
  2. e621 tagging system — solo / duo / group participant tags.

Presentation & medium

5 dimensions · v1.0.0 · CC-BY-4.0

How a pose is presented to the viewer: the source medium (adult video vs. print/magazine), the camera framing commonly associated with it, its physical difficulty, its prevalence, and an intensity tier used for compliance gating. The intensity tier mirrors the three-level model already used in the platform's generation pose library (illustrative / suggestive / explicit) so that downstream consumers can filter by the same contract. Adults only.

Dimensions

  • Source medium

    categorical

    The medium in which the pose characteristically appears.

    Distinguishes moving-image (video) from still-image (print/magazine) depiction conventions.

    Values (4)

    • videoVideoPrimarily a moving-image position (motion is intrinsic).
    • magazineMagazine / printPrimarily a still-image display pose.
    • bothBothCommon in both video and print.
    • illustrativeIllustrativeChiefly a reference or diagrammatic depiction.
  • Common camera framing

    categorical

    The camera framing most characteristic of the pose. A pose may support several; this records the typical primary framings.

    Cinematographic framing descriptors as applied in adult media production.

    Values (8)

    • pov_penetratingPOV (penetrating)Point-of-view from the penetrating partner.
    • pov_receivingPOV (receiving)Point-of-view from the receiving partner.
    • side_profileSide profileLateral full-body framing.
    • overheadOverheadFrom above.
    • low_angleLow angleFrom below.
    • rearRearFrom behind the receiving partner.
    • close_upClose-upTight framing on the point of contact.
    • wide_full_bodyWide / full bodyBoth participants fully in frame.
  • Physical difficulty

    ordinal

    How physically demanding the pose is (strength, flexibility, balance).

    Qualitative rating of strength, flexibility, and balance demand.

    Values (4)

    • lowLowComfortable and sustainable for most people.
    • moderateModerateSome flexibility or stamina required.
    • highHighNotable flexibility, strength, or balance required.
    • advanced_athleticAdvanced / athleticDemanding; often performed for camera rather than comfort.
  • Prevalence in adult media

    ordinal

    How commonly the pose appears in adult video and print.

    Qualitative estimate of how frequently the position is depicted.

    Values (4)

    • very_commonVery common
    • commonCommon
    • occasionalOccasional
    • nicheNiche
  • Intensity tier (compliance)

    ordinal

    Compliance-gating tier. Consumers may filter to suggestive-and-below where an explicit surface is not authorised.

    Mirrors the platform generation pose-library tier model (Sfw/Suggestive/Explicit).

    Values (3)

    • illustrativeIllustrativeClothed or diagrammatic; safe for general reference.
    • suggestiveSuggestiveImplied or partially clothed; no explicit act depicted.
    • explicitExplicitAdult/explicit act depicted. Gate behind an age-verified surface.
References (2)
  1. Internal: LatinaCastings.Services PoseSceneLibrary PoseTier (Sfw/Suggestive/Explicit) — the tier contract mirrored here.
  2. Adult-media production framing conventions (POV, side, overhead, rear).

Keypoint schema

1 dimensions · v1.2.0 · CC-BY-4.0

Specification for the per-participant pose keypoints shipped with the catalog. Keypoints use the 17-joint COCO Keypoint layout (the de-facto standard for 2D human pose), so the catalog's named positions can be consumed by any pipeline already built around COCO. Coordinates are IDEALISED, SCHEMATIC TEMPLATES — not motion-captured or image-derived. Each catalog participant references a base posture from the base-skeleton library and a placement transform (translate, scale, rotate, flip); the build step composes the final scene-space keypoints from those. This provides the symbolic-to-geometric bridge that named-position vocabularies (Wikidata, Wikipedia) lack and that keypoint datasets (COCO, MPII) provide only as unnamed coordinates.

Dimensions

  • Base posture

    categorical

    The canonical body posture a participant is composed from. Scene arrangement is produced by the placement transform.

    Canonical schematic skeletons defined in keypoint-base-skeletons.json (v1.0.0).

    Values (17)

