Jump to content

Dave Martin

  • Content Count

  • Joined

  • Last visited

  • Days Won


Everything posted by Dave Martin

  1. Niki, If it is a resource issue due to the size of the images, one way to identify if it is absolute image size that is causing a problem, or just the total demand on the system, might be to try processing a reduced number of your hi-res images. I don't know how many you have in your full set - I have 50-100 typically - but if you maybe tried with maybe just ten images? - the RTI obviously wouldn't be as good but it might help to diagnose the problem? I'm sorry that, as just another user, I can only use my RTI and computer troubleshooting experience, I haven't any access to the innards of the programs! Dave
  2. Nikki, another tiny thought in case it helps shed light on the problem. You might just see something if you have memory monitoring displayed when it crashes on your big images? and there might also be something written into the Windows logs? Dave
  3. Thanks Nikki, sounds as if moving (slowly!) towards possible indicators. I'm glad it works for at least sample images so software setup looks OK. I don't know about any absolute image size limits, I just did a quick search here and found this thread which does describe processing large images 10328x7760px, albeit I think on a Mac. I'm sorry that I can't help much more at present - I'm not in front of my (PC based) RTI workstation, I think it may be a case of waiting until Carla and team at CHI are on-air (San Francisco time). mfG Dave
  4. Also Nikki, another thought - if you download one of the sample image sets from the CHI website, which are known to work OK, do they work on this PC? Dave
  5. Nikki, Firstly, now that error message is visible, there is a thread here on this forum about exactly that message which might give you some clues Secondly, have you tried on another PC? Dave
  6. Nikki, path looks OK as far as I can tell. Couple of questions: 1) you mention it fails after cropping, does it fail if you don't crop? 2) you mention you're trying PTM, you have downloaded and installed the PTM Fitter, haven't you? (page 10 of the processing manual) 3) if you try HSH rather than PTM, any change? Dave
  7. Niki - can you post a screenshot showing the file path?
  8. Jack, I think this is an important question to consider, and the answer may depend on the purpose and the longevity expected. You mention Archive and Display, and I think there are useful parallels in terms of images, especially as we may have one format for long-term assured storage and another for speedier manipulation and display. You mention DNG, and DNG can contain TIFFs. Top-level archives which accept images will often specify they must be un-compressed TIFFs. This seems counter-intuitive in that they are taking on storage obligations which could be significantly reduced, but they are protecting against future licensing changes. The TIFF standard has no proprietary issues, but some of the algorithms behind compression standards, whilst open to read and implement at present, are private property, and there is the danger that permission may be revoked. Whilst not implying any concerns over or criticism of Khronos, it states at the top of https://www.khronos.org/gltf/ “glTF™ (GL Transmission Format) is a royalty-free specification” – in other words, it looks to be remaining Khronos’s property. There are precedents to be a little concerned, especially for the long-term. The Graphics Interchange Format (‘GIF’) is now thought of by most as a container for handy little images or animations etc. The GIF was developed by the Compuserve BBS team, and internally the data is losslessly compressed using LZW. LZW is proprietary, and from what I recall, there were attempts by those who owned them to assert their rights the LZW algorithm, and this led to the development of the open PNG standard. I don’t have an answer in the ‘model archiving’ space but will be interested to learn. Dave
  9. Hi P fried, I don't know if there is anything off-the-shelf yet, but a couple of articles on computer-vision jigsaw solving: https://www.i-programmer.info/news/181-algorithms/4380-faster-jigsaw-solving.html https://www.researchgate.net/publication/220843444_An_Innovative_Algorithm_for_Solving_Jigsaw_Puzzles_Using_Geometrical_and_Color_Features Another related area which is receiving attention is the DARPA Shredder Challenge - reconstructing documents from images of shredded strips or fragments. Let us know how you get on! cheers/Dave
  10. P Fried, I don't know how well it would translate to such a context, but there has been work in the computer vision field trying, and succeeding, in solving jigsaws, so there might be something there. Suspect they're geared around 2D, but might be possible to use them to produced an 'unwrapped' solution? Dave
  11. Katie - have you downloaded the PTM from the Hewlett Packard research site?
  12. Crystal, I'm not in front of my PhotoScan workstation, but a couple of quick thoughts: There are several options for masks, one of which is a separate mask for each image, in which case all you do is select the folder containing the masks - and the symptoms sound like possibly that's the route you're currently going down. I'm not sure if you have visited it already, but - especially as the issue isn't particularly subject-matter specific - I'm sure you would find it worthwhile to visit the dedicated PhotoScan forum operated by Agisoft themselves, where there's support from Agisoft staff and fellow users http://www.agisoft.com/forum/index.php?board=7.0 Also, on the Agisoft forum there is info on new masking methods (single image, volumetric...) which have been introduced in recent versions of PhotoScan. Dave
  13. Thanks Carla, yes understand can create a grouping for the dome, and apologies if I didn't make question clear. The point was that the dome structure is (relatively) invariant, so rather than have to associate pictures of the dome or rig with every image set taken with the dome, the 'normalised' solution would be to associate the pictures with the equipment. Similar might apply to, say, a photogrammetry turntable arrangement which was used repeatedly. Dave
  14. Thanks Erich, Have just tried that and, exactly as you say, ticking '[ ] Show associated?' does allow an operator to be added with another role. I would observe/suggest: 1) Not sure if it might be better either completely without the need to tick that box, or reverse the default / pre-tick the box. It certainly isn't obvious (until you know how!) 2) As you say, once that box is ticked, not only can you associate multiple roles, but you can add several instances of the same actor/operator with exactly the same role. Would strongly suggest that is not a valid result, so basic validation when associating an operator with an image set would be to present only those roles which they don't have already - so adding a completely new operator would in effect offer all roles; adding an operator for a second or subsequent time would offer a reduced set of roles (NOT IN ...) Dave
