PoseMate: A Conceptual Framework for AI-Generated Socially Driven Pose Guidance for Gen Z Photo-Taking
Archives of Design Research (AoDR), 39(2), 77-95 (2026)

Summary
Gen Z spends substantial time on social media and often experiences fear of missing out (FoMO), motivating them to align everyday decisions, including photo-taking, with socially visible trends. While recent design and human-computer interaction (HCI) research has examined socially driven decision support, how such support should be designed to support Gen Z’s expressive photo-taking remains underexplored. We conducted a formative study to examine Gen Z’s photo-taking practices and their reliance on socially shared visual references. Building on these insights, we introduced PoseMate, a conceptual artificial intelligence (AI) photo-taking assistant that frames posture trends from social media as designable knowledge. Then, we conducted a participatory design study to inform PoseMate’s initial analytic workflow, focusing on how Gen Z expects AI to curate the input photo set from which pose trends are derived. The studies showed that Gen Z prefers socially shared, user-generated pose references over expert-driven, rule-based guidance, and that pose trends become meaningful only when derived from carefully filtered photo sets rather than all available images. In the participatory study, participants articulated key filtering dimensions that shape how socially driven pose trends are constructed and interpreted, suggesting design strategies for AI-supported pose-trend guidance generation. This work contributes empirical and design insights into socially inspired photo-taking in Gen Z and reframes photo filtering as a trend-making infrastructure for socially driven pose-guidance systems. By foregrounding user-controllable photo set construction, PoseMate shifts AI support away from prescribing “best” poses toward enabling interpretable, socially grounded pose references derived from shared visual practices.
Citation
Nur Izzatty Binti Mohamad Jamal and Bokyung Lee. (2026). PoseMate: A Conceptual Framework for AI-Generated Socially Driven Pose Guidance for Gen Z Photo-Taking. Archives of Design Research (AoDR), 39(2), 77-95. http://dx.doi.org/10.15187/adr.2026.05.39.2.77
Project Team
Nur Izzatty Binti Mohamad Jamal, Bokyung Lee
Keywords
AI in photo-taking, AI for FoMO, socially-driven AI guide, pose guidance, pose trends, Gen Z, social media, photo filtering
Acknowledgement
This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIT) [RS-2024-00340828].

PoseMate conceptualizes AI pose guidance as a pipeline from user-controllable photo filtering to anonymized pose analysis and socially grounded pose recommendations.
PoseMate conceptualizes AI pose guidance as a pipeline from user-controllable photo filtering to anonymized pose analysis and socially grounded pose recommendations.
Socially Driven Photo-Taking
PoseMate starts from a simple observation: for Gen Z, taking photos is not only about recording a moment, but also about participating in a shared visual culture. Social media makes certain poses, places, and moods highly visible, and young people often use these images as references when deciding how they want to appear. Existing photo-taking assistants usually focus on technical quality, such as framing, composition, or expert-defined posing rules. This project instead asks how AI can support photo-taking when the goal is not simply to make a technically good image, but to help users interpret socially meaningful pose trends and adapt them to their own situation.
PoseMateλŠ” Gen Zμ—κ²Œ 사진 촬영이 λ‹¨μˆœν•œ 기둝을 λ„˜μ–΄μ„ λ‹€λŠ” μ μ—μ„œ μΆœλ°œν•©λ‹ˆλ‹€. 사진은 μ†Œμ…œ 미디어에 μ˜¬λΌκ°€κ³ , μΉœκ΅¬λ“€μ—κ²Œ 보이며, λ•Œλ‘œλŠ” νŠΉμ • λΆ„μœ„κΈ°λ‚˜ νŠΈλ Œλ“œμ— μ°Έμ—¬ν•˜λŠ” 방식이 λ©λ‹ˆλ‹€. κ·Έλž˜μ„œ ν¬μ¦ˆλ„ λ‹¨μˆœνžˆ λͺΈμ„ μ–΄λ–»κ²Œ λ‘˜μ§€μ˜ λ¬Έμ œκ°€ μ•„λ‹™λ‹ˆλ‹€. μžμ‹ κ°, μ†Œμ†κ°, 유머, μŠ€νƒ€μΌμ„ λ³΄μ—¬μ£ΌλŠ” ν•˜λ‚˜μ˜ μ‹œκ°μ  μ–Έμ–΄κ°€ 될 수 μžˆμŠ΅λ‹ˆλ‹€. ν•˜μ§€λ§Œ κΈ°μ‘΄ 사진 보쑰 μ‹œμŠ€ν…œμ€ 주둜 κ΅¬λ„λ‚˜ ν”„λ ˆμ΄λ°, μ „λ¬Έκ°€κ°€ μ •ν•œ 쒋은 포즈처럼 기술적인 기쀀을 μ€‘μ‹¬μœΌλ‘œ μž‘λ™ν•©λ‹ˆλ‹€. PoseMateλŠ” μ—¬κΈ°μ„œ μ§ˆλ¬Έμ„ λ°”κΏ‰λ‹ˆλ‹€. 쒋은 사진을 λ§Œλ“€μ–΄μ£ΌλŠ” AIκ°€ μ•„λ‹ˆλΌ, μ‚¬μš©μžκ°€ μ‚¬νšŒμ μœΌλ‘œ 의미 μžˆλŠ” 포즈 νŠΈλ Œλ“œλ₯Ό μ΄ν•΄ν•˜κ³  μžμ‹ μ˜ 촬영 상황에 맞게 ν™œμš©ν•˜λ„λ‘ λ•λŠ” AIλŠ” μ–΄λ–€ λͺ¨μŠ΅μ΄μ–΄μ•Ό ν• κΉŒμš”?
