Wikidata talk:WikiProject sum of all paintings/Archive/2021

From Wikidata
Jump to navigation Jump to search
This page is an archive. Please do not modify it. Use the current page, even to continue an old discussion.


500.000 paintings

A lot of paintings to look at!

We hit just hit the milestone of 500.000 paintings on Wikidata. See Wikidata:WikiProject sum of all paintings/Property statistics for details (updates every Monday morning). Multichill (talk) 22:49, 10 January 2021 (UTC)

HELP! Merging gone wrong.

@Multichill: @Jarekt: @Jane023:
Sorry everybody, I thought it was a good idea (I still think it is) to merge Q3044768 (Département des peintures du Louvre, whom nobody gives a f*** about outside of Paris) with Q19675 (Louvre Museum). For some reason, it did not worked as planned at all, and I cannot explain why! I tried to undo the merger but with no results. Again, sorry!! --Edelseider (talk) 08:46, 11 January 2021 (UTC)

Copy from User_talk:Jane023
First a technical explanation: it is impossible to merge two items on Wikidata when there is two different page on a Wikimedia project. So here, if you want to merge, Department of Paintings of the Louvre (Q3044768) and Louvre Museum (Q19675), first you need to merge fr:Département des peintures du musée du Louvre and fr:Musée du Louvre.
Next the editorial point of view: merging these items (and articles) seems to be a very bad idea. It would be like merging "France" with "Europe". Plus, pages like Wikidata:WikiProject sum of all paintings/Top collections by number of Wikipedia articles could easily be fixed to take into account all departments of a museum.
Cheers, VIGNERON (talk) 13:23, 13 January 2021 (UTC)
Do not merge Department of Paintings of the Louvre (Q3044768) with Louvre Museum (Q19675). One is a subsidiary department of the other. Note they have distinct external identifiers (A vs B). Whether you or I give a "f***" about the department is irrelevant, clearly they are distinct entities and multiple libraries and catalogs recognize the distinction. -Animalparty (talk) 17:58, 13 January 2021 (UTC)

Participant list

@Multichill: linking it means that people can't see how large it. I also see no reason why a larger list of participants shouldn't be included. Why do you think it shouldn't be there? ChristianKl13:34, 17 January 2021 (UTC)

I'm trimming down the page because it has too much noise and clutter on it. I added the number of participants so people can see that. Multichill (talk) 16:12, 17 January 2021 (UTC)

"error decoding response body: expected value at line 1 column 1"

The Wikidata:WikiProject sum of all paintings/Top collections by number of Wikipedia painting articles cannot be updated because of ERROR: "error decoding response body: expected value at line 1 column 1". Could somebody please fix that? Thank you very much! --Edelseider (talk) 11:11, 28 November 2020 (UTC)

Looks like we hit the timeout limit on this one. I took out the "list per museum" part and it updated just now. Onwards and upwards to the next time this list times out! Jane023 (talk) 10:01, 30 November 2020 (UTC)
@Jane023: thank you! But why does the same thing not work here? I tried everything! --Edelseider (talk) 12:46, 30 November 2020 (UTC)
Oh you just forgot to take out the code for ?name too - since it's not used, the list won't update. Fixed now. Jane023 (talk) 13:50, 30 November 2020 (UTC)
¡Muchas gracias! --Edelseider (talk) 13:54, 30 November 2020 (UTC)
Same problem with Wikidata:WikiProject sum of all paintings/Top creators by number of identifiers. CONCAT use is the problem?--Jklamo (talk) 09:13, 5 February 2021 (UTC)
When debugging, can you please copy the list to your sandbox and experiment there? Multichill (talk) 19:11, 5 February 2021 (UTC)

Drawings by Leonardo da Vinci

Hi all. Over at enwiki Aza24 and I would like to create a list of drawings by Leonardo da Vinci, based on Frank Zöllner's 2003 catalogue raisonné which has been replicated here. I'd like to generate a Listeria list from Wikidata, and ideally re-arrange the works into chronological order, as Zöllner's numbers are based on thematic groupings. I'd then reformat the resulting table to make it acceptable for enwiki. Does anyone have a sense of what would need to be done in order to generate Wikidata items for every work in the Commons gallery I've linked to above (paintings as well as drawings)? There's also a version in the form of a table here. Adding titles, accession numbers, etc., would not be a problem as both Aza and I have print copies of the catalogue. I realise that this is strictly speaking outside this project's scope, but the technical issues must be the same for any catalogue of artworks. Ham II (talk) 15:06, 6 February 2021 (UTC)