    • standingStandingUpright, weight on both feet. Profile, facing +x.
    • kneelingKneeling (upright)Upright torso, weight on the knees, shins back along the ground. Profile, facing +x.
    • kneeling_sitKneeling, sitting on heelsKneeling with the pelvis lowered onto the heels. Profile, facing +x.
    • seatedSeatedSitting on the ground, knees raised. Profile, facing +x.
    • seated_lapSeated on lapUpright seat on a partner's lap, legs wrapped forward around them. Profile, facing +x.
    • supineSupineLying on the back, body horizontal, face up. Profile, head-left.
    • supine_legs_upSupine, legs raisedOn the back with hips flexed, knees up and ankles raised. Profile, head-left.
    • proneProneLying face down, body horizontal, forearms under. Profile, head-left.
    • plank_overPlank overBody inclined over a lying partner: arms extended to the ground, knees down, hips low. Profile, head-left.
    • all_foursAll foursOn hands and knees, torso horizontal. Profile, facing +x.
    • side_lyingSide-lyingLying on one side, knees drawn forward in a shallow Z. Profile, head-left.
    • astrideAstride (kneel-straddle)Kneeling straddle over a horizontal partner, torso upright, shins back. Profile, facing +x.
    • invertedInverted (shoulder-stand)Shoulders and head on the ground, hips raised vertically, legs folding back. Profile.
    • squat_overSquat overStanding deep squat above a partner, knees strongly bent. Profile, facing +x.
    • arch_supineArched supineOn the back with hips lifted and back arched, feet planted (display arch/bridge). Profile, head-left.
    • standing_bentStanding, bent overStanding with the torso folded forward at the hips, hands toward the floor or a support. Profile, facing +x.
    • standing_carryCarried (standing)Held off the ground by a standing partner, torso upright, legs wrapped or held forward. Profile, facing +x.
References (2)
  1. Lin T-Y et al. (2014). Microsoft COCO: Common Objects in Context. ECCV.
  2. COCO Keypoint Detection Task — 17-keypoint person layout.

Wardrobe state

3 dimensions · v1.0.0 · CC-BY-4.0

What the participants are wearing and how it sits on the body. Included because coverage state is the single most prominent control in commercial adult generators (it is typically the first menu a user touches), yet it is absent from every structured position taxonomy. Coverage and disarray are separated deliberately: 'topless' describes how much is covered, 'pulled aside' describes how a garment has been displaced, and the two combine freely. Per-participant, not per-pose. Adults only.

Dimensions

  • Coverage state

    ordinal

    How much of the participant's body is covered.

    Ordered by extent of body coverage, from fully unclothed to fully clothed.

    Values (8)

    • nudeNudeNo garments.
    • accessories_onlyAccessories onlyJewellery, footwear, or hosiery only; no torso or hip coverage.
    • toplessToplessChest uncovered, hips covered.
    • bottomlessBottomlessHips uncovered, torso covered.
    • sheerSheerCovered by transparent or semi-transparent fabric.
    • lingerieLingerie / swimwearMinimal opaque coverage of chest and hips.
    • partially_clothedPartially clothedSome outer garments removed; underwear or a partial outfit remains.
    • clothedClothedFully dressed in ordinary clothing.
  • Disarray mode

    categorical

    How the garments have been displaced from their normal worn state.

    Descriptor of how a garment departs from its worn position. Independent of coverage_state.

    Values (8)

    • intactIntactGarments worn normally.
    • pulled_asidePulled asideA garment is displaced sideways without removal.
    • pushed_upPushed upA garment is raised above its normal position.
    • pulled_downPulled downA garment is lowered below its normal position.
    • unfastenedUnfastened / openClosures released; the garment hangs open.
    • half_removedHalf removedPartly off the body, still attached at a limb.
    • disorderedDisorderedRumpled and out of place without a specific displacement.
    • tornTornFabric damaged.
  • Garment class

    categorical

    The kind of clothing worn, where any is worn. Multi-valued.

    Descriptive enumeration of the garment families that appear as selectable options in commercial adult generators.

    Values (11)

    • noneNone
    • lingerie_setLingerie set
    • hosieryStockings / hosiery
    • swimwearSwimwear
    • casualCasual wear
    • formalFormal wear
    • uniformUniform / occupational
    • fetish_gearFetish gear
    • costumeCostume / character
    • footwear_onlyFootwear only
    • jewellery_onlyJewellery only
References (3)
  1. Danbooru tag group: attire, garment folksonomy, consulted for garment-family coverage; wording original.
  2. Coverage/disarray split is original to this vocabulary; no consulted source separates the two axes.
  3. Internal: pose-vocabularies/presentation.json `intensity_tier`, the gating contract that `intensity_floor` references.

Scene & setting

4 dimensions · v1.0.0 · CC-BY-4.0

Where the depiction takes place and how it is lit and shot. Included because a location control and a production-look control appear in effectively every commercial adult generator's interface, and because production mode materially changes the prompt: the same position rendered as 'studio professional' and as 'phone selfie' needs different lighting, lens, and grain language. Adults only.

Dimensions

  • Location class

    categorical

    The kind of place depicted.

    Descriptive enumeration of the location families offered as presets across commercial adult generators.

    Values (15)

    • studio_seamlessStudio (seamless)Plain backdrop, no environmental detail. The default for reference and catalogue imagery.
    • bedroomBedroom
    • bathroomBathroom / shower
    • living_spaceLiving room / kitchen
    • hotelHotel room
    • officeOffice
    • casting_roomCasting / audition roomSparse room with a couch and a visible camera rig; a recognised adult-media set convention.
    • gymGym / locker room
    • poolPool / spa
    • vehicleVehicle interior
    • outdoor_naturalOutdoor (natural)Beach, forest, field.
    • outdoor_urbanOutdoor (urban)Street, rooftop, alley.
    • club_stageClub / stage
    • industrialIndustrial / warehouse
    • fantasy_setFantasy / constructed setNon-naturalistic environment built for the depiction.
  • Lighting setup

    categorical

    The dominant light source and its quality.

