Jump to content

Dave Martin

Members
  • Content Count

    72
  • Joined

  • Last visited

  • Days Won

    4

Posts posted by Dave Martin


  1. SpeiVerta,

    OK, I think may best not to try to climb Everest as first mountain! :)

    CHI have provided a sample set of images - which definitely work. If there is any problem, still doing this, there is no point going on to your relatively-massive pictures.

    Please forgive me if these duplicate what you know or have tried, but this is the simplest set of steps, could you try these as they definitely work.

     

    Please try these and let us know if you successfully make and view a basic HSH RTI as below.

     

    Dave

     

    1. On your PC, if you don’t already have one, make a folder RTI at the top level on your C: drive, and within that, make a folder Fish_1000 so you end up with C:\RTI\Fish_1000
       

    2. Download the Fish_1000 set of example files from http://culturalheritageimaging.org/What_We_Offer/Downloads/Process/
      And then – somewhere else on your PC - expand the fish_fossil-data-set_1000.zip file. In the zip file, down one level, you will see a folder jpeg-exports      Copy that folder and its contents into C:\RTI\Fish_1000 so you end up with C:\RTI\Fish_1000\jpeg-exports which will contain 36 photographs named fish_fossil_01.jpg through fish_fossil_36.jpg
       

    3. Start RTI Builder

      In the Project Name box, type Fish_test
      In the Operation Sequence block, click on the first option – Highlight based (HSH Fitter)
      Then click the [Start new project] button
       

    4. On the Image selection screen which opens with an initial blank list, click the [Open folder] button at the right-hand edge of the screen

      In the Windows selection box which pops-up, it usually seems to default browsing to ‘My Documents’.
      Click on the ‘Computer’ icon (fourth one down) in the left-hand bar
      Double-click on C:
      Double-click on the RTI folder
      Double-click on the Fish_1000 folder
      Then click [Open]   (N.B. please don’t click on the ‘jpeg-exports’ folder)

      You should then see a pop-up ‘Loading images’ with a progress bar
      When all the images have loaded, you should then see 36 images in the selection pane
      after which you should be able to proceed with RTI processing; and as you proceed with building your RTI model, you should see an xml and other folders appear in the project folder, so eventually, if you just proceed (using default options and not cropping) you will have:
      C:\RTI\Fish_1000\assembly-files
      C:\RTI\Fish_1000\finished-files
      C:\RTI\Fish_1000\jpeg-exports
      C:\RTI\Fish_1000\Fish_test.xml
      and your finished RTI will be
      C:\RTI\Fish_1000\finished-files\ Fish_test_1000.rti
       

    5. You will then have an HSH RTI which you should try to open with the viewer C:\RTI\Fish_1000\finished-files\ Fish_test_1000.rti


  2. SpeiVerta,

    Apologies that I can only reply in English.

    I can see there is definitely a problem, the size - as read by the HSHfitter - is enormous. The height and width used by the HSH fitter should, with no re-size, be the same as the 'Actual size' (see attached screenshot).

    It may well be that your images are too big for that HSH fitter, there are other options if that is the case but some questions/suggestions that may help identify the problem:

    1) Did this work before for these exact images and just stop now?

    2) Is this the first time you tried to build an HSH RTI?

    3) Have you successfully made an HSH RTI for other images of your own?

    4) IF you made one OK before, if you go back to that set, does it still work?

    5) Have you tried making an HSH RTI with the sample set of fish-fossil images you can download from the CHI website?

    6) If you can make the fish-fossil OK, it may be your images or the memory demand are too big. How many images are you using? / what is their size in MB?

    7) If you only process, say, eight of your images, does the same happen?

    8) If you - just for testing - make a copy of your images with smaller resolution, say 1/4 size, does that work?

    9) Also, It is always useful to also report what operating system/version you are running?; also the source of your images?

    Dave

    Capture.JPG


  3. Mike,

    Just some more info to back your report up and maybe shed a little light for the developers:

    I've just tried this on my installation - with both CHI supplied test data and my own data.

    I had never tried for a PG set before, I get exactly the same as you. BUT if I highlight an RTI image set, it will allow me to export an RDF.

    Dave


  4. Hi Alessandro,

    Thanks for that information.

    By 'source' for the images I meant what camera and what type of file are you capturing (ARW, RAW, DNG, TIF etc.) and resolution; and any processing you're doing before attempting to make the RTI PTM.

    If you are able to share, if you could upload two folders (i) the files directly from the camera, and (ii) whatever you're trying process, to DropBox or similar, then I'd be glad to have a look and try to build PTM on my own workstations (will be a few days before I would have time though).