@Ham II: sounds like a nice project! Did you already create an item for the catalogue raisonnée? If not, please create one and link it to the gallery on Commons. That would be a good starting point.
Next step would be to add the catalog code (P528) qualified with the item to the paintings.
When the paintings are done, you can add the catalog code to the already existing drawings.
With all the existing works linked up, you can go to import the missing drawings. But let's get to this point first. Shouldn't be too much work. Multichill (talk) 17:07, 6 February 2021 (UTC)
@Multichill: Thank you! I've now finished all these steps; what's next? Ham II (talk) 06:44, 10 February 2021 (UTC)
@Ham II: great! Now that you matched up the existing drawings you can add the new ones using Quickstatements. This is some example input that you can paste in to get an idea:
qid,P18,P528,qal972,P31,P170,Den,Dnl
,"""Leonardo da Vinci - Studien zu einer Anbetung der Hirten, um 1480, 21488.jpg""","1",Q105337930,Q93184,Q762,"drawing by Leonardo da Vinci","tekening van Leonardo da Vinci"
,"""Leonardo da Vinci - Etude pour une Adoration des bergers, NI1776;AI658.jpg""","2",Q105337930,Q93184,Q762,"drawing by Leonardo da Vinci","tekening van Leonardo da Vinci"
,"""Leonardo da Vinci - Sleeve study for the Annunciation, 1470-73.jpg""","3",Q105337930,Q93184,Q762,"drawing by Leonardo da Vinci","tekening van Leonardo da Vinci"
,"""Leonardo da Vinci - 1860,0616.100, Verso Three studies of the Virgin and Child, seated.jpg""","4",Q105337930,Q93184,Q762,"drawing by Leonardo da Vinci","tekening van Leonardo da Vinci"
,"""Leonaredo, studio per l'adorazione dei magi, uffizi.jpg""","5",Q105337930,Q93184,Q762,"drawing by Leonardo da Vinci","tekening van Leonardo da Vinci"
,"""Leonardo da Vinci - Adoration des Mages, RF 1978, Recto.jpg""","6",Q105337930,Q93184,Q762,"drawing by Leonardo da Vinci","tekening van Leonardo da Vinci"
,"""Leonardo da Vinci - Etude pour la Nativité de Jésus, 19-506106.jpg""","7",Q105337930,Q93184,Q762,"drawing by Leonardo da Vinci","tekening van Leonardo da Vinci"
,"""Leonardo da Vinci - Etude pour la Nativité de Jésus, 19-506105.jpg""","8",Q105337930,Q93184,Q762,"drawing by Leonardo da Vinci","tekening van Leonardo da Vinci"
Q52707816,"""Leonardo study Madonna of the rocks.jpeg""","15",Q105337930,Q93184,Q762,"drawing by Leonardo da Vinci","tekening van Leonardo da Vinci"
This is incomplete, please expand with fields like descriptions in other languages and the label in one or more languages. You can use the gallery as starting point and a spreadsheet to format it like this. Multichill (talk) 17:25, 10 February 2021 (UTC)
@Multichill: I've started a spreadsheet here, for pasting into QuickStatements. Am I doing this right so far? I've included QIDs for existing items, but I can remove those if they shouldn't be there. I'm planning to improve the formatting of the dates (i.e. adding sourcing circumstances (P1480) = circa (Q5727902)) at the end; first I want to copy them as they appear in the book. I've tried using the Wikipedia and Wikidata Tools add-on to Google Sheets in order to find the item numbers for collections, but that doesn't seem to have worked. My idea there is to delete the column with the collections' names at the end after the Wikidata items for all of them have been found. Ham II (talk) 15:08, 20 February 2021 (UTC)
Hi Multichill, I've continued trying to learn the QuickStatements syntax by experimenting with another catalogue raisonné, Michelangelo: The Complete Paintings, Sculptures and Architecture (Q105636252) (of which I've also got a print copy); this is in order to build my confidence before tackling the longer one for Leonardo drawings. I think I understand better now; copying and pasting from a spreadsheet to QuickStatements will only work if I click 'Import CSV commands', whereas the comma-separated code you wrote above works with 'Import CSV commands'. I was basing my spreadsheet on the code above, so it looks as if I either have to reformat it to work better with V1 commands, or continue with it as it is and find-and-replace all the tabs with commas after pasting it into QuickStatements. Not sure which of these I'll do yet; as I said I'll continue experimenting with Michelangelos in order to build confidence. Thanks for your help so far! Ham II (talk) 09:41, 28 February 2021 (UTC)
@Multichill: Sorry for the repeated pinging. I realise now that the spreadsheet's viewing option was set to 'Restricted'; I've now changed it to 'Anyone with the link'. Would you mind taking a look at it (here) and offering your comments? I'm leaning towards going for CSV commands – i.e. pasting this spreadsheet to QuickStatements and replacing all the tabs with commas. Thanks, Ham II (talk) 11:00, 6 March 2021 (UTC)
I will take a look. Jane023 (talk) 19:51, 6 March 2021 (UTC)
Here's a list, but I don't understand the numbering system Wikidata:WikiProject sum of all paintings/Catalog/Leonardo da Vinci, The Complete Paintings and Drawings. Jane023 (talk) 19:58, 6 March 2021 (UTC)
@Jane023: It might be clearer from the Commons gallery. Paintings and cartoons have Roman numerals, and are in chronological order except for the Salvator Mundi, a recent rediscovery, at the end. Drawings have Arabic numerals and are grouped thematically. Ham II (talk) 22:41, 6 March 2021 (UTC)
OK, thanks! Now if I only knew the regex to list Roman numerals in order! Jane023 (talk) 08:07, 7 March 2021 (UTC)

