The purpose of this paper is to construct and validate a model centered on architectural emotion, to explore the role of emotion in architectural design, and to provide theoretical support for emotion-oriented design. This study collected 614 terms related to architectural emotion, screened 30 core terms, constructed a two-dimensional architectural emotion model, and verified the scientific and practicality of the model through three measurement methods. First, the one-dimensional scale analysis identified two dimensions of pleasure and charm, which portrayed the range of word variation; second, the Principal Component Analysis confirmed the periodic ordering pattern of words, which revealed its systematic relationship; and, lastly, the Multidimensional Scaling Analysis demonstrated the distribution of emotion words based on cognitive similarity in the multidimensional space. Based on this model, this paper proposes a three-layer circular model of “architectural emotion-architectural cognition-architectural elements”, which constructs the correspondence between architectural emotion and design elements, as well as how architectural cognition and architectural elements can synergize to create a spatial experience that triggers specific emotions. The model provides theoretical support for emotion-oriented architectural design and evaluation, and helps designers to better understand the relationship between emotion and space so as to create more valuable architectural works.

With the increasing complexity of human–computer interaction scenarios, conventional digital human facial expression systems show notable limitations in handling multi-emotion co-occurrence, dynamic expression, and semantic responsiveness. This paper proposes a digital human system framework that integrates multimodal emotion recognition and compound facial expression generation. The system establishes a complete pipeline for real-time interaction and compound emotional expression, following a sequence of “speech semantic parsing—multimodal emotion recognition—Action Unit (AU)-level 3D facial expression control.” First, a ResNet18-based model is employed for robust emotion classification using the AffectNet dataset. Then, an AU motion curve driving module is constructed on the Unreal Engine platform, where dynamic synthesis of basic emotions is achieved via a state-machine mechanism. Finally, Generative Pre-trained Transformer (GPT) is utilized for semantic analysis, generating structured emotional weight vectors that are mapped to the AU layer to enable language-driven facial responses. Experimental results demonstrate that the proposed system significantly improves facial animation quality, with naturalness increasing from 3.54 to 3.94 and semantic congruence from 3.44 to 3.80. These results validate the system’s capability to generate realistic and emotionally coherent expressions in real time. This research provides a complete technical framework and practical foundation for high-fidelity digital humans with affective interaction capabilities.

Micro-expressions, characterized by brief and subtle facial muscle movements, are essential for conveying nuanced emotions in digital humans, yet existing rendering techniques often produce rigid or emotionally monotonous animations due to the inadequate modeling of temporal dynamics and action unit interdependencies. This paper proposes a graph-driven framework for micro-expression rendering that generates emotionally diverse and lifelike expressions. We employ a 3D-ResNet-18 backbone network to perform joint spatio-temporal feature extraction from facial video sequences, enhancing sensitivity to transient motion cues. Action units (AUs) are modeled as nodes in a symmetric graph, with edge weights derived from empirical co-occurrence probabilities and processed via a graph convolutional network to capture structural dependencies and symmetric interactions. This symmetry is justified by the inherent bilateral nature of human facial anatomy, where AU relationships are based on co-occurrence and facial anatomy analysis (as per the FACS), which are typically undirected and symmetric. Human faces are symmetric, and such relationships align with the design of classic spectral GCNs for undirected graphs, assuming that adjacency matrices are symmetric to model non-directional co-occurrences effectively. Predicted AU activations and timestamps are interpolated into continuous motion curves using B-spline functions and mapped to skeletal controls within a real-time animation pipeline (Unreal Engine). Experiments on the CASME II dataset demonstrate superior performance, achieving an F1-score of 77.93% and an accuracy of 84.80% (k-fold cross-validation, k = 5), outperforming baselines in temporal segmentation. Subjective evaluations confirm that the rendered digital human exhibits improvements in perceptual clarity, naturalness, and realism. This approach bridges micro-expression recognition and high-fidelity facial animation, enabling more expressive virtual interactions through curve extraction from AU values and timestamps.

With the increasing complexity of human-computer interaction scenarios, conventional digital human facial expression systems show notable limitations in handling multi-emotion co-occurrence, dynamic expression, and semantic responsiveness. This paper proposes a digital human system framework that integrates multimodal emotion recognition and compound facial expression generation. The system establishes a complete pipeline for real-time interaction and compound emotional expression, following a sequence of "speech semantic parsing—multimodal emotion recognition—Action Unit (AU)-level 3D facial expression control."First, a ResNet18-based model is employed for robust emotion classification using the AffectNet dataset. Then, an AU motion curve driving module is constructed on the Unreal Engine platform, where dynamic synthesis of basic emotions is achieved via a state-machine mechanism. Finally, Generative Pre-trained Transformer (GPT) is utilized for semantic analysis, generating structured emotional weight vectors that are mapped to the AU layer to enable language-driven facial responses. The software interface is shown in Fig .Experimental results demonstrate that the proposed system outperforms traditional methods in terms of recognition accuracy, expression naturalness, and interaction efficiency, effectively supporting realistic, natural, and semantically aligned facial expressions. This research provides a complete technical framework and practical foundation for high-fidelity digital humans with affective interaction capabilities.

Micro-expressions, characterized by brief and subtle facial muscle movements, are essential for conveying nuanced emotions in digital humans, yet existing rendering techniques often produce rigid or emotionally monotonous animations due to inadequate modeling of temporal dynamics and action unit interdependencies. This paper proposes a graph-driven framework for micro-expression rendering that generates emotionally diverse and lifelike expressions. We employ a 3D-ResNet-18 backbone for joint spatiotemporal feature extraction from facial video sequences, enhancing sensitivity to transient motion cues. Action units are modeled as nodes in an asymmetric graph, with edge weights derived from empirical co-occurrence probabilities, and processed through a graph convolutional network to capture structural dependencies and asymmetric interactions. Predicted action unit activations are interpolated into continuous motion curves using B-spline functions, modulated by emotion-specific intensity parameters, and mapped to skeletal controls in a real-time animation pipeline (Unreal Engine). Experiments on the CASME II dataset demonstrate superior performance, achieving an F1-score of 0.623 and accuracy of 64.1%, outperforming baselines in temporal segmentation and expression diversity. Subjective evaluations confirm enhanced perceptual clarity, naturalness, and authenticity in rendered digital humans. This approach bridges micro-expression recognition and high-fidelity facial animation, enabling more expressive virtual interactions.

Modern dance conveys complex emotions through continuous bodily movement, but assigning recorded motion to continuous affective coordinates remains challenging. This study examines the mapping of Laban Movement Analysis (LMA)-informed kinematic descriptors to predefined emotion prototypes in Russell’s valence–arousal space. Thirty-one frame-level descriptors were extracted from recorded dance motion and temporally aggregated for circular-target regression. Four-quadrant classification and direct valence–arousal regression were also evaluated as separate modeling tasks. In the 12-category circular experiment, the selected ExtraTrees model was evaluated on 117 motion sequences using five-fold stratified cross-validation and obtained an angular mean absolute error of 36.765 ± 6.414° and a quadrant macro-F1 of 0.670 ± 0.094. These values measure prediction against assigned category-level affective targets. Shortest-arc interpolation of predefined prototype angles, combined with linear interpolation of their radii, produced continuous numerical paths in valence–arousal space; generated skeletal transitions were not evaluated. A separate questionnaire involving 36 observers assessed 24 videos of the recorded dancers. The findings provide preliminary evidence that the selected descriptors recover assigned affective direction in this recorded-dance dataset. Perceptual evaluation of generated skeletal or avatar motion and investigation of potential applications in immersive or digital-human settings remain future work.