paulduchesne
commented
1 year ago new thread for manifestations here @ladislav-nfa @natashafairbairn @heftberger @stephenmcconnachie @torbjornbp
Closed paulduchesne closed 1 year ago
paulduchesne
commented
1 year ago new thread for manifestations here @ladislav-nfa @natashafairbairn @heftberger @stephenmcconnachie @torbjornbp
paulduchesne
commented
1 year ago I suppose we should start with Manifestation Type 2.2.1, D.5 has a fairly extensive vocabulary:
I would be very interested in hearing whether these are normally implemented as is, or anyone is aware of reductions or expansions to the vocabulary.
torbjornbp
commented
1 year ago The manifestation entity is an entity I feel highlight the laser focus on published materials and libraries, the original bibliographic standards had. We have struggled to find a practical approach to the entity for a lot of the complex archival materials we have (fragmentary production materials for example).
In our initial discussions on the manifestations for film here at the NLN, we saw two approaches to the manifestations: purism or pragmatism. Keep in mind that these discussions were centred around CMS implementation, more than modelling.There are quite a few suggested manifestation attributes related to technical format etc (appendix K: from 2.3 and down). I'm not particularly fond of those attributes being there to be honest.
In the purist approach the manifestation represented an ideal item so to speak. The reasoning were that if we were to put technical format attributes on the manifestation level, these attributes would also have to be true for the child item records. For example we would not group material with different ideal technical attributes, if we put that information at the manifestation level. A good example is extent/duration (this would necessarily differ between the original cut negative, the camera negatives, the myriad magnetic sound elements, etc etc)
It would lead to a high degree of inheritance to its items, which again could the ease the cataloging of “published materials” (prints, dvds, etc) where we have many identical items (then again this could also be solved by duplicating identical items). The perceived cons though were that, as we have lots and lots of “unpublished materials”, we would be left with a TON of manifestations. In many cases as many manifestations as we had items… In essence the manifestation would then have little value for structuring items for most posts.
In the pragmatic approach we would put very little technical attributes at the manifestation level, and rather use it to group all material related to some kind of production event. Eg. all material related to the production of the original 35mm theatrical release in one manifestation (whether to group negatives/prints is not decided), all material related to the 16mm distribution, several home video manifestations, and most likely some digital preservation manifestations.
The manifestation then could be used quite effectively in grouping/ordering items on posts with a lot of items (films having undergone several rounds of both analog and digital copying etc.) The cons were that it would be sort of non-standard. In addition it would possibly demand a quite complex ruleset for manifestation boundaries for our cataloguers to follow.
This touches on the manifestation type question you raise @paulduchesne. Different approaches and manifestation boundaries will have different granularity/detail needs of typification. Perhaps some other type field can solve it in a CMS, but still. We are still viewing some variation of the pragmatic approach as the only feasible one moving forward at the moment.
I guess another headache the manifestation has given us is where to put various duplications. Should the digital carrier of a digitised video cassette spawn a new manifestation, or does it belong under the manifestation of what its a copy of? I reckon we would have to have some sort of hybrid approach to this, but I don't find it to be particularly straightforward...
Very curious to hear how others are thinking about the manifestation entity! I find it raises more questions than the variant to be honest.
paulduchesne
commented
1 year ago @torbjornbp thank you for framing this discussion so well, I would agree that an issue for using technical terms at a manifestation level is that many "manifestations" will comprise different components, e.g. different gauges, stock, technology for production components, or the not insubstantial time when an "original release" comprised both a 35mm print and a sibling DCP.
On the other hand, I am aware that this is where archives can retain information about known characteristics of a film which is not held (e.g. the gauge and duration of a film premiere), but a possible solution which just occurred to me - why could you not have an "empty" item? I have always operated under the assumption that an Item equates to a physical item which is held in the collection, but there is nothing to prevent it from being simply a record of the characteristics of one of the items which sits under the manifestation, whether it is extant or not.
In addition it would possibly demand a quite complex ruleset for manifestation boundaries for our cataloguers to follow.