The Difficulty of Reading Pose Trends
In the formative study, participants described using social media photos as practical pose references, especially in travel contexts. They looked through Instagram, Pinterest, and other platforms to understand what kinds of photos people commonly take at a place. However, this was not a straightforward copying process. Participants had to browse many images, infer repeated pose patterns, judge whether a pose suited their body, clothing, preferences, or group, and then adapt it to real-world constraints such as lighting, crowds, and spatial layout. The study shows that the challenge is not a lack of reference images, but the difficulty of making sense of them.
ν˜•μ„± μ—°κ΅¬μ—μ„œ μ°Έμ—¬μžλ“€μ€ μ—¬ν–‰μ§€μ—μ„œ 사진을 찍기 μ „ μ†Œμ…œ λ―Έλ””μ–΄λ₯Ό 자주 μ‚΄νŽ΄λ³Έλ‹€κ³  λ§ν–ˆμŠ΅λ‹ˆλ‹€. μΈμŠ€νƒ€κ·Έλž¨μ΄λ‚˜ ν•€ν„°λ ˆμŠ€νŠΈμ—μ„œ μ‚¬λžŒλ“€μ΄ μ–΄λ–€ μž₯μ†Œμ—μ„œ μ–΄λ–€ 포즈λ₯Ό μ·¨ν•˜λŠ”μ§€ 보고, 그쀑 μžμ‹ μ˜ 사진에 μ°Έκ³ ν•  λ§Œν•œ 이미지λ₯Ό μ°ΎλŠ” κ²ƒμž…λ‹ˆλ‹€. ν•˜μ§€λ§Œ 이 과정은 λ§ˆμŒμ— λ“œλŠ” 사진 ν•˜λ‚˜λ₯Ό 골라 κ·ΈλŒ€λ‘œ 따라 ν•˜λŠ” κ²ƒμ²˜λŸΌ λ‹¨μˆœν•˜μ§€ μ•Šμ•˜μŠ΅λ‹ˆλ‹€. μˆ˜λ§Žμ€ 이미지λ₯Ό λ„˜κ²¨ 보며 λ°˜λ³΅λ˜λŠ” 포즈λ₯Ό 슀슀둜 νŒŒμ•…ν•΄μ•Ό ν–ˆκ³ , κ·Έ ν¬μ¦ˆκ°€ μžμ‹ μ˜ μ²΄ν˜•μ΄λ‚˜ 옷차림, ν•¨κ»˜ μ°λŠ” μ‚¬λžŒ, 촬영 μž₯μ†Œμ— μ–΄μšΈλ¦¬λŠ”μ§€λ„ νŒλ‹¨ν•΄μ•Ό ν–ˆμŠ΅λ‹ˆλ‹€. μ‹€μ œ ν˜„μž₯μ—μ„œλŠ” μ‘°λͺ…, μ‚¬λžŒ 수, 곡간 λ°°μΉ˜λ„ 달라지기 λ•Œλ¬Έμ— 레퍼런슀λ₯Ό κ·ΈλŒ€λ‘œ μž¬ν˜„ν•˜κΈ° μ–΄λ ΅μŠ΅λ‹ˆλ‹€. κ²°κ΅­ ν•„μš”ν•œ 것은 더 λ§Žμ€ 이미지가 μ•„λ‹ˆλΌ, 이미지λ₯Ό 읽고 μ •λ¦¬ν•˜κ³  λ‚΄ 상황에 맞게 해석할 수 μžˆλŠ” μ§€μ›μ΄μ—ˆμŠ΅λ‹ˆλ‹€.