    Regards
    Dave


  5. (This thread would really be better in the 'RTI Processing' section, maybe one of the Admins could move it?)

    Hi Alessandro,


    A couple of questions, then a suggestion.

    1) Have you ever built successfully using PTM?
    2) Can you advise what operating system and version you're running?
    3) Can you advise size and source of your images?

    To identify if it is a problem with your images or with the installation on your computer, could I suggest you try using one of the sample image sets which CHI provide, as they are known to work.

     

    Please forgive me if these duplicate what you know or have tried, but this is the simplest set of steps, could you try these as they definitely work (below are written for a Windows operating system).
    Dave

     

    ===

    1. On your PC, if you don’t already have one, make a folder RTI at the top level on your C: drive, and within that, make a folder Fish_1000 so you end up with C:\RTI\Fish_1000
       

    2. Download the Fish_1000 set of example files from http://culturalheritageimaging.org/What_We_Offer/Downloads/Process/
      And then – somewhere else on your PC - expand the fish_fossil-data-set_1000.zip file. In the zip file, down one level, you will see a folder jpeg-exports      Copy that folder and its contents into C:\RTI\Fish_1000 so you end up with C:\RTI\Fish_1000\jpeg-exports which will contain 36 photographs named fish_fossil_01.jpg through fish_fossil_36.jpg

       

    3. Start RTI Builder

      In the Project Name box, type Fish_1000_PTM
      In the Operation Sequence block, click on the first option – Highlight based (PTM Fitter)
      Then click the [Start new project] button

       

    4. On the Image selection screen which opens with an initial blank list, click the [Open folder] button at the right-hand edge of the screen

      In the Windows selection box which pops-up, it usually seems to default browsing to ‘My Documents’.
      Click on the ‘Computer’ icon (fourth one down) in the left-hand bar
      Double-click on C:
      Double-click on the RTI folder
      Double-click on the Fish_1000 folder
      Then click [Open]   (N.B. please don’t click on the ‘jpeg-exports’ folder)

      You should then see a pop-up ‘Loading images’ with a progress bar
      When all the images have loaded, you should then see 36 images in the selection pane
      after which you should be able to proceed with RTI processing; and as you proceed with building your RTI model, you should see an xml and other folders appear in the project folder, so eventually, if you just proceed (using default options and not cropping) you will have:
      C:\RTI\Fish_1000\assembly-files
      C:\RTI\Fish_1000\finished-files
      C:\RTI\Fish_1000\jpeg-exports
      C:\RTI\Fish_1000\Fish_1000_PTM.xml
      and your finished model will be
      C:\RTI\Fish_1000\finished-files\ Fish_1000_PTM_1000.ptm

       

     


  6. Re Beta 1.0.3 (DB version 2.2.3) Build Apr 26 2018 15:55:31

     

    1. Fully appreciate that the priority in the first beta was to expose all the functionality, but structure of the Dashboard might benefit from reorganisation - order? titles? move ‘paramaterisation’ items to a ‘Setup’ menu?
       

    2. System tray icon – when the focus is not on DLN:CC, the icon is just a blank white sheet – suggest something more meaningful would be useful – maybe something like the blue notebook DLN logo used on the web pages? or the CHI logo?
       

    3. Capture Teams – there are traces of ‘Teams’ (e.g. in export) but no UI access that I can spot anywhere in the dashboard. Suspect that Teams were at one point in there (something like equipment subassemblies) but have since been dropped, with which I personally would agree. Needs either maintenance and ability to use, or remove Teams from DLN:CC ?


  7. Re Beta 1.0.3 (DB version 2.2.3) Build Apr 26 2018 15:55:31

     

    1. When creating or modifying entries on the reference lists, there doesn’t appear to be any check on whether a duplicate is being created.

      DLN:CC appears to (universally as far as I can see) allow the user to create duplicate entries. In the case of, say, Operator Roles, one can create two roles both called Supervisor. When one comes to associate Operators with an Image Set, you have to pick a role for the association, and you will see two indistinguishable Supervisor roles to choose from, and even if the tooltip / description had been populated, it isn’t visible.

      In some cases there may legitimately be multiple items such as tripod or light stand, but that would be covered by having Quantity > 1 in the equipment record as there would be no need to differentiate between them.

