arXiv:2512.11199v1 Announce Type: cross 
Abstract: Crafting computer-aided design (CAD) models has long been a painstaking and time-intensive task, demanding both precision and expertise from designers. With the emergence of 3D generation, this task has undergone a transformative impact, shifting not only from visual fidelity to functional utility but also enabling editable CAD designs. Prior works have achieved early success in single-part CAD generation, which is not well-suited for real-world applications, as multiple parts need to be assembled under semantic and geometric constraints. In this paper, we propose CADKnitter, a compositional CAD generation framework with a geometry-guided diffusion sampling strategy. CADKnitter is able to generate a complementary CAD part that follows both the geometric constraints of the given CAD model and the semantic constraints of the desired design text prompt. We also curate a dataset, so-called KnitCAD, containing over 310,000 samples of CAD models, along with textual prompts and assembly metadata that provide semantic and geometric constraints. Intensive experiments demonstrate that our proposed method outperforms other state-of-the-art baselines by a clear margin.

تم تقديم CADKnitter كإطار جديد لتوليد CAD التركيبي يستخدم استراتيجية أخذ العينات بالتشتت الموجه بواسطة الهندسة لإنشاء نماذج CAD من النصوص والإرشادات الهندسية. تتناول هذه الطريقة المبتكرة قيود طرق توليد CAD السابقة التي تركز على قطعة واحدة من خلال تمكين تجميع عدة قطع تتماشى مع المتطلبات الدلالية والهندسية.

CADKnitter se ha presentado como un nuevo marco de generación de CAD composicional que utiliza una estrategia de muestreo por difusión guiada por la geometría para crear modelos CAD a partir de indicaciones textuales y restricciones geométricas. Este enfoque innovador aborda las limitaciones de los métodos anteriores de generación de CAD de una sola pieza al permitir el ensamblaje de múltiples piezas que cumplen con los requisitos semánticos y geométricos.

CADKnitter a été présenté comme un nouveau cadre de génération de CAD compositionnel qui utilise une stratégie d'échantillonnage par diffusion guidée par la géométrie pour créer des modèles CAD à partir de prompts textuels et de contraintes géométriques. Cette approche innovante répond aux limitations des méthodes précédentes de génération de CAD à une seule pièce en permettant l'assemblage de plusieurs pièces respectant à la fois les exigences sémantiques et géométriques.

CADKnitter has been introduced as a new compositional CAD generation framework that utilizes a geometry-guided diffusion sampling strategy to create CAD models from text prompts and geometric constraints. This innovative approach addresses the limitations of previous single-part CAD generation methods by enabling the assembly of multiple parts that adhere to both semantic and geometric requirements.

CADKnitter: Compositional CAD Generation from Text and Geometry Guidance

Was this article worth reading? Share it

artypa

PlantFCE Model Builder

File Architect

Z3D

Artefacts.ai

Archipeg

Ready to build your own newsroom?