PoseMate Concept
PoseMate is proposed as a conceptual AI workflow that treats pose trends from social media as designable knowledge. The workflow begins by constructing a photo set through user-controllable filtering, rather than assuming that all available social media photos are equally meaningful. It then abstracts posture information into skeleton-based representations to reduce reliance on identifiable visual details, and identifies recurring pose patterns through pose similarity or clustering. The central idea is that pose trends are not simply extracted from a large dataset. They become meaningful through the way users and AI construct the input photo set before analysis begins.
PoseMateλŠ” μ†Œμ…œ λ―Έλ””μ–΄ 속 포즈 νŠΈλ Œλ“œλ₯Ό AIκ°€ λ‹€λ£° 수 μžˆλŠ” μ§€μ‹μœΌλ‘œ λ°”κΎΈλŠ” κ°œλ…μ  μ›Œν¬ν”Œλ‘œμš°μž…λ‹ˆλ‹€. μ—¬κΈ°μ„œ μ€‘μš”ν•œ 점은 λͺ¨λ“  사진을 λ˜‘κ°™μ΄ λΆ„μ„ν•˜μ§€ μ•ŠλŠ”λ‹€λŠ” κ²ƒμž…λ‹ˆλ‹€. λ¨Όμ € μ‚¬μš©μžκ°€ 촬영 λͺ©μ μ΄λ‚˜ λ§₯락에 맞게 μ°Έκ³ ν•  사진 집합을 μ’νž™λ‹ˆλ‹€. κ·Έλ‹€μŒ μ‹œμŠ€ν…œμ€ 사진 속 인물을 κ·ΈλŒ€λ‘œ λΆ„μ„ν•˜κΈ°λ³΄λ‹€ μŠ€μΌˆλ ˆν†€ ν˜•νƒœμ˜ 포즈 μ •λ³΄λ‘œ 좔상화해 κ°œμΈμ •λ³΄ λ…ΈμΆœμ„ 쀄이고, μœ μ‚¬ν•œ ν¬μ¦ˆκ°€ λ°˜λ³΅λ˜λŠ” νŒ¨ν„΄μ„ μ°Ύμ•„λƒ…λ‹ˆλ‹€. PoseMateκ°€ λ§ν•˜λŠ” νŠΈλ Œλ“œλŠ” κ±°λŒ€ν•œ 데이터셋 μ–΄λ”˜κ°€μ— 이미 μ •ν•΄μ Έ μžˆλŠ” 닡이 μ•„λ‹™λ‹ˆλ‹€. μ–΄λ–€ 사진을 μ°Έκ³  λŒ€μƒμœΌλ‘œ 삼을지 μ‚¬μš©μžκ°€ μ–΄λ–»κ²Œ κ³ λ₯΄κ³  κ±ΈλŸ¬λ‚΄λŠ”μ§€μ— 따라 νŠΈλ Œλ“œμ˜ μ˜λ―Έκ°€ λ‹¬λΌμ§‘λ‹ˆλ‹€.
Filtering as Trend-Making
The participatory design study examined how Gen Z expects AI to curate the photo set behind pose guidance. Participants reviewed Instagram photos from two travel destinations and discussed which images felt relevant, inspiring, or worth following. Their responses showed that filtering is not just a technical preprocessing step. It is the place where pose trends become visible, trustworthy, and usable. Participants wanted filtering to consider social metadata such as likes and hashtags, environmental context such as season or weather, human representations such as clothing, body type, and group size, and micro-locations within a destination. These criteria shape not only which photos are selected, but also what counts as a meaningful trend.