Also in terms of data transformation, while we may agree on a structure, we are also going to be parsing data from institutions who may (will!) have different definitions.
I guess another headache the manifestation has given us is where to put various duplications. Should the digital carrier of a digitised video cassette spawn a new manifestation, or does it belong under the manifestation of what its a copy of? I reckon we would have to have some sort of hybrid approach to this, but I don't find it to be particularly straightforward...
I have always felt uncomfortable about film scans of original release going in under the "original release" manifestation, e.g. a DPX image sequence produced from an archival film scan is very different to a DPX image sequence which was used as part of the production - my inclination would be to place such scans under a D.5.9, preservation/restoration manifestation?
stephenmcconnachie
commented
1 year ago We’ve always worked to the pragmatic principle that a digital derivative without change to the content, inherits the same parent Manifestation as the physical / analogue source. So where we digitise a videotape carrier for preservation, for example, the FFV1 Matroska sits under the same Manifestation as the DigiBeta Item.
Where the digital derivative undergoes transformation – classic example being a restoration of a feature film – the digital Items live under the Manifestation for that Restoration. So imagine a raw ungraded DPX scan from a film element, would stay in the same Manifestation as the film Item; when the restoration creates a new set of DPX files, those live under the new Restoration Manifestation. And if we create a ProRes MOV specifically for distribution or BFI Player, that ProRes MOV Item lives under the Manifestation for the BFI Player.
We explored the idea of a format migration at the Item triggering a new Manifestation for the new derivative (I believe CNC and maybe others use this approach) but we ruled it out immediately due to the immense overhead of data creation and – more critically – the negative implications for understanding of the collection in context. We believe our method makes it better and easier to understand the Items in their context, within the lifecycle of the Work. Not all Events merit a Manifestation, and in our logic, an identical clone of physical to digital at the Item level does not merit a new Manifestation. The opposite has been argued, as I said I believe CNC followed that route.
torbjornbp
commented
1 year ago We’ve always worked to the pragmatic principle that a digital derivative without change to the content, inherits the same parent Manifestation as the physical / analogue source. So where we digitise a videotape carrier for preservation, for example, the FFV1 Matroska sits under the same Manifestation as the DigiBeta Item. Where the digital derivative undergoes transformation – classic example being a restoration of a feature film – the digital Items live under the Manifestation for that Restoration.
Good to hear, this is quite similar to what we have been thinking lately! “Unprocessed copies” (like DPXes straight from the scanner) at the same manifestations as what its copied from, with a set amount of processing being needed to trigger a new manifestation. It will make video digitisation very orderly, but for film preservation I imagine it can become a bit messy. Preservation projects picking from many generation of digitisations from many different elements. The fact that it becomes messy might just be a fact of film preservation in general though...
I guess some order could also be reinstated by relationships to preservation events. The work going on in the Preservation/Restoration Documentation task force is probably relevant here.
It would be interesting to hear the thoughts of someone going for a strict new format/triggers new manifestation ruleset.
By the way, I guess that means you are still putting format information on the manifestation at the BFI? We figured if we were to put different format items on a manifestation, we would omit putting any such information at the manifestation level (or just put information that would be true for all its items there). I just had an idea that having these format attributes at the manifestation level could be used to differ between primary and secondary items in a manifestation. Sort of like “representative item attributes”. The items matching the attributes at the manifestation levels are primary/original, while the items with differing attributes are secondary/copies.
paulduchesne
commented
1 year ago By the way, I guess that means you are still putting format information on the manifestation at the BFI?
This would mitigate the issue I was describing in trying to determine if DPXs under a preproduction manifestation were digitised preproduction elements, or preproduction elements themselves. I was also keen to put forward the idea that what we are discussing here is the appropriate use of the structures rather than the structures themselves, so while this would be crucial for appropriate transformation, maybe something we can avoid making firm decisions on during the modelling phase.
I would be very keen to test cutting the technical elements for manifestation as @torbjornbp has suggested, although a related "interpretation" issue is where archives have used technical manifestation info to mean "all of the child items are 35mm" vs "the ideal for this manifestation is 35mm film", which are clearly different things.
ladislav-nfa
commented
1 year ago Before I write my comments about Manifestation, I would like to ask you to clarify for me the concept of parent-child relationship as I am bit confused now.