      I can envisage relatively few cases where near-duplicate entries might be required – but they would still need visually distinct names to allow them to be correctly associated. For example, if a user had two camera bodies of identical model, but different serial numbers – and the user wished to keep track of which body was used on a shoot because of the age of the body, or perhaps different firmware. Two equipment records could be created – but unless there was something distinctive about the names, the user would not know which body to associate with the image capture.

      It should be relatively easy for the alert user to monitor this on small lists (especially once they are sorted alphabetically) but once lists get longer, or perhaps a DLN:CC instance has multiple users, the risk of inadvertent creation of duplicate entries will increase.

      I would strongly suggest that DLN:CC shouldn’t allow creation of any records with identical names.


       

    2. List ‘shuffles’ – a disconcerting phenomenon – only seen, I think, on the Equipment Overview list.

      If you open the Equipment Overview and then highlight a row and click [Details], or double-click a row and then [Close] the details screen, the underlying list remains exactly unchanged.

      However, if you open the details and then change something (not necessarily the Name) and then click [Save], when you then [Close] the details screen, the entry you clicked on has, at first sight, disappeared!

      On further inspection though, it has either moved to display row 9 or the bottom of the list.

      This is somewhat disconcerting!

      If the operator has manually sorted the equipment list before activating & changing details, then the list seems to survive un-molested in the chosen order (unless the list was alphabetical and the name has changed in which case the resultant change to maintain alphabetical order is performed correctly).

     


  8. Re Beta 1.0.3 (DB version 2.2.3) Build Apr 26 2018 15:55:31

     

    On the Image Set screens, as well as the ‘Created’ dates, it also exposes two date fields captioned ‘Data entry some time within’. At first, as these are only (as far as I can see) on image sets, I wondered if ‘Created’ was the capture date and ‘Data entry’ was the date on which the images were processed to build an RTI model or photogrammetric product.

     

    However, the ‘data entry’ fields are read-only, and appear to be some form of internal audit record which has both start and end date auto-populated with the date on which the record was keyed in DLN:CC?

     

    Suggest:

    1. ‘Created’ date might be better captioned ‘Capture’ date to make it clear it relates to the date the images were captured, not when the DLN record was created

    2. Audit fields not displayed elsewhere, so not needed here?

    3. It would, though, be useful to allow recording of date on which image sets were processed?

     

    (also, previous comments / discussions on date formats apply)


  9. Re Beta 1.0.3 (DB version 2.2.3) Build Apr 26 2018 15:55:31

     

    1. On many list displays (e.g. Category/Type/Role) the first column of the tabular display, to the left of the ‘ID’ field, is a monotonically ascending integer which I suspect is a display index of some kind, which could be useful for debugging but can’t envisage functionality for normal users, and similarity in values with those in the ID column could potentially be confusing. Is this un-named column necessary? / should it be hidden or zero-width?
       

    2. Default sort order – appreciate that lists can be sorted by clicking on the column-head label, but would suggest that a consistent approach is taken to the order in which items are sorted for display when a screen first opens. This is currently not the case, for example:

    • Equipment Categories defaults to alphabetically by name

    • Person Overview appears to default to ascending ID

    • Equipment Overview appears random – neither alphabetical or in ID order

    • The list of Role options when associating an Operator with an Image Set -  neither alphabetical or in ID order

                I would suggest that alphabetically A..Z for the ‘name’ of the item would be a reasonable and useful default.


  10. Hi Carla,

     

    Thanks very much for all the info, especially on Rights, and on Image Bundles.

     

    I’m glad to contribute a little bit back to the work CHI does; and from experience developing and leading development teams, and owning projects for deployment to users across all inhabited continents, I know how when a beta is exposed for scrutiny just merely having different pairs of eyes can raise questions, some of which are valid, others may be down to misunderstanding or differing expectation!

     

    Cheers

    Dave


  11. Re Beta 1.0.3 (DB version 2.2.3) Build Apr 26 2018 15:55:31

     

    This is initially relates to silent corruption of a field in the Equipment Overview, but may have wider implications…

     

    If one is using DLN:CC Equipment Overview as a mini asset register as per the sample database, and populates a price, there are silent and insidious problems if one enters anything other than digits 0 to 9 or a decimal point by including say  a currency symbol (e.g. £ for GBP) or a comma (used by many to delimit thousands, or in some regions as the separator twixt the integer portion and decimal fraction of a real number).

     

    What happens is:

    On the equipment overview details screen, the data entry field (whether adding a new record or editing an existing record) allows one to enter costs such as

    123.45

    1234.56

    1,234.56

    1.234,56

    £123.45

    £1,234.56

    GBP 123.45

    without any objection either when the Price field is exited or when the [Save] button is clicked.