This Study for the Madonna of the Cat (Q11765908) needs to be split. It's not useful to keep both sides together when you want to discuss these as separate lines in a catalog. So total you need 3 items: one for the recto, one for the verso, and one for the series that connects them as object. Jane023 (talk) 21:46, 6 March 2021 (UTC)

I am not sure how to explain my own process for item creation with quick statements because it is an integral part of the way I set up my catalog spreadsheets that I use to check. Looking at your spreadsheet I see that you are including measurements, but these cannot be added with QS yet. In any case, the main problem is the transformation of a line item (catalog code) with its various columns in your sheet, to a multi-line input for item creation. See this image for a screenshot of a quick copy-paste that came out of my little "create item from commons file" spreadsheet, which I use quite often for single-file uploads. I pasted this into QS and then hand-editted the item further. Jane023 (talk) 09:42, 7 March 2021 (UTC)
@Jane023: Thank you for this; most helpful. I was wondering how measurements were going to work – where the units of measurement would come in. I've deleted the measurements from the spreadsheet and will input them from en:User:Ham II/Leonardo when all the works have been added to Wikidata.
If I understand correctly, the "multi-line" method of inputting data from a spreadsheet (in your screenshot) is the V1 option on QuickStatements, and the "single-line" method in Multichill's example above is the CSV option. I'd like to keep the spreadsheet's formatting simple – one line per item – so I think I'll keep it in its current CSV-esque formatting, and replace all the tabs with commas after pasting the finished spreadsheet into QuickStatements. Ham II (talk) 11:32, 7 March 2021 (UTC)

Proposal for incomplete creator and collection lists

Readers don't know if a collection is complete, so I suggest to put a note (maybe with a template) in the manner "(this list is complete)" or "(this list is incomplete)". --Mateus2019 (talk) 14:03, 9 February 2021 (UTC)

As ping project doesn't work for this project (too many participants), here is a note:

Please see Wikidata:Project_chat#Interpreting_"described_at_URL"_(P973). --- Jura 10:02, 19 February 2021 (UTC)

Canvasing about canvas

Unfortunately I can't ping the members of this project. Please have a look at Wikidata talk:WikiProject Visual arts#Canvasing_about_canvas. Multichill (talk) 17:47, 26 April 2021 (UTC)

Adding fabrication method (P2079)?

I'm wondering if it's useful to add fabrication method (P2079) to paintings. It's not used a lot now (about 1700 times). Based on the top materials it would be dominated by oil painting (Q174705) followed by things like watercolor (Q50030) and fresco painting (Q134194). Two questions:

  1. Is it useful to (mass) add fabrication method (P2079)?
  2. Should we add fabrication method (P2079) to existing items based on made from material (P186)? So for example add fabrication method (P2079) -> oil painting (Q174705) based on made from material (P186) -> oil paint (Q296955)

Multichill (talk) 11:17, 5 June 2021 (UTC)

Yes I think so. There are so many types of oil paint, and also ways to put it on a panel or canvas, that starting this way enables efforts to refine work on those. Jane023 (talk) 13:58, 5 June 2021 (UTC)