Torbjørn wrote: "In the purist approach the manifestation represented an ideal item so to speak. The reasoning were that if we were to put technical format attributes on the manifestation level, these attributes would also have to be true for the child item records."
I understand the ideal Manifestation concept in the way that Manifestation and Item share some (but not all elements) and the difference is between values of these elements, not betwen the elements.
So the ideal physical extent of Manifestation would be, let´s say, 3100 metres (something we need to establish from sources like distribution information not from the film material itself), whereas the combined print (Item) in the archive will have about 2600 m, indicating the Item is incomplete. Do you understand it in the same way?
torbjornbp
commented
1 year ago @ladislav-nfa Yes, that is what I meant. In the strict purist approach we figured manifestation could hold an ideal value of an element. The elements would also be shared in some extent with the items. Differing elements would serve as manifestation boundaries.
In the pragmatic approach I would also group together things that have different elements, not just values. For example physical and digital items. The DPX of a scanned 35mm film will necessarily not have the same format elements, as its physical counterpart.
Someone else should probably chime in here. Here at the NLN this is still theoretical and not implemented in any practical sense...
stephenmcconnachie
commented
1 year ago So we've been using this approach to the Manifestation since 2011, and it's the only approach we could use in our high volume archive operations - the new-Manifestation-per-format-change would be unworkable as a data model for use in managing the collection. It works, I believe. The Manifestation for us is tied to the event in the work lifecycle, not to the formats of the Items beneath. So though we do capture some formats in our Manifestations, they are high level not specific - eg digital terrestrial broadcast, not MPEG-TS; Digital Video not FFV1 MKV. There is an exception for digital cinema, where for new theatrical releases we tend to state DCP in the Manifestation - while the Items beneath could comprise a DCDM, a DCP and maybe a ProRes MOV.
We create Manifestations for moving image works that we collect and for those we don't - eg we create works / manifestations for every feature film released in mainstream cinemas in the UK. The Manifestation represents the UK theatrical release. In reality, we don;t collect non-UK feature films in the collection, and we only manage to collect a percentage of UK feature films. But the Manifestation is the record of the UK cinema release event in the work lifecycle.
So the Manifestation may have some format info, but those are indicative / ideal - the specific, granular formats are held in the child Items, for those where we do acquire an ITem into the collection. Hope this is useful!
ladislav-nfa
commented
1 year ago In our archive, we do not have a manifestation level in our film collections system - there is Work and Item level only. We do have a manifestation level in our other system (filmographic information about Czech feature films), primarily the original manifestation of the manifestation type “Theatrical distribution (country of origin)” but these manifestations are not connected to the Items described in our older system.
So the way of describing manifestations is something we will have to specify.
In our old system, production materials are described as individual Movie Image Works, so for instance there are “Baron Prášil” (Munchhausen) and “Baron Prášil. Denní práce” (Munchhausen. Rushes). The rationale is that these production materials have different content, at least there could be different cast (actors whose scenes were cut off in the final cut).
/Production magnetic tapes (containing sound effects and/or music and/or dialogues) are not yet described (we have a huge collection from rather new acquisition) but they will be described differently and connected to Sound Work, not Moving Image Work. This is a rather complicated story which I will skip for the time being./
The manifestation for production materials - cataloged under the separate Works umbrella, such as “Baron Prášil. Denní práce” example - could be described just by type element (Manifestation type = “Production material” or something like that) and all other information will be on the Item level only. This is, I guess, a kind of similar way of handling production material manifestations as Torbjørn described.
