3D Archiving Houses and Settlements Alternative Technologies to support Lasting Diachronic Memory KOBAYASHI, Hideyuki (DR. Eng.)* Abstract In order to preserve 3D digital data describing houses, we propose alternative technologies to keep the readability of the data against ever changing data formats, and lasting recording media, through usage of ‘virtual converter’, and ‘electronic record’ attached to each building, as follows: Background and purpose 1. Cases of organizations in charge of maintaining lasting data 2. Historical review of media recording digital data 3. Currently used data formats for modeling 3D objects 4. Development of alternative technologies 5. Next steps Conclusion Keywords: diachronic archiving, 3D data format, recording media, houses and settlements *Research Coordinator for Housing Information System, Research Center for Advanced Information Technology, National Institute for Land and Infrastructure Management (NILIM), Ministry of Land, Infrastructure, Transport and Tourism (MLIT), JAPAN E-Mail: kobayashi-h92qa@nilim.go.jp Background and purpose In Japan, housing stock already far exceeds the total number of household, therefore, maintenance of existing houses and buildings became more important, rather than massive construction and provision. (million units) Total house unit (left index) Vacant house units (left index) Vacant rate (right index) 1958 1963 1968 1973 1978 1983 1988 1993 1998 2003 2008 Figure-1 Trend of total house units, vacant units and vacant rate in Japan 1958-20081) According to the national census on housing and land 2008, 49.99 million households were counted, while total housing stock was 57.59 million units. 7.56million units were identified as ‘empty house’ (fig 1). Also, the average length of life of houses is extending. There are some different methods for measuring ‘lifetime’, but most precise analysis was performed in Osaka, which showed the ever increasing trend of it. Average service time (Years) Statistic Forecast Chuoh Ward Higashi-Yodogawa Ward Hirakata City , in Osaka Prefecture Anno Domino Figure-2 Trend of average lifetime of timber houses in 2 wards and 1 city in Osaka2) That means, existing houses are expected to serve for generations of families through re-use, or different families through resale of long-life house as stock. This situation is quite different from that after 1945 until 1980s, where new construction and/or ‘scrap and build’ type reconstruction of short-life houses were dominant, which served only for one generation of ‘core families’. In order to cope with this new situation, the governmental housing policy pays more attention not only to the quality (length of life) of houses, but also provision of reliable data/records of houses attributed to physical buildings will assist the alteration and transactions of in future. Record of houses and settlements are also valuable in case of disasters. From the technical viewpoint, instead of conventional blue prints and written documents on paper, digital data are becoming popular for synchronic communication. Usefulness of them will further better in future. However, the coded data need media to record and replay, and durability of current digital recording media is still insufficient to cover the whole lifetime of houses. Also, data formats tend to change frequently, making the diachronic transfer of data between ages suspicious. 2D images like drawings and photos could be at least printed on long-life papers and stored. However, coded 3D data cannot be printed on papers as objects of description are seen. They need drives and decoders to be replayed through computer graphics or three dimensional printers etc. This nature of 3D data is common to such coded data as sounds and movies. Therefore, we need durable recording devices, and method for assure the ‘readability’ of various and changing formats of data for the length of time which is, at least, beyond the length of life of physical buildings. In order to strengthen technical basis for to cope with this new demand, ‘Basic Research on Technologies for Permanent Archiving of 3D Data of Houses’ was performed by National Institute for Land and Infrastructure Management (NILIM, in Tsukuba-science city) between fiscal years 2012-2014. This paper reports some alternative technologies we developed, toward the lasting use of 3D data archiving houses and settlements. 1. Cases of organizations in charge of maintaining lasting data Several representative organizations in Japan that are in charge of keeping important and massive digital data were surveyed and technologies, methods and media which they use today for storing ever increasing data were monitored. Many of them are keeping data in own data centers, and carefully considering the out-sourcing of the activities, paying attention to the security of the data. (1) Hokkaido Building Guidance Center The center provides charged service for filing and preserving data of individual houses since 2005, earliest among local authorities in Japan. The houses registered should meet the standard for cold region, which is stricter than the minimum conditions required by the national building standard law. The data can contain digital files (CAD data, image data of drawings, photos etc.) up to 100MB. The data is saved in the data center (out-sourcing to the private firm), and also 3 pieces of CD-R are provided. These three disks are separately filed at the center, at the contractor who built the house and at the owner of the house. The center selects a qualified product of CD-R that is expected to last for more than ten years. When the ten years have passed, the center sends mails to the contractors and house owners to refresh the media. They say that this renewal procedure is also a good chance to make them recall the existence of such archives. The data center is located in the industrial complex near to Sapporo city, where average temperature is rather low, which is good condition for data center. The data center is also considering the usage of waste heat from the hard discs for regional air conditioning. (2) National Diet Library This organization keeps more than 38 million documents. Most of them are made of paper, however, 670 thousands of recorded media and 8.9 million of documents in the form of micro films are also filed. Besides preservation of original documents, they are elaborating to create digital archives for the purpose of disclosure, without physical damage to the original documents. Sometimes books disclosed are attached with digital data in the form of diskette or CD-ROM. For the purpose of reading, the library keeps machines and drives to read these legacy media. (3) National Archives of Japan Historical documents of the Japanese government are filed in this organization. Important documents are scanned and disclosed in the form of digital archive, accessible