Notable work notable work (P800) for collections

I was thinking about how I have been adding estate catalogs to collectors' items with P800 but then I realized it would also be useful for collections. Either for notable exhibitions or notable sub-collections. Is this a valid use of P800 though? I thought it was mostly for people. Jane023 (talk) 14:02, 5 June 2021 (UTC)

Like this edit you mean? Yes, that could be useful. It is a curated property so it comes with the problem of users unable to make a choice and adding a ton of statements. Would probably be good to write down some guidelines if that hasn't been done already. Multichill (talk) 14:59, 5 June 2021 (UTC)
Hmm that is an interesting edit. I guess that is significant also because the museuem was designed to be centered around it. I don't think we would want all colleection highlights under P800. Additionally we don't want all exhibitions there either. I was thinking along the lines of Rembrandt. Schilderijen bijeengebracht ter gelegenheid van de inhuldiging van Hare Majesteit Koningin Wilhelmina, 8 September-31 October 1898 (Q63930925) and Museum van der Hoop (Q29053170). Jane023 (talk) 15:49, 5 June 2021 (UTC)

How to model provenance

As most of you know, I started WikiProject Provenance (Q98801351) and it has been quite an adventure this summer exploring the various aspects of painting provenance. My original intention was to build a long list of large owners and work from there. We already had quite a bit of information in owned by (P127) statements but that property is used for country Qids, city Qids and museums as well as people. I think some general guidelines for data modelling are in order and this won't be easy because everyone seems to it differently. My question today is whether we can set some basic rules of the road forward and also to gather suggestions for missing ways to model certain painting events (inheritance/auction/loss/split/rejoined/overpainted).

General property usage

In general, properties that come into play when modelling provenance are these: creator (P170), owned by (P127), and collection (P195). Bonus points go to museum items that include location (P276), and donated by (P1028).

Good faith versus bad faith provenance presentation

In the case of "good faith" provenance, I will assume all of used painting properties to be undisputed in the present usage in the present moment. However, if you consider that much of older art has been repeatedly looted & rediscovered in the past, then I will also assume that all old art with zero provenance between creation and present day has a "bad faith" provenance somewhere in its past. Examples of bad faith provenance are hiding previous catalog references by: 1) changing previous attributions to artist's pupils/colleagues, 2) changing/mispelling former owner names, 3) omitting previous collection names, 4) for depicted people, changing/mispelling sitter names and 5) for depicted places/topics, changing/mispelling these.

Marking possible "bad faith" provenance

As I catalog works with catalog code (P528) I have started to include previous attributions with the deprecated statement mark. Sometimes (but not always), I reference the deprecated attribution with a catalog link. I also occasionally add old titles in the alias field when the title includes information that is deprecated (sitter/place/topic).

Inheritance

After wracking my brains over text like "thence by descent" and trying (and often failing) to reinstate an owner's inheritance lines, I have come to the conclusion that wherever possible, this is the best way forward: use P127 with start date as death date of person bequeathing and end date as death of next person bequeathing. When successful I find that there is a lot of added benefit (family connections) to doing this properly. So I would like to propose using P127 statements for old art primarily for human items (occasionally a duo for married couples, or groups of children as heirs). What to do about P127 for countries and cities? I propose using country code country (P17) with some link to a generic national decree or something listing the reason for state appropriation (yes it occasionally exists, but many country-owners in P127 are currently unreferenced). In this case of national ownership, I do think it should be accompanied by country of origin (P495) as a qualifier, even if it's the same country (which for older paintings is probably a different item anyway).

1933-1945 lacunes

WWII provenance is a popular problem and will probably remain so going forward. It would be nice to agree on a basic set of properties to handle these. I am open to suggestions. We have Lost Art ID (P1428) and items for things like Database of Art Objects at the Jeu de Paume (Q58100718) plus a few others which I should probably gather up. It's confusing and I don't understand why I only see Dutch entities on View of the Rhine near Rhenen (Q27286520) that is claimed to have been in the Führermuseum (Q475667).

Qid types for provenance entities

  1. human (Q5) should be used ideally in owned by (P127) but not in collection (P195). For people who were collectors, dealers, and critics, they may have a role in cataloguing, auctioning, sales, and inheritance.
  2. art museum (Q207694) or collection (Q2668072) should be used whenever possible in collection (P195) and location (P276), with occasional use as commission (Q760089) of commissioned by (P88).
  3. I feel we miss properties for "restorer" which could be used to indicate physical alterations made a apoint in time, or for "agent" which could be used to describe parties that made purchases for female collectors unable to attend public auctions.
  4. I also feel that we miss specific properties for auctioneers or for art dealers, if you consider that owners have P127 and collections have P195. I can't think of a way to give them a specific property though. Currently they are occasionally used in those and in significant event (P793) but not in any structured way. I am open to suggestions.