Besides production materials, I would propose the approach mentioned in the manual that “internal” copies of Item (copies made by an archive for in-house use) will be just another Item under the same Manifestation. So all beta copies, digitized version (telecine), “print-screens” (video recording of film projection - we use it during the in-house projection of films for cataloging committee projections) as well as black-and-white copies of color films (made for economical reasons) or 35mm-to-16 mm reductions will be different Items under the same Manifestation the source Item belongs to. There is some kind of analogy to book cataloging as you can find print copy (“xerox”) of the book among the Items under the same Manifestation (the issue, having ISBN). This is not uncommon in libraries (if you lose a book in a university library, sometimes you are asked to provide them with a copy of it).
I am trying to see Manifestation (at least in the case of analogue films) like something you could ideally see and hear in the theater whereas the film Item is something you have to explore “in hands”. So most physical aspects of film (material brand name, number of splices etc.) may be cataloged solely on the Item level. The most important information about an Item is probably its physical condition and access status. This is also not uncommon to libraries, where you can have information about the book Item such as “missing 10 pages”.
I have many questions about the Manifestation level. I start with following.
First off, Manifestation is not just an ideal physical entity (technical properties) but also the Publication Event. I do not see any difference between D.9 (Work/Variant Publication Type) and D.10 (Manifestation Publication Type). In my opinion, Work cannot have a Publication Event per se, for the Publication Event = Manifestation. Also, I do not see much difference between the “D.10 Manifestation Publication Types” and Manifestation Types. For the former, there is for example “Theatrical distribution, Non-theatrical distribution, Not for release” - it is the same for D.5.2., D.2.3 and D.5.4. If the Publication Event was meant to specify Manifestation Type (type “Release” looks like that), it could be useful. For you can have different dates (sub-events) in one Manifestation - for example - Release date in the country, Release date in a city (we do have this data), Distribution date range (from release date to last projection date) etc. Or to make distinction between Release and Re-release?
Secondly, there was theatrical distribution of the original manifestation in two “versions” in Czechoslovakia. Both 35mm film (original) and 16mm film (reduction for village cinemas) was distributed across the country. Do they constitute two manifestations (both type Theatrical distribution, with distinction on carrier (35mm vs. 16 mm - it has consequences on physical and logical extent as well)?
Thirdly, we have an archival theater which screens on a one-film-one-day basis (unique screening). So in the last ten years, one film could have been screened five times, one day every two years. Does all of this screening mean separate manifestations (type D.5.3 Non-theatrical distribution, subtype film club screenings)?
natashafairbairn
commented
1 year ago Coming late to the Manifestations party...
Linking this to Ladislav's points above, I would say yes the 35mm and 16mm versions should have separate Manifestations and both would be Theatrical; but that screenings data re. an archival theatre is an Event rather than Manifestation - thus the Theatrical Manifestation of the Work would have five linked Events reflecting each screening. [NB. Just to totally contradict what I have just said we keep a record of when a film was shown at BFI Southbank - with details about the version, running time, and suppliers (sometimes it is from our own Archive but often not) - but we link this to the Work. Being purist it should be linked to Manifestation, but it was a) easier in terms of migrating data from different databases and b) easier from a user point of view (both internal and external) to be able to see quickly and easily all the different screenings data for the one film in one place rather than scattered across multiple Manifestations (sometimes under differing titles depending on the version screened).]
I think a lot will depend on the nature and amount of an archive's holdings and acquisition/internal copying rates. In terms of having Items with a differing format to the linked Manifestation, as mentioned that occurs where an in-house copy of an Item is made and linked to the same Manifestation as its source. The FIAF Manual references a choice here, either doing the former or having a separate Preservation Manifestation and linking such Items under that. Either is fine. An Archive deciding the rather more purist latter line would also then have to consider whether it should create a new Preservation Manifestation for each time a copy was made, or else create a new one only if the format differed, or else create it once and link all subsequent in-house copies on different digitial or other formats to that one Preservation Manifestation. Because the BFI has made so many in-house copies over the years (eg. viewing copies on 35mm, video, DVD for research use; digital copies on different digital types, etc. having disparate Preservation Manifestations would lead to confusion and a massive increase in the number of Manifestations (when we already have loads due to migrated legacy data - much of which needs merging).