  15. DLN:CC Beta – software version 1.0.3 (build 26 Apr 2018 15:55:31) / DB version 2.2.3 First few observations / comments: Needs Regionalisation, in particular dates – looking at examples logged today (4 June 2018) it looks like 6 April 2018 (6-4-2018). If the system can’t/won’t reference the users’ locale, or maybe even as a ‘better’ fix, suggest that the date format, when displayed at least, is dd-mmm-yyyy so it shows 7-Feb-2018 etc which is then un-ambiguous world-wide. Registering an RTI image set has various string (Egyptian, not computer!)-related prompts. Might I suggest there’s a tick-box/radio-button or similar to indicate string or dome? Talking of domes, I can see how one would create an entry for a dome in the equipment register. Domes are rarely off-the-shelf and thinking of how one could/would document a dome, I wonder if the ability to associate photographs with project setups could usefully be extended to allow one or more photos of a piece of equipment (e.g. a dome type rig) to be associated with that equipment entry to save having to re-associate them with every group or set of images acquired using the same rig. In documenting an Image Set, it allows Operators to be associated, and to specify their role. However, it only allows one role per operator. In real life there may well be multiple roles for an actor/operator (e.g. image acquisition, computer processing). Could ‘fudge’ this by creating hybrid roles (e.g. 'photographer / computer operator'), or having schizophrenic clones of the same operator (‘Dave Martin – photographer’, ‘Dave Martin – computer operator’) but that is de-normalised and doesn’t reflect reality; so multiple roles are needed. Dave
  16. Trying to use DLN:Inspector with a DLN database (the sample database which installs with DLN:CC). The 'Project' at the top of the DLN:Inspector GUI does allow you to select a project; but the drop-down 'Group' to select image groups drop-down/combo is always empty so cannot get any further. Same if run DLN:Inspector RTI or PG. The DLN database has both the LACMA tests (RTI and PG) and some of my own initial tests in as well. None of them are accessible from DLN:Inspector (also, unlike DLN:CC, there's no 'About' which allows one to report more detailed version/build info, only the 'Version 1.0 Beta' in the title bar of the application)
  17. UPDATE: Have found a short-term work-around. I guess that there is a password embedded in the DLN applications, and if DLN:CC starts it must have the correct password. I noticed that on the DLN:CC Admin menu there was a 'Run script' option, so I first of all tried running the 'createpg.sql' script from DLN:CC - the script started OK, but then failed because the three sub-scripts that created the sequences, tables and then inserted the ref data only had relative paths when called from within 'createpg.sql'. So, I was going to modify a copy of 'createpg.sql' to embed the paths, but instead from within DLN:CC which was logged in to the database, I just ran the 'createsequence.sql', then 'createtable.sql' and finally 'insertlookupdata.sql' - all of which ran OK to completion. I can now start DLN:Inspector and connect to the DLN database. Dave
  18. Hi Carla, Have tried to install DLN:Inspector. Once I found out how to get DLN:CC to start (separate thread in DLN:CC section) I tried to start DLN:Inspector again and it came up with the first message shown below that the database had not been update for DLN:Inspector and to see the documentation for instructions. Checked online and in the installed program groups and could not spot the documentation. In the end I started DLN:Inspector 'without the DLN' and then, by clicking [Help] it opened the PDF. I then followed the instructions (which I guess are OK for beta testing but possibly not ideal for end-users long term) but, when the update is run, it fails with the message shown in the second screenshot. Dave
  19. UPDATE: - I had a look at/for the (not) installed database, then repeated the installation, and I've got it working now. I think problem when application was trying to connect to the database was not so much 'no password' as in fact 'no database'! The installer's default option under 'DLNCC Database' is 'Upgrade existing database'. If there is no database to install, it silently skips that stage. For first install, manually changing the option to 'New database, replacing the existing' seems to rectify the situation. This also removes the complaint when DLN:Inspector starts) Iain - suggest you try re-installing and tick the 'New database' option and see if that helps. I think / would suggest: the installer script needs to be 'cleverer' so if no database installed it will install skeleton one, or at least prompt the user; or at least the database option text in the installer screen above needs extra (like the PostgresSQL) so it maybe says 'New database, replacing the existing (check if first installation)' or at least some pre-install info/advice for the user . Dave
  20. Carla - similar (I think) question to IainS. When starting DLN:CC, it attempts to connect to the database - it pops up a database login (first pic), username defaults to 'postgres' with an option of two databases to connect to ('cpt_db' and 'cpt_db_v2.1'). Default password is blank, but there must be a password set as can't get beyond this point. Edit: I guess because DLN:CC hasn't got beyond this stage, DLN:Inspector complains at startup as well (second pic) Dave
  21. Thanks Carla, I had suspected that ability was not in there yet. Is there a suggestion list anywhere for possible features in the RTI Viewer? - if so, I'd like to add this feature to that list. Dave
  22. Is there any way to specify in a model that a particular bookmark should be activated when the viewer opens the model? The 'high noon' default can be less than ideal for first view, especially for a novice viewer. Rather than trying to coach the user to go down and open specific a bookmark, it would be great if one could specify, rather than 'NONE', a bookmark (with all its power) which would be activated as soon as the model is loaded (broadly similar to the way one can set the 'initial view' for a PDF). Dave
  • Create New...