Asianart.com | Associations | Articles | Exhibitions | Galleries


Visitors' Forum

Asian Art  Forums - Detail List
Asian Art Forums

Message Listing by Date:
Message Index | Back | Post a New Message | Search | Private Mail | FAQ
Subject:General question
Posted By: Richard Rodeheffer Sat, May 10, 2025 IP: 66.188.193.117

I have a number of Japanese scroll paintings, Nanga and Zenga, and it can be challenging to identify artists. I have a modest library of books, mostly in English, dealing with Japanese painting, but this is of limited value when it comes down to this level of information. Are there helpful reference books or web resources that would help identify signatures and seals?



Subject:Re: General question
Posted By: rat Sun, May 11, 2025

If the pictures are not by established artists, it can be difficult to identify their creators other than by reading the seals, which are typically carved in a variety of archaic script types, and sometimes in obscure variant characters. There are good references for Japanese seals and signatures (I. Nagy knows these better than I do; some are quite extensive but also not readily available unless someone has posted them online somewhere); however, they obviously don't include every artist. Matters are complicated by the common artistic habit of using multiple names, so even if you can read the content of a seal, you also need to determine who used the name in question (sometimes multiple people did). There are a few AI and OCR-based apps that try to read seals (here is one from Taiwan that you can download; it tries to read "seal script" characters but is not particularly effective: https://apkpure.com/tw/zhuan-shu-shi-bie/com.xiamo.zuanshu). There may be better ones available for Japanese seals (which can use character variants that don't appear often in Chinese).

Subject:Re: General question
Posted By: Richard Rodeheffer Mon, May 12, 2025

Thank you. One would think that a pattern recognition challenge like identifying seals would be ripe for an AI based approach. One could envision a database with known seals that could’ve expanded over time as new information becomes available.

Subject:Re: General question
Posted By: rat Tue, May 13, 2025

Exactly. There is a database along those lines maintained by China's Zhejiang University (not always accessible at any given moment though); it combines images for Chinese paintings/calligraphy from leading published compilations of seals and I expect that someone is or will be training AI with it. Here is another(?) database that a Chinese team has posted, presumably for that purpose. I have not downloaded it: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5079459

I know much less about Japanese resources, but imagine similar databases exist. Here is one that compiles images of collectors' seals, for instance: https://seal.dhii.jp/sealdb/

The trick I suspect will be managing GenAI's hallucination problem.


Asianart.com | Associations | Articles | Exhibitions | Galleries |