μ°Έμ—¬ λ””μžμΈ μ—°κ΅¬μ—μ„œλŠ” AIκ°€ μ–΄λ–€ 사진을 μ°Έκ³ ν•΄μ•Ό 포즈 κ°€μ΄λ“œκ°€ 더 μœ μš©ν•΄μ§ˆμ§€ ν•¨κ»˜ μ‚΄νŽ΄λ³΄μ•˜μŠ΅λ‹ˆλ‹€. μ°Έμ—¬μžλ“€μ€ 두 μ—¬ν–‰μ§€μ˜ μΈμŠ€νƒ€κ·Έλž¨ 사진을 보며 μ–΄λ–€ μ΄λ―Έμ§€λŠ” μ°Έκ³ ν•  λ§Œν•˜κ³ , μ–΄λ–€ μ΄λ―Έμ§€λŠ” μ œμ™Έν•˜κ³  싢은지 μ΄μ•ΌκΈ°ν–ˆμŠ΅λ‹ˆλ‹€. 이 κ³Όμ •μ—μ„œ 필터링은 λ‹¨μˆœνžˆ 사진을 κ²€μƒ‰ν•˜κ±°λ‚˜ μ •λ¦¬ν•˜λŠ” 단계가 μ•„λ‹ˆλΌλŠ” 점이 λ“œλŸ¬λ‚¬μŠ΅λ‹ˆλ‹€. μ–΄λ–€ 사진을 남기고 μ–΄λ–€ 사진을 λΉΌλŠλƒμ— 따라 λ³΄μ΄λŠ” 포즈 νŠΈλ Œλ“œ μžμ²΄κ°€ 달라지기 λ•Œλ¬Έμž…λ‹ˆλ‹€. μ°Έμ—¬μžλ“€μ€ μ’‹μ•„μš” μˆ˜μ™€ ν•΄μ‹œνƒœκ·Έ, κ³„μ ˆκ³Ό 날씨, 옷차림과 μ²΄ν˜•, ν•¨κ»˜ μ°λŠ” μ‚¬λžŒ 수, 촬영 μ§€μ μ²˜λŸΌ μ—¬λŸ¬ 기쀀이 반영되길 μ›ν–ˆμŠ΅λ‹ˆλ‹€. 이런 기쀀듀은 ν¬μ¦ˆκ°€ μœ ν–‰μ²˜λŸΌ λ³΄μ΄λŠ”μ§€, λ‚΄ 상황에 λ§žλŠ”μ§€, μ‹€μ œλ‘œ 따라 ν•  수 μžˆλŠ”μ§€λ₯Ό ν•¨κ»˜ κ²°μ •ν•©λ‹ˆλ‹€.
Design Implications
PoseMate reframes AI photo-taking assistance as a form of social sensemaking. Rather than telling users the single best pose, AI can help them understand which pose references matter, why they feel relevant, and how they can be adapted. The paper’s key design implication is to treat filtering as trend-making infrastructure. Filtering constructs social legitimacy, supports expressive intent, situates trends in context, and negotiates whether a pose feels attainable for a particular body, outfit, group, or place. Because the system relies on socially shared visual data, the paper also emphasizes ethical design directions, including user-controllable data construction, skeletonized pose representations, and privacy-aware processing.
PoseMate의 핡심은 AI 사진 보쑰λ₯Ό μ •λ‹΅ μΆ”μ²œμ΄ μ•„λ‹ˆλΌ 의미 ν•΄μ„μ˜ 문제둜 λ³Έλ‹€λŠ” 데 μžˆμŠ΅λ‹ˆλ‹€. AIκ°€ “이 ν¬μ¦ˆκ°€ κ°€μž₯ μ’‹λ‹€"κ³  λ§ν•˜λŠ” λŒ€μ‹ , μ‚¬μš©μžκ°€ μ–΄λ–€ 레퍼런슀λ₯Ό μ™œ μ°Έκ³ ν•˜κ³  싢은지 μ΄ν•΄ν•˜κ³ , 그것을 μžμ‹ μ˜ λͺΈκ³Ό 옷차림, ν•¨κ»˜ μ°λŠ” μ‚¬λžŒ, μž₯μ†Œμ— 맞게 λ°”κΏ”λ³Ό 수 μžˆλ„λ‘ 도와야 ν•œλ‹€λŠ” κ²ƒμž…λ‹ˆλ‹€. λ”°λΌμ„œ 이 λ…Όλ¬Έμ—μ„œ 필터링은 λ‹¨μˆœν•œ μ „μ²˜λ¦¬κ°€ μ•„λ‹ˆλΌ νŠΈλ Œλ“œλ₯Ό λ§Œλ“œλŠ” 기반이 λ©λ‹ˆλ‹€. μ–΄λ–€ 사진을 ν¬ν•¨ν•˜λŠλƒμ— 따라 μ‚¬νšŒμ μœΌλ‘œ κ·ΈλŸ΄λ“―ν•œ 포즈, ν‘œν˜„ν•˜κ³  싢은 λΆ„μœ„κΈ°, μ‹€μ œλ‘œ κ°€λŠ₯ν•œ 포즈의 λ²”μœ„κ°€ λ‹¬λΌμ§‘λ‹ˆλ‹€. λ™μ‹œμ— μ†Œμ…œ λ―Έλ””μ–΄ 사진을 λ‹€λ£¨λŠ” 만큼, μ‚¬μš©μžκ°€ 데이터 ꡬ성을 μ‘°μ ˆν•  수 μžˆμ–΄μ•Ό ν•˜κ³ , 식별 κ°€λŠ₯ν•œ 이미지보닀 μŠ€μΌˆλ ˆν†€ 기반 포즈 ν‘œν˜„κ³Ό ν”„λΌμ΄λ²„μ‹œλ₯Ό κ³ λ €ν•œ 처리 방식이 μ€‘μš”ν•©λ‹ˆλ‹€.