Generative AI (GenAI) is accelerating design space exploration and multimodal prototyping in industrial design (ID), bringing new efficiencies and possibilities to early-stage ideation and cross-media expression. Yet many studies do not clearly define stage-wise human–GenAI roles, preserve constraints as traceable cross-stage artifacts, or provide verifiable stage-wise evaluation, undermining traceability in both concept convergence and concept-to-engineering handover. To address these issues, this paper proposes GID-HGCC, a GenAI-driven human–GenAI co-creation ID framework that links four core stages: requirements confirmation, concept generation, concept evaluation, and 3D modeling. First, it specifies stage-wise responsibilities and defines the corresponding inputs and outputs. Second, it establishes a traceable cross-stage artifact flow—“structured prompts–candidate concepts–evaluation outputs–3D engineering issue list”—to support continuous constraint transmission and explicit documentation. Third, it integrates a multi-dimensional evaluation criteria system with IVIFNs–CRITIC–TOPSIS for concept ranking, and further strengthens convergence reliability via preference–consistency diagnostics. The framework is validated through a case study on a portable passive cervical spine rehabilitation training device. Expert preferences over stage-wise co-creation artifacts exhibit an overall medium-to-high level of consistency, and the Top-5 overlap between each expert and the group ranking ranges from 0.80 to 1.00. These results demonstrate that GID-HGCC offers an operational reference for constraint-guided human–GenAI co-creation in ID, improving traceability and handover reliability from requirements confirmation to engineering refinement.

Camouflage design necessitates region-consistent colour and texture synthesis. However, practical datasets are frequently constrained by scarce paired supervision and cross-source distribution shift, which impede reliable generalisation across heterogeneous environments. To address this problem, we propose a region-feature-driven framework for camouflage generation from unpaired background sequences. The main framework-level contribution lies in constructing a decoupled sequence-conditioned regional representation and using it to drive a fused-background-first camouflage-generation pipeline under unpaired supervision. Specifically, each background sequence is encoded into a continuous region descriptor and combined with a discrete Environment × Season code; under unpaired supervision, this decoupled condition guides StyleGAN2-based fused-background synthesis via style modulation, after which a deterministic digital-camouflage renderer converts the synthesised fused background into a deployable pattern. On a stratified test split, the synthesised backgrounds achieve FID = 15.2 ± 0.5, KID = 4.7 ± 0.3 × 10⁻³, while also showing stable condition adherence under the tested settings. Auxiliary sequence-consistency proxy measures are PSNR = 34.56 dB, SSIM = 0.776, and LPIPS = 0.134. Under a paired forest-scenario detector protocol, the generated camouflage reduces mean image-best IoU from 0.716 to 0.400 for a one-stage detector and from 0.690 to 0.460 for a two-stage detector. Overall, these results support the effectiveness of the proposed condition design and transfer mechanism under the present experimental setting.

On-machine measurement (OMM) enables real-time dimensional feedback in production, yet accuracy is often degraded by thermal drift, sensor noise, and environmental disturbances. This motivates intelligent error-prediction methods to ensure reliable, high-precision machining. This study proposes a hybrid deep learning model integrating a Convolutional Neural Network (CNN), a Bidirectional Long Short-Term Memory (Bi-LSTM) network, and a Whale Optimization Algorithm (WOA) for precise OMM error prediction. Initially, raw measurement data underwent preprocessing to remove noise and outliers. Subsequently, we use a CNN to extract features and a Bi-LSTM to model time-dependent patterns. Finally, WOA optimizes model hyperparameters globally, further boosting predictive accuracy. Comparative experiments show that the proposed model reduces RMSE, MAE, and MAPE by approximately 53.58%, 54.96%, and 57.65%, respectively, while improving the R² score by about 11.17% over baseline methods. Results confirm the method’s superior nonlinear prediction capabilities, significantly enhancing machining accuracy and production efficiency, and demonstrating promising industrial application potential.