In summary, I would like to propose changes that better highlight "1933-1945 lacunes" and also changes that better highlight original owners, even if the original commissioners are defunct (such as the original organization housed in Kloveniersdoelen (Q3421329) that commissioned the The Night Watch (Q219831). Jane023 (talk) 07:35, 31 July 2021 (UTC)

Thanks for bringing this up again. We have been tackling some difficult modeling issues recently and I would love to get this one more clear.
The provenance of a painting should start with creator (P170), inception (P571) & location of creation (P1071). I think we have a well established practice how to use these properties and how to use these properties in uncertain and conflicting situations. Documentation at Wikidata:WikiProject Visual arts/Item structure.
Than we have properties that describe a state. These are generally qualified with start time (P580) & end time (P582) (or related properties like point in time (P585)) in case of uncertain or missing data:
  • location (P276) - all (most specific) places this painting has visited and where it currently is
  • collection (P195) - all the collections the painting has been in and the one(s) it's currently in. This is generally an organization
  • owned by (P127) - all the people and organizations who owned the painting. Usually quite hard to figure out so not even included in the statistics yet
These three are fairly consistent and decently documented. Besides that we also use exhibition history (P608) for the exhibitions a painting was in and catalog code (P528) qualified with catalog (P972) for different catalogs the painting is mentioned in (with all sorts of qualifiers because of different title and/or different attribution).
I think everything I mentioned above doesn't need any major overhaul, maybe a bit more fine tuning. Agree?
The hard part seems to be the state changes (events) like a donation, a sale, bequest and much more. Here we have three things we want to model:
  1. The type of event. For example a bequest (Q211557)
  2. The first party in the event. For example Adriaan van der Hoop (Q2778946)
  3. The second party in the event. For example Amsterdam (Q727)
For the type of event we use significant event (P793), but how to document the parties? Depending from the event, it's viewed from the first or the second party. The example bequest (Q211557) is viewed from the first party, but acquisition (Q22340494) is viewed from the second party. I think we should first focus on solving this puzzle for paintings that have a relatively straightforward provenance. Once we get this sorted out, it's probably also easier to model the more advanced cases.
I'm considering using participant (P710) as a qualifier to significant event (P793). Than we can do it in two ways:
Opinions? Multichill (talk) 12:32, 31 July 2021 (UTC)
Thanks for this - it's very true that many events have giver/receiver roles, whether it's a donation, a bequest, or a military theft (as in the case of Napoleon or Hitler). I agree we should settle the easy ones first. Writing this down reminds me of the discussions we had about reliquaries, because the little relics needed to be blessed and so on. How do other handover actions take place I wonder? Are there examples in law or treaties? For the large group of paintings commissioned by institutions that are then absorbed by other institutions, I am unclear whether we want to include all iterations of those institutions in the ownership or just keep the last one. See this edit as an example (again in Amsterdam). Jane023 (talk) 13:24, 31 July 2021 (UTC)

Provenance, continued

Starting a new section because this is getting a bit off-topic, but considering the issue of bequests you often have the owner dying before the museum acquires the painting, sometimes by several years due to various reasons of inheritance rights. Looking at the case of convictions in law, this is also handled by two properties - see Mark Hofmann (Q4354485) used as an example for convicted of (P1399) which appears to need penalty (P1596) and they may not use the same timestamp qualifiers because though they might not occur in the same year though they are related to the same crime. We need something similar. Jane023 (talk) 14:22, 31 July 2021 (UTC)