    Standard photographic lighting descriptors as applied to adult media production.

    Values (9)

    • soft_keySoft keyLarge diffuse source; gentle shadow edges.
    • hard_keyHard keySmall undiffused source; crisp shadow edges.
    • window_naturalNatural windowDaylight through a window; directional and soft.
    • low_key_rimLow key / rimDark field with an edge-defining backlight.
    • practical_ambientPractical / ambientLit by lamps visible in the scene.
    • neon_colourNeon / colouredSaturated coloured sources.
    • on_camera_flashOn-camera flashDirect frontal flash; flat subject, hard background shadow.
    • candle_firelightCandle / firelightLow warm flickering source.
    • overcast_flatOvercast / flatEven shadowless illumination.
  • Production mode

    categorical

    What the imagery appears to have been shot on and by whom. Drives lens, grain, framing discipline, and colour.

    Descriptor of the apparent capture apparatus and its photographic consequences.

    Values (8)

    • studio_professionalStudio professionalCrewed shoot; controlled light, clean framing, full-frame lens character.
    • amateur_handheldAmateur handheldConsumer camera, available light, imperfect framing.
    • phone_selfiePhone selfieFront-facing phone camera held at arm's length; wide lens distortion.
    • mirror_selfieMirror selfiePhone photographed in a mirror; the device is visible.
    • webcam_streamWebcam / streamFixed low-mounted webcam, compressed, front-lit by a screen.
    • film_vintageFilm / vintageAnalogue grain, period colour, softer resolution.
    • editorial_glamourEditorial glamourMagazine production values; retouched, styled, deliberate.
    • surveillance_lookSurveillance lookHigh-mounted fixed wide camera, low frame rate, monochrome or desaturated.
  • Time of day

    categorical

    Apparent time of day.

    Standard temporal descriptors, recorded because they drive colour temperature and light direction.

    Values (6)

    • unspecifiedUnspecified
    • morningMorning
    • middayMidday
    • golden_hourGolden hour
    • duskDusk
    • nightNight
References (3)
  1. Photographic lighting terminology (key quality, rim, practical, flat) as applied in production practice.
  2. Location and production-mode value sets derive from the preset menus commonly exposed by commercial adult generators; wording original and no vendor menu is reproduced verbatim.
  3. Internal: pose-vocabularies/presentation.json `camera_framing`, static viewpoint, distinct from these axes.

Expression & gaze

3 dimensions · v1.0.0 · CC-BY-4.0

Where a participant is looking and what their face is doing. Included because gaze direction is one of the highest-frequency descriptors in the folksonomies that open adult models were trained on, and because it is the axis with the largest effect on how a still reads while costing the least to control. Per-participant, not per-pose. Adults only.

Dimensions

  • Gaze target

    categorical

    Where the participant is looking.

    Direction-of-look descriptor; the highest-frequency single descriptor across booru-style adult folksonomies.

    Values (8)

    • at_viewerAt viewerDirect eye contact with the lens.
    • at_partnerAt partnerLooking at another participant.
    • awayAwayLooking off-frame, no specific target.
    • upwardUpwardEyes raised.
    • downwardDownwardEyes lowered.
    • over_shoulderOver shoulderHead turned back toward the lens or partner behind. Characteristic of from-behind arrangements.
    • eyes_closedEyes closedEyes shut.
    • obscuredObscuredFace not visible or covered.
  • Facial affect

    categorical

    The emotional register the face is playing.

    Expression descriptors as used in performance direction, restricted to states readable from a single frame.

    Values (9)

    • neutralNeutral
    • smilingSmiling
    • playfulPlayful / teasing
    • sultrySultryHalf-lidded, composed, deliberate.
    • focusedFocusedConcentrated, absorbed in the act.
    • pleasure_restrainedPleasure (restrained)Contained reaction; brow and mouth engaged, composure held.
    • pleasure_overwhelmedPleasure (overwhelmed)Loss of composure; brow raised, mouth open, eyes rolled or shut. Corresponds to the folksonomy term rendered as 'ahegao' in anime-domain tagging.
    • surprisedSurprised
    • distressedDistressedReads as discomfort or unwillingness.
  • Mouth state

    categorical

    The configuration of the mouth, recorded separately from affect because it is independently controllable and frequently mis-rendered.

    Discrete mouth configurations readable from a single frame.

    Values (6)

    • closedClosed
    • partedParted
    • openOpen
    • tongue_outTongue out
    • biting_lipBiting lip
    • occupiedOccupiedMouth engaged in the act.
References (2)
  1. Danbooru tag group: face tags and expression folksonomy, consulted for coverage of readable expression states; wording original.
  2. Gaze/affect/mouth three-way split is original to this vocabulary; consulted folksonomies bundle these into single tags.

The action layer, action beats, tempo, camera moves, loop types and contact continuity, has its own reference at the action and camera catalog, where each value carries the mechanical wording used for video prompting.