The "ideal" running time for a Manifestation has been mentioned, and often the Item acquired will match this. However, sometimes the acquired Item may be slightly shorter, e.g. if it had some censorship edits but that hadn't been identified yet, or if only a partial acquisition, i.e. only 3 reels of an original 4 reel film had ever been acquired by an archive. This is why running time is in both Manifestation and Item.
natashafairbairn
commented
1 year ago On a new tack... Manifestation Type = “A word or phrase denoting the relationship between the manifestation and the variant or cinematographic work that it manifests. May be omitted if no other manifestation is known. An unknown relationship should be indicated by a value of "unknown". Values for this attribute should be taken from a controlled vocabulary…”
There is no rule within EN15907 as to the categories of Manifestation Type there can be, only that its values come from a controlled vocabulary list. The only one stipulated value that exists is that of a category of ‘Unknown’. In the FIAF Manual we came up with a possible controlled vocabulary list incorporating types that covered the usual expected main types of manifestations of Theatrical release, Non-theatrical release, Home Viewing, etc. but also types that reflect the in-production or post-production process, e.g. Pre-Release, and for use with other moving image acquisitions an archive may have, e.g. home movies that were never intended for or had a theatrical release, screen tests, in-production materials, etc.
the Manifestation Type is one of the fundamental key elements of EN 15907 which are crucial to the strength and flexibility of the structure and its ability to incorporate and reflect both the nature of the Work and its potential multiple differing embodiments/manifestations.
torbjornbp
commented
1 year ago So the Manifestation may have some format info, but those are indicative / ideal - the specific, granular formats are held in the child Items, for those where we do acquire an ITem into the collection. Hope this is useful!
Super useful! I guess this is outside this discussion, but I would be very interested in seeing the BFIs manifestation boundary ruleset (particularly for digitised materials), if they exist in some shareable form. We are working on this at the NLN right now, drawing up different variants of manifestation granularity and their different consequences.
Thirdly, we have an archival theater which screens on a one-film-one-day basis (unique screening). So in the last ten years, one film could have been screened five times, one day every two years. Does all of this screening mean separate manifestations (type D.5.3 Non-theatrical distribution, subtype film club screenings)?
I would interpret it as not needing a new manifestation. The manifestation holds the information around initial release/production and items holding the history of the individual prints. Even though a film was released in theatres in 1970, it doesn’t mean that a 1970 print from that release was only screened in 1970. We currently keep track of such later screenings using a combination of loan/movement/treatment information on the particular screened item. Not optimal, but I reckon events could clean this up a bit.
paulduchesne
commented
1 year ago The "ideal" running time for a Manifestation has been mentioned, and often the Item acquired will match this. However, sometimes the acquired Item may be slightly shorter, e.g. if it had some censorship edits but that hadn't been identified yet, or if only a partial acquisition, i.e. only 3 reels of an original 4 reel film had ever been acquired by an archive. This is why running time is in both Manifestation and Item.
What did you think of my "ghost item" suggestion above? This is the concept that rather than applying characteristics to a manifestation, you instead add an empty item to carry that data. I realise this could be challenging if used in a CMS, as I imagine most systems expect that anything at Item level link to something physically on a shelf, but I could see this being useful if a single manifestation comprised items and screening data exhibiting different characteristics (if this does not already force a new manifestation by definition).
An example, the long-lost premiere 70mm screening print of "Lawrence of Arabia", and a subsequent significantly-shorter 35mm original release prints. I would have thought these would both live under the same original "Theatrical distribution" manifestation, however what is the "ideal length" and "ideal gauge" in this instance? Hmmm another problem, if you did hold the 70mm print you would want to explicitly link this to the premiere screening, which would get lost if both items/events had a generic manifestation in the middle.
For you can have different dates (sub-events) in one Manifestation - for example - Release date in the country, Release date in a city (we do have this data), Distribution date range (from release date to last projection date) etc.
That was my understanding, an original release would have a "Theatrical distribution" manifestation release with attached screening events. These could be date ranges, although I'm excited about the concept of merging in highly granular screening data (eg https://www.cinemacontext.nl/).