We also need to add art gallery (Q1007870) and auction (Q177923) and the Auktion Houses conducting such transactions. Sometimes the artwork stays there for quite some time. On Commission, or on temporary ownership. Sometimes the auction house was just the agent of transfer. But especially in the case of restitution of Nazi-looted art (Q2146005) and especially with the earlier seizure (Q1151709) of artworks some names pop up more often than others, with different ways of involvement - criminally some, more with the intend of saving others, some ambivalent. And making them visible in our data, might also help shed some light on these involments if we enable them to pop up in our queries.
I also find it difficult to display the proper timeline of such transfers and ownerships if exact dates are not available and we just know that one has to have been before the other, but we do not know if there was a direct transfer, or if there is, or must be a missing link.
Than there are persons we don't have a Wikidata-Item for. Should we create one for them, just because they were transient owners by inheritance? Sometimes names pop up that do have a Wikidata-item, but can we be sure it is the identical person?. Is there a way to note such persons as write-in, so the name is recorded in a proper fashion. Look at how I tried to solve it with this Le Palais Ducal (Q48976445). I think it is not perfect and should be improved upon. --Wuselig (talk) 10:20, 1 August 2021 (UTC)
Thanks for your thoughts! You touch on a lot of different issues. First of all, with sales, see the comments by Multichill above, where he is considering two significant events per transaction when the loss is not the same date as the acquisition. I hadn't even thought of using participant (P710) with multiple significant event (P793) statements. An example of significant event (P793) with auction (Q177923) is A Panoramic View of London, from the Tower of St. Margaret's Church, Westminster (Q55451381). Years ago I tried to get a property for auction for use with lot number (P4775) but I am still grad we have one half of that puzzle. Basically all auction houses (contemporary and defunct) that handled top art should have items. Same for art galleries, and we have a lot of them already. Feel free to make more as you see fit. As far as owners that just inherited a house with pictures and passed it on to the next generation, I agree these people are not collectors, but I have decided we need them and we should declare them art collectors anyway. This includes a lot of poor art critics who could not afford art, but owned something anyway. This also includes the illegal agents and anybody else with art in their possesion. I have been bold and made Hitler and Göring the owner "when it comes to nazi loot, because I think owned by (P127) should be human items when possible, and not states or cities. I agree it gets fuzzy with art dealers, but if I have a choice, I use the person associated as owner of the gallery at time of sale. I totally agree about the owners when you think they might only have owned one painting and you want to record the same. I wouuld really like a property for "registered owner string" or something that looks like author name string (P2093). I have created lots of items for owners where I found just two paintings however. Two is enough for one item I think. Thanks for your work on Le Palais Ducal (Q48976445)! I do exactly the same thing without dates - I just add the items and hope that I can fix it eventually. One tip I have is to google owner names for relatives who might be the collector. Then once I have one "provenance chain" I can reuse it for other items that are mentioned as coming from the same family. See Nymphs dancing to a Pipe-playing Shepherd (Q107148574) as an example for the Lowther family. Jane023 (talk) 09:35, 4 August 2021 (UTC)
On more thing about owners is that I have noticed omissions in the leading documentation when it comes to female owners, especially widows who outlived their husbands by a decade or more. It's very important to set up the dates for ownership for them, because often they made purchases and donations in their husband's names, but they were the agent. A big problem when you dont do this is where everyone talks about Mr. XXX collection but they really mean something that Mr. XXXX never even saw. Jane023 (talk) 09:43, 4 August 2021 (UTC)

Ideas for more structured work

Good morning from Vienna, currently I am trying to participate on this project. Especially I attempt to add claims concerning the property "ICONCLASS"(P1257). Many questions arise. I would like to cooperate with any experienced/interested user. Is there a way to discuss a joint targeted approach with community/special users? It could be very helpful if I were able to find out who recently published changes regarding the decisive items (all paintings, particularly with P1257). The website with the url "https://www.wikidata.org/wiki/Special:RecentChanges" provides some filters, but I am not capable to restrict results to our Wikiproject. Is there any method to build a sparql query by my own? In my opinion it would be very useful to add such a subpage to a sitemap (or to "https://www.wikidata.org/wiki/Category:WikiProject_sum_of_all_paintings"). Any ideas are appreciated. Many thanks in advance, regards forom Austria Hippokrene (talk) 07:24, 1 September 2021 (UTC)

Welcome to the sum of all paintings project! I have added iconclass to paintings in the past, and as I recall, the difficulties are mostly to do with the lack of publicly available data in datasets. If you have this, please add it. There are two major parts to this: adding iconclass data to paintings, and adding iconclass data to art genres, popular art themes, and depicted objects in paintings. If tell me what kind of list you want I will build it for you to play around with and you can start adding some of your own. Do you want just paintings located in or from Vienna? Jane023 (talk) 08:09, 1 September 2021 (UTC)

Thank you for your reply. My personal list contains about 34000 items of religious art, I am going to split it into some subgroups related to biblical scenes or persons. This topic is from my point of view most suited for assignments in Iconclass. Hippokrene (talk) 13:09, 1 September 2021 (UTC)