     

    However, if one then saves, closes and goes back to view the record details, in all except the first two cases, the Price is now ‘0’.

     

    Without seeing the code, from the above it appears the failure is because the UI is accepting (meaningful and valid) ‘text’ but the database is only accepting a real with a decimal point separator. Insidiously, the whole database INSERT / UPDATE isn’t failing – if for example the serial number and price are changed, the serial number is saved/updated but the price can be corrupted to 0 if it contains anything except 0 to 9 or a decimal point.

     

    Couple of points:

    1. As there is unlikely to be any arithmetic on equipment prices, in this specific case might be best to just make the Price column in the database a text varchar, rather than numeric, column.

    More importantly though:

          2.   If there are restrictions as to what can be stored in a field, the UI needs up-front validation either when the field is exited or when the [Save] button is clicked.

          3.  IF the UI is posting the input from the user intact and it is being rejected by the database, then it raises the possibility that the result of commands which modify the database may not be being monitored.

     

    Dave


  12. Re Beta 1.0.3 (DB version 2.2.3) Build Apr 26 2018 15:55:31

     

    A tick-box ‘Show associated’ appears in the Filter block on a number of screens. I fully understand this is a beta; I can understand what ‘Show associated’ does, but I’m not sure if it is needed (as much / at all), and if it is needed, its placing on the screen and its nomenclature.

     

    The data in DLN builds the database by associating, say, equipment, operators and documents with a project or image set. In pretty much every case, as it is a normalised database, one operator or document can be associated with multiple projects.

     

    What is in question here is the view ‘from the project’ side – and from the project perspective, the question of how many, and how, the higher-level objects are associated with, say, a project.

     

    ‘Show associated’ appears when the operator is given the opportunity to link higher-level objects with, say, a project.

     

    In the current beta, the left-hand pane displays a list of items that can be associated, and the right-hand pane displays a list of those have (already) been associated. Items can be added to and removed the associated list, and the corresponding entry appears/disappears in both right and left hand panes; the list on the left shows those that have not already been associated.

     

    The current system effectively operates a default mechanism to allow one of each type of entry to be associated with the project; ticking the ‘Show associated’ box populates the left-hand list with all possibilities, not just those which have already been associated.

     

    First two points:

     

    1. The ‘Show associated’ tick box is in the Filter pane at the top of the display, which otherwise filters the right-hand list – it would (if/when still needed) be more intuitive if ‘Show associated’ was above the left-hand pane which it influences, alongside the title of the left-hand list.

    2. The tick box’s name ‘Show associated’ would be more meaningful as ‘Show already associated’ (if/when still needed).

     

    Looking deeper though, I would question the need for ‘Show associated’ in many cases. Taking a few examples:

    • Linking a Document to a Project – this, when established, is a direct, unqualified link. There would be no sense in linking the same document to the same project more than once, so having ‘Show associated’ on that screen is not needed. In this case, ‘Show associated’ isn’t needed.

    • Linking Equipment to an Image Set - this, when established, is a direct, unqualified link. In this case though it is possible that multiples of the same item are used. The equipment list will probably be significantly ‘longer’ and only ‘shrink’ a little when an item is associated,  so having ‘Show associated’ on that screen is of minimal benefit. If helpful validation was to be offered, then when an item is being associated, quickly check if it has already been associated and pop-up a quick message to ask the operator to confirm they wish to associate a second instance.

    • Linking a Stakeholder to a Project – this establishes a qualified link as a role must be selected at the time the stakeholder is linked to the project. The default only allows a Stakeholder to be associated once; if a second role needs to be documented, the ‘Show associated’ needs to be ticked so that the stakeholder can be selected a second time and a second association recorded. However, there are no checks that a pre-existing role isn’t being duplicated, so by ticking ‘Show associated’ it is possible to accidentally assign a meaningless duplicate role. In this case, some safeguard is needed but ‘Show associated’ isn’t really the most appropriate – I would suggest the Stakeholder list is always left fully populated (i.e. no ‘Show associated’), and when an association is being created the pop-up linkage details screen either automatically limits the list of roles to those ‘… NOT IN …’ the existing, or quickly check before saving that it isn’t a duplication.

     

    So, 
     

           3.  I would query the appropriateness or need for ‘Show associated’ as the data model is refined.

     

    Just to repeat, these are suggestions which could only be made after the valuable first beta; and I fully appreciate workflow and data model will evolve. ‘Show associated’ allowed testing to progress and these comments to be made!