Secondly, there was theatrical distribution of the original manifestation in two “versions” in Czechoslovakia. Both 35mm film (original) and 16mm film (reduction for village cinemas) was distributed across the country. Do they constitute two manifestations (both type Theatrical distribution, with distinction on carrier (35mm vs. 16 mm - it has consequences on physical and logical extent as well)?
I may have mentioned this above, but another example which was quite prevalent for many years in the later 2000s, is that we would get a couple of matching 35mm prints and a DCP for big releases, which we would play in parallel as a balance between not having completely converted to digital vs the cost of striking prints. I feel these would be of the same manifestation in this case, they would often arrive in the same box.
paulduchesne
commented
1 year ago Another question to throw in while we are talking about manifestations and release events, is it common to see many-to-one manifestation/item relationships? Here I am thinking about a print which is part of an original "theatrical distribution" release, and then later used for "non-theatrical" screenings - or is it preferred that manifestation reflects "original intention" of the item?
ladislav-nfa
commented
1 year ago "An example, the long-lost premiere 70mm screening print of "Lawrence of Arabia", and a subsequent significantly-shorter 35mm original release prints. I would have thought these would both live under the same original "Theatrical distribution" manifestation, however what is the "ideal length" and "ideal gauge" in this instance? Hmmm another problem, if you did hold the 70mm print you would want to explicitly link this to the premiere screening, which would get lost if both items/events had a generic manifestation in the middle."
Paul, I would suggest this is the case of Variant type, not the Manifestation/Item.
torbjornbp
commented
1 year ago Another question to throw in while we are talking about manifestations and release events, is it common to see many-to-one manifestation/item relationships? Here I am thinking about a print which is part of an original "theatrical distribution" release, and then later used for "non-theatrical" screenings - or is it preferred that manifestation reflects "original intention" of the item?
Does the standard even allow for items being attached to multiple manifestations? In our proposed data model here at the NLN we would not allow for an item to be attached to several manifestations. I'm struggling to see it resulting in anything but a labyrinthine web of relationships.
ladislav-nfa
commented
1 year ago Does the standard even allow for items being attached to multiple manifestations? In our proposed data model here at the NLN we would not allow for an item to be attached to several manifestations. I'm struggling to see it resulting in anything but a labyrinthine web of relationships.
I understand the conceptual risk of this but this could happen very easily. It was one of my next questions as I was thinking about it very often.
If we loan the combined print to festival screening and a year later the same Item will be screened in our archive theater, than there are two manifestations - the former perhaps theatrical distribution, sub-type: festival screening (like in Bologna), the latter being non-theatrical distribution: sub-type: film club screening.
ladislav-nfa
commented
1 year ago Some time ago I tried to formalise manifestation types and sub-types (using FIAF manual and some other sources) like that. (It is just the first sketch).
paulduchesne
commented
1 year ago Thank you for sharing that @ladislav-nfa!
I think going into the next meeting I have two questions I would like to explore:
1) Should technical attributes (ie 2.3.X) be available for manifestations? I see two arguments above for their use, convenience in manifestation disambiguation through a CMS, and expressing data which refers to ideal state supplied from secondary sources. In the first case, I think we are working with systems which enable "pulling" through layers, so that a manifestations can be defined by the characteristics of the item underneath it. With the "ideal" attributes, I vaguely feel this would be better expressed between item and the event which corresponds with the characteristic being observed. Having said this, it would be foolish to go against a popular understanding, especially if this enables conforming data which is built based on a common understanding - ie the model should serve the archives, not the archives serve the model.