That's a lot! We currently have 3914 concepts mapped to their Iconclass equivalent using Iconclass notation (P1256) )query is here) and we have 14273 artworks depicting such iconclass concepts using depicts Iconclass notation (P1257) (query is here). Jane023 (talk) 19:02, 1 September 2021 (UTC)

Adding creator lists to items of the form "list of paintings by <creator name>"

Can't believe this didn't occur to me before! See this edit Jane023 (talk) 11:17, 1 October 2021 (UTC)

have a look at lighthouse lists ;) --- Jura 11:25, 1 October 2021 (UTC)
Seems that items don't show on Petscan for Wikidata: https://petscan.wmflabs.org/?psid=20653014 --- Jura 14:08, 14 November 2021 (UTC)

Painting cut and reframed

At Ramon Casas and Pere Romeu on a Tandem (Q4889277). I wasn't quite sure how to add that, so I used resizing of painting (Q109500248).

Checking further, I found two paintings that use cutting (Q196751) in significant event (P793).

There is also disassembly (Q52161698). --- Jura 18:29, 9 November 2021 (UTC)

Interesting! There is also reassembly after an unknown cutting event. Not sure if this is reflected in significant event though. I suppose as forensic analysis of wood samples and so forth increases we will see more reassembly events. Also maybe overpaint removal events, but I am not sure how to model those. Jane023 (talk) 07:42, 15 November 2021 (UTC)
Depending on the extent of the "change", a new item should probably made for the resulting (state of the) work.
For my case (Q4889277), I was wondering how I should cross-reference the significant event with the changes in some of the statements (size, inscription, image). --- Jura 11:41, 15 November 2021 (UTC)

length (P:P2043) on paintings

Looks like some items use this instead of width (P2049) (or possibly height (P2048)) (sample, query with ca. 1000). I added a corresponding constraint to the property [1]. --- Jura 12:43, 15 November 2021 (UTC)

±1 (in width and height)

Some time ago, ±1 was removed from many values, given that the formerly required value was generally mostly guesswork or a default value.

It seems to be still present in some painting sizes, e.g. Q9015019#P2049.

Sample query for height:

SELECT * { 
          ?item wdt:P31 wd:Q3305213 .
          ?item p:P2048/psv:P2048 ?st . 
          ?st wikibase:quantityAmount ?value .
          ?st wikibase:quantityUpperBound ?valueLower .
          ?st wikibase:quantityLowerBound ?valueUpper .
          ?st wikibase:quantityUnit ?unit 
         BIND( ?value - ?valueLower as ?diffLower )
         BIND( ?valueUpper - ?value as ?diffUpper )
         }

Try it!

Currently gives ca. 10000 items.

Personally, I'd remove them here as well. --- Jura 14:08, 14 November 2021 (UTC)

I made a request at Wikidata:Bot_requests#request_to_remove_±1_(in_width_and_height)_from_paintings_(2021-11-15). --- Jura 09:40, 16 November 2021 (UTC)

units of height and width

SELECT  *
{
  ?item wdt:P31 wd:Q3305213 .
  ?item p:P2048 /psv:P2048 [wikibase:quantityAmount ?height ; wikibase:quantityUnit ?h_unit ] .
  ?item p:P2049 /psv:P2049 [wikibase:quantityAmount ?width ; wikibase:quantityUnit ?w_unit ] . 
  FILTER( ?h_unit != ?w_unit )
}

Try it!


Seems odd that values from the same source are sometimes (correctly) imported in different units (mm and cm).

Unsurprisingly some big computing company tends to get them wrong.

Not that the last point matters, but, in general, I think it would be preferable to use just set of units, e.g. "cm" for 25 cm x 15.5 cm at Q24175739.

An exception may be Q3362985 (2.6 m x 25 cm), but then, maybe P31=Q3305213 shouldn't be used there. --- Jura 07:33, 17 November 2021 (UTC)

museum etc. "owner of" painting (inverse of "owned by")

SELECT ?item ?itemLabel ?itemDescription ?count 
WITH
{
  SELECT ?item (COUNT(*) as ?count) 
  {
    ?painting wdt:P31 wd:Q3305213 .
    ?item p:P1830 / ps:P1830 ?painting . 
    hint:Prior hint:rangeSafe true .  
    ?painting p:P127 / ps:P127 ?item .
  }
  GROUP BY  ?item
} as %count
WHERE
{
  INCLUDE %count
  SERVICE wikibase:label { bd:serviceParam wikibase:language "[AUTO_LANGUAGE],en". }
}
ORDER BY DESC(?count) ?item

Try it!