     

    Dave


  13. Re Beta 1.0.3 (DB version 2.2.3) Build Apr 26 2018 15:55:31

     

    In addition to the technique-specific fields on image sets, currently:

    • there is an unformatted free-form ‘Notes’ or ‘Description’ field/area on many of the register screens;

    • the Equipment Overview has a number of hard-coded auxiliary fields, which are enabled in specific cases if the equipment is a lens / lens filter, or illumination source / illumination filter; and

    • the Subject Overview screen has an option to populate four free format label:value pairs, one of which whose label indicates it is intended to be ‘is part of’.

     

    The auxiliary fields on the Equipment records have fixed purpose and are stored in hard-coded columns.

     

    However, the auxiliary fields on the Subject are to all effects uncontrolled. As it stands, the user, on a per-subject basis, decides on the purpose they wish to use the field for that single subject and  types in what they want the field name / item label to be and then types in the value to be stored – both of these are totally free format.

     

    What this means is that these auxiliary fields are, I would argue, of possibly even less value than the free-format Notes field – whereas the auxiliary fields could actually be so much more useful….

     

    With the current system, there are two issues:

    1. There is no control over the labels, so on one record the user could type in “Gift of” as the title or label for the first field (as in the sample data), and on another subject, they could type in the first field’s label as “Donated by” or “Donor”.
      If the user then sought to search for all records where label1 = “Donor” … then they would miss ones where the label had been typed as “Gift of” or “Donated by”.
      (as it happens, I suspect “Gift of” as cited in the example database might actually be better as a stakeholder role, but the “Gift of” serves well to illustrate the point.)

    2. Also, as there is no control over which of the fields is used, even if there was standardisation to call an entry, say, “Catalogue number”, as that could be stored in subject auxiliary field 1, 2 or 3, then any query would have to be some form of (… WHERE subject_id1_lbl = ‘Catalogue number’) UNION  (… WHERE subject_id2_lbl = ‘Catalogue number’) UNION ( … WHERE subject_id3_lbl = ‘Catalogue number’)


    As it stands, having the free-format labels and data fields allow highlighting (a bit like having the ability to underline or embolden text in the Notes field) but doesn’t contribute as much as it could in terms of structured data.

     

    I think that a simple change could resolve these, and add to the power of DLN. Instead of having multiple auxiliary fields embedded in the Subject rows, store the auxiliary data in a linked table, and control the list of possible labels via a ‘label type’ register?


  14. Re Beta 1.0.3 (DB version 2.2.3) Build Apr 26 2018 15:55:31

     

    1. Location entries need a ‘Notes’ section like most other registers

    2. Whilst some well-known institutions and monuments may have entries in the Getty Thesaurus of Geographic Names, possibly the majority of locations will not appear therein, or at least with insufficient granularity. Need field(s) in which the address can be entered.

    3. As mentioned by IainS earlier, the ‘GPS coordinates’ field needs provision to record information on projection etc.

    4. Need a field to use to store local coordinates and type thereof

    5. Suggest allow recording of EPSG code for sets of co-ordinates, and also ability to have local name for a co-ordinate system (e.g. “OSGB36(15)”)


  15. Re Beta 1.0.3 (DB version 2.2.3) Build Apr 26 2018 15:55:31

     

    Other UI comments / across multiple registers:

     

    1. Deleting, for example, equipment entries – there is no check for dependencies so equipment which is associated with an image set can be deleted / referential integrity is not enforced.

    2. Would expect [Ctrl][F4] to activate [Close] as it does for almost every other Windows application

    3. Would expect [Esc] to activate [Cancel]/[Close]

    4. Some maintenance screens have an explicit [Close] button, others don’t

    5. To see / update details, some maintenance screens have button [Details], others have one called [Edit]

    6. Even without the use of accelerator keys, would suggest not ideal having three visually-similar buttons adjacent to each other [Details] [Duplicate] [Delete]

    7. Associating e.g. equipment with image set – currently works by highlight then click [ >> ] – it would be good if it also responded to the commonly used double-click to move.

    8. Where there is a sub-assembly, this allows filtering of the left-hand list. Would be good to be able to just associate the entries on the list.

    9. Filter block - most maintenance screens just have a [Clear] but some have an explicit [Filter] button?

    10. Filter block - several maintenance screens have [Export RDF] in the Filter block?

    11. File / About - more usually under Help / About?

    12. File / About - make the text on the screen so it can be copied and pasted into error reports etc. to ease reporting / accuracy

     

×
×
  • Create New...