2) What are the triggers to force/define a new manifestation? As mentioned this is not really a question about structure, as we only have to define the vocabulary, not how the vocabulary should be applied, although if we start to experiment with applying to real data it will quickly arise. A bit of history, when I was working with the Murnau project this was one of the most difficult aspects, merging manifestations from different institutions, some with very elaborate manifestation data and some without the tier at all. In the end I "built" manifestations by grouping items by format information (eg 35mm nitrate film, dvd) but this was not very satisfactory as it provided no insight into whether the 35mm nitrate film was a production component, theatrical release print, answer print etc etc, which is what is explicitly addressed by the vocabulary. This leads on to the question of what to do with manifestations where there is minimal data - should you default to grouping (they are all 35mm film!) or keep separate until you are confident that they belong together.
paulduchesne
commented
1 year ago Here is the current representation for manifestation, both turtle and webvowl. As you can see, without technical elements it is pretty lean!
ladislav-nfa
commented
1 year ago As far as the ideal manifestation is concerned, the matter is more complicated due to the fact that data on ideal manifestation may vary from one source to another. One of the sources for the Czech environment is (especially for Extents information), for example, the magazine Filmový přehled intended for persons responsible for movie theater operating. The magazine however took technical information from different sources. Another source is, for example, censorship lists (for films before 1945). So the information about the ideal manifestation itself is more a matter of probability or estimation than certainty. The ideal case is, perhaps, to record different values (for example multiple physical extents) for one Manifestation if this is the case, with quoting the source for each of the values.
As for Item, an Item footage that differs slightly from the ideal Manifestation footage does not necessarily mean that the Item is incomplete (see original negatives for silent films being shorter than prints as negatives are cued by flash intertitles).
Our archive practice is always that each Item (analogue film material) is physically inspected and measured (footage, image format, etc.). In the future, when we have a manifestation level, we will probably first create a new manifestation for each new Item, and only later possibly merge (deduplicate) it with existing ones, or keep it as a new one and describe it in more detail. Item and Manifestation descriptions will almost certainly not be done by the same cataloguers. I think the case of digital manifestations (such as DCPs) is a bit different as most of the technical properties could be extracted using software characterization tools such as MediaConch. I do not want to go in detail for the time being.
I don't know whether this forum is a place to discuss the particular technical metadata elements common to Manifestation and Item. Anyway, this would be very useful. See for example the distinction between image format and projection format that is very important for Manifestation - Item relationship. See the case of 35 mm films with format image roughly about 4:3 which were intended to be projected widescreen (i. e. involves matting out the top and bottom of the film frame). This is something which cannot be for sure established by rough Item inspection and there were cases when we - having not this information - screened these films without matting out (so you could see mike on the bottom of the projected film, to give an extreme example of the issue).
paulduchesne
commented
1 year ago Discussion supports that manifestation level technical data is widely implemented, so I will add in these elements over the coming days and look forward to any feedback!
paulduchesne
commented
1 year ago Quick tech note, I have added a Github actions pipeline which combines the individual turtle files into a single ontology file, validates for syntax errors and writes documentation out to the README.md.
See for example the distinction between image format and projection format that is very important for Manifestation - Item relationship. See the case of 35 mm films with format image roughly about 4:3 which were intended to be projected widescreen (i. e. involves matting out the top and bottom of the film frame).
I feel like I have seen this spit into "aperture" and "aspect ratio" as two different attributes (maybe you suggested this?). EDIT: This is present in the manual under 2.3.4.2.
~
I also had a question around duration - is it just an extant with "minutes" as the unit of measurement or something fundamentally different?
ladislav-nfa
commented
1 year ago I feel like I have seen this spit into "aperture" and "aspect ratio" as two different attributes (maybe you suggested this?). EDIT: This is present in the manual under 2.3.4.2.
Yes, I meant that. I would prefer the term "image format" vs. "projection ratio" (synonyms in manual), as image format is also a kind of aspect ratio.
However, I discussed the aperture term and values (D.7.15 Aperture) with colleagues working with film materials and they do not understand what is meant by that in this context.
Currently, we do record “image format” information (the actual exposed image on the film frame) on the Item level, with aspect ratio values like 1 : 1,19; 1 : 1,33, 1 : 1,37 etc.
D.7.15 values like Full Height, Full Screen or 3D do not seem to them as workable for image format description.
So I would appreciate any explanation of what is meant by aperture.
torbjornbp
commented
1 year ago How about using a single image format attribute, with typification (camera/gate format, projection format) and multiple cardinality?
ladislav-nfa
commented
1 year ago How about using a single image format attribute, with typification (camera/gate format, projection format) and multiple cardinality?
