Many digital musical instrument design frameworks have been proposed that are well suited for analysis and comparison. However, not all provide applicable design suggestions, especially where subtle but important details are concerned. Using traditional lutherie as a model, we conducted a series of interviews to explore how violin makers ``go beyond the obvious'', and how players perceive and describe subtle details of instrumental quality. We find that lutherie frameworks provide clear design methods and have substantial empirical backing, but are not enough to make a fine violin. Success comes after acquiring sufficient tacit knowledge, which enables detailed craft through subjective, empirical methods. Testing instruments for subtle qualities was suggested to be a different skill to playing. Whilst players are able to identify some specific details about instrumental quality by comparison, these are often not actionable, and important aspects of ``sound and feeling'' are much more difficult to describe. In the DMI domain, we introduce NIMEcraft to describe subtle differences between otherwise identical instruments and their underlying design processes, and consider how to improve the dissemination of NIMEcraft.