Not sure if there is much use in these inverse statements, e.g. at Q95569#P1830. The information is already available on the items themselves. I'd remove them. --- Jura 06:08, 17 November 2021 (UTC)

I added a request at Wikidata:Bot_requests#request_to_remove_inverses_with_P1830_for_paintings_(2021-11-23). --- Jura 12:09, 23 November 2021 (UTC)

catalogue raisonné (P9969) and catalog code (P528)

SELECT ?painter ?painterLabel ?catalogue ?catalogueLabel ?catalogueDescription ?count ?sample_painting
WITH
{
  SELECT ?painter ?catalogue (COUNT(DISTINCT ?painting) as ?count) (SAMPLE(?painting) as ?sample_painting) 
  {
    hint:Query hint:optimizer "None".
    ?painter wdt:P9969 ?catalogue .
    ?painting wdt:P170 ?painter .
    MINUS { ?painting p:P528 / pq:P972 ?catalogue } 
    ?painting wdt:P31 wd:Q3305213 .
  }  
  GROUP BY ?painter ?catalogue
} as %inc                  
{
  INCLUDE %inc
  SERVICE wikibase:label { bd:serviceParam wikibase:language "[AUTO_LANGUAGE],en". }
}  
ORDER BY DESC(?count)

Try it!

Above an attempt to cross-check them .. it needs some work for those with separate items for volumes. --- Jura 13:32, 18 November 2021 (UTC)

Resolution of image

Please see Wikidata:Request_a_query#Resolution_of_image_(on_item_for_painting). --- Jura 05:36, 13 November 2021 (UTC)


After some back and forth, here the scale now at this list (2nd column):

# label benchmark criteria dependency
1 very low image width < 300px (Wikipedia infobox size) file
2 low Wikipedia infobox size image width < 800px and image heigth <600px (file description page default) file
4+ high or very high A4 print >300 dpi at size A4 (larger of width and height compared to 297 mm) file
3+ medium or higher Commons file description page default all other (not 1, 2, 4+)
5 very high print in original size currently not used. Should be based on dpi at original painting size. Reasonable dpi depends on size file and original size

The query checks for 5% within the limits. Small paintings (<A4) could be at "3+" while having higher resolution.

It's an approximation that could allow to find paintings that could benefit from better images. --- Jura 09:47, 16 November 2021 (UTC)

I added a proposal at Wikidata:Property_proposal/dpi. --- Jura 12:17, 27 November 2021 (UTC)

image with color scale

Many photographers pay attention to this, but it's not always easy to verify if colors are accurate.

A few images on Commons include a visible color scale, e.g. the one to the right above.

Similar to image with frame (P7420) maybe we should cross-reference them here too. @Jane023: --- Jura 13:09, 15 November 2021 (UTC)

Interesting - sometimes the colors are indeed pretty ghastly compared to the actual thing. Not sure if it would be widely used though. Jane023 (talk) 13:36, 15 November 2021 (UTC)
Initially, it might be like P:P7417: something good to have, but generally not available.
However, it's easier to make and it might encourage people to upload it (or make more systematically available). --- Jura 17:48, 15 November 2021 (UTC)
Some I found since on Commons. --- Jura 18:55, 15 November 2021 (UTC)
image with color chart (P10093) is now available. --- Jura 12:19, 27 November 2021 (UTC)

Property proposal - Harvard Art Museums ID

I've proposed a new property at Wikidata:Property proposal/Creative work#Harvard Art Museums ID. Comments and suggestions for improvement welcomed! - PKM (talk) 00:59, 27 November 2021 (UTC)

Sounds good.
The general approach with "described by url" should probably be reviewed. While the Listeria lists can be modified to also include external id with them, I'm not sure of a good way to do it for the Commons artwork template. --- Jura 12:28, 27 November 2021 (UTC)
I wouldn't mind thhe external ID for the museum record disappearing from the artwork template on Commons if it was added back in the location subtemplate somehow. Now for location you just see the institution template, but this should be linked to a new subtemplate with the Accession number and accompanyinng link. Jane023 (talk) 08:38, 29 November 2021 (UTC)

Cleveland Museum of Art - project chat discussion

There is a proposal at Wikidata:Project_chat#Bot_work_for_Cleveland_Museum_of_Art (includes some paintings, but not primarily). --- Jura 12:22, 27 November 2021 (UTC)