I would like to have that!
paulduchesne
commented
1 year ago I had included this as projection characteristic as it is collected under 2.3.4.2, but I can see what you are saying. Maybe it would be better expressed as image characteristic (with the subclasses @torbjornbp has suggested above)?
I can also see some confusion between the definition of aperture/image format, which is very explicitly about exposed camera image area, and the D.7.15 vocabulary which includes terms which are related but not completely appropriate for that definition (eg pan and scan, 3d).
annahoegner
commented
1 year ago I also had a question around
duration- is it just anextantwith "minutes" as the unit of measurement or something fundamentally different?
I don't see a major difference here. In 0.2.1 Elements of description across Works, Variants, Manifestations, and Items it is mentioned as a "temporal extent"
paulduchesne
commented
1 year ago I don't see a major difference here. In 0.2.1 Elements of description across Works, Variants, Manifestations, and Items it is mentioned as a "temporal extent"
Excellent! I must have missed that detail, have updated the description above.
paulduchesne
commented
1 year ago Updates following discussion, has carrier renamed to has format due to intention of including a carrier tier, and removing ambiguity. projection characteristic renamed to image characteristic as some attributes (e.g. aperture) do not pertain to projection requirements.
Discussion around the modelling of Manifestation, using the FIAF Cataloguing Manual as primarily source.
Manifestation Elements
Define the
manifestationclass.2.2.1 Manifestation Type
Subclasses of
manifestationtaken verbatim from D.5 of the manual.2.3.1 Identifier 2.3.1.1 Identifier Type
At this level
identifieris less likely correspond to an external resource, rather internal archival ids.2.3.2 Title 2.3.2.1 Title Type
One area of complexity with
titleis that it is available for all tiers, but has subtly different vocabularies (see Manual pg 85) .2.4.1 Agent(s) 2.4.1.1 Agent Activity
2.4.1 highlights which agent activities are appropriate for each tier.
2.3.3 Language 2.3.3.1 Language Terms 2.3.3.2 Usage Type
This structure does feel slightly unintuitive -
manifestationconnected to type of language, and then the language itself, e.g.https://www.filmovyprehled.cz/cs/film/396690/sedmikrasky>hasLanguageUsage>blank node (of type "Spoken", subclass of "Language Usage")>hasLanguage>Czech.2.3.6 Notes
As with
work/variantall elements terminating in text blocks have been removed.2.4 Relationships
This is covered by the explicit relationship declarations (eg
hasAgent,hasEvent).2.4.2 Events D.4 contains a table of event types appropriate to each tier.
2.4.3 Other Relationships
2.4.4 Items
2.3.4 Format
2.3.4.1 Carrier Type 2.3.4.1.1 Carrier Type: General 2.3.4.1.2 Carrier Type: Specific
Carrier hierarchy can be drawn from vocabulary D.7.2
2.3.4.3 Sound Characteristics 2.3.4.3.1 Sound systems 2.3.4.3.2 Sound Channel Configuration
Manual suggests a vocabulary of sound characteristics under D.7.4.
2.3.4.4 Colour Characteristics
Manual suggests a vocabulary of colour characteristics under D.7.11.
2.3.5 Extent 2.3.5.1 Logical Extent 2.3.5.2 Physical Extent
extentcan be expressed with a blank node ofextent typefollowed by anextent value: e.g.Example Theatrical Manifestation>hasExtent>blank node (type Reels)>hasExtentValue>6.2.3.4.2 Projection Characteristic
Possible full implementation of this could be
Example Theatrical Manifestation>hasProjectionCharacteristic>blank node (type Aspect Ratio)>hasProjectionCharacteristicValue>1.75. As allaspectandaperture ratiosshould be numeric it might have been possible to represent as float, although a controlled vocabulary of entities is probably preferable (ie many numerical values are not legitimate ratios in this context).2.3.5.3 Duration 2.3.5.3.4 Duration Precision 2.3.5.3.5 Duration Type
Duration expressed as
extentwith type: e.g. seconds, minutes, hours