When played with the new features of Houdini 17, this idea crossed my mind...
The pipeline in a nutshell:
British Museum 3D scan model -> study scene in Houdini -> drop a fabric with Vellum -> render with low sampling and other tricks and use Optix Denoiser.
I found a great opportunity using the AI (deep learning instead...) based nVIDIA Optix Denoiser with low sampling settings: it makes a render similar to a painting.
Originally I made the first render for my Houdini 17 review for 3D World mag.
The Sketchfab model preview is below the images.
British Museum 3D scan collection: https://sketchfab.com/britishmuseum