through internet. Annex storage is located in Tsukuba, c.a. 50 km from the main storage in Tokyo. (4) Statistical Bureau, Ministry of Internal Affairs National censuses are performed by this organization, and digital master data are coded and filed in their own data center. (5) Geography Survey of Japan This institute is in charge of geographical survey, especially underground soil conditions and mineral resources, and today under AIST, located in Tsukuba-city. (6) Geospatial Information Authority of Japan This institute is in charge of geographical survey and publishing various maps. Digital maps (DM) are also available. In 2011, the damaged area of the great eastern Japan earthquake was intensively surveyed and precise maps of changed land shape, damaged settlements were issued. (7) Remote Sensing Technology Center (RESTEC) Satellite image data are collected, kept and delivered from this institute. (8) Building statistic Building certification procedure is performed in Japan under control of Ministry of Land, Infrastructure, Transport and Tourism(MLIT), and newly started construction works are tabulated and issued from the Ministry of Land, Infrastructure, Transport and Tourism. The master data is kept in a governmental data center and also in the form of MO (see 2(4)). (9) Registered copyright of software/programs Copyrights are registered in the Agency for Cultural Affairs, Especially computer programs are registered in Software Information Center (SOFTIC). They had filed the registered works in the form of micro films for lasting memory, however, started to receive digital data since June 2011. Up to march 2013, 11,557 programs have been registered. (10) Public buildings We monitored data on public buildings provided and kept by the local governments. In case of Asahikawa city, CAD data of public offices, schools and rental houses are kept in a file server maintained in the city hall. (11) Technical office of MLIT (SXF) Design drawings and completion drawing of national public works are recorded in a form of standardized 2D data format named SXF. CAD data of infrastructure provided by consulting firms are converted in the common SXF format and kept in the local technical offices of the MLIT in the form of file server and CD-R, and the data is stored in the local technical offices annex to the 8 provincial branch of the MLIT. They are summarized into GIS and utilized for operation and maintenance works. This GIS data is managed by the NILIM. (12) Japan Patent Office Industrial Property Digital Library (IPDL) keeps digital data of patent and provides searching services. 2. Historical review of media recording digital data Vulnerability and risks of historical digital records are generally warned since 1990s.3) From this viewpoint, we traced various memory devices which have been used to record digital data (statistics, CAD, GIS etc.) in Building Research Institute (BRI since 1946) and National Institute for Land and Infrastructure Institute (NILIM, established in 2001, partially inherited those old recording media from BRI). They including MT, CMT, FD, MO, DAT, HDD, CD-ROM, CF, SD etc. and we evaluated from viewpoint of durability. In general, devices that keep the data in the form of electrons on silicon chips have the shortest life, followed by magnetic devices. The devices that record the data in the form of physical shape (e.g. punched papers and pressed CD-ROM, not CD-R) can last longest. The durability of listed digital devices promised by the makers has not exceeded 100 years. Beside the deterioration of recording media itself, the social life of the ‘drives’ will be also problematic. Some of the media still correctly keep the recorded data; however, it has become difficult to find the ‘drive’ to read out the data from the media, for example ‘5 inch MO disc’. (1) Paper media for recording program/data In 1970s, holes on paper ribbon or card was used to record program/data. Some of them still remain in our institute for a kind of memory (Figure-3, 4). Through observation, the data recorded on them are still readable through eyes or scanners. However, machines to read them had disappeared (card reader, tape reader, etc.). Figure-3 One piece of punched card Figure-4 One series of punched cards containig statistical data Also, some data printed on papers by ‘line printer’ in 1970s still remains (Figure-5). They seem to be still readable through OCR. Figure-5 Statistics output by line printers, blueprinted (2) Magnetic media for recording program/data Open-reel type magnetic tape was a media to keep statistic data in 1960-70s. Large tape drive was equipped in main-frame computer rooms. Magnetic disc was also attached to mini-computers in ‘80s. In those days, open-reel type magnetic tape was still used to record sounds, however the specification was quite different and the tape decks for musical tape are not useful to the tapes containing digital data. In 1970s, compact cassette was introduced to record speech and music. Some of initial micro-computers utilized this for recording and replaying digital data through ‘MODEM’ interface, which was also used to exchange data through telephone. Digital compact cassette was used to record digital data for automation in factories. Tape players for compact cassettes are still produced and sold in the market. DAT, which was once popular for digital recording of audio, was also utilized as back-up media for hard disc in 1990s. Today they are already obsolete. Floppy discs that enable random access between tracks and sectors throughout a disc unit were introduced in BRI (one of previous body of NILIM) in ‘80s. At first, the size was 8 inch, and the density was 250KB. The recording capacity per disc was enhanced through usage of both sides of a disc and doubling the recording density. In 1990, satellite image data was delivered from RESTEC, in the form of 8inch floppy discs in sequential recording format. (5 inch double sided double density in MS-DOS format) Figure-6 One of the most popular types of floppy disk In order to regulate random access to multiple small files kept in a disc, DOS (disc operation system) was introduced. At first, several DOS’s inherent to various systems (word processor, games etc.) were used, and finally MS-DOS and Macintosh formats became main streams. More compact 5 inches, and 3.5 inches types followed. The last 3.5 inch type had been used until 2010. The 3.5 inch type MS-DOS discs were adopted as official digital media for ‘building certification’ procedure in 1990s. In short, floppy disks had once flourished, and had disappeared rapidly, and during their era (c.a. 20 years), their size, shape, density and format were very varied and rapidly changed, making the future handling very difficult. Table-2 Types of floppy disks, used in BRI 1980-2000 Diameter(inches) Sides Density Capacity 8 1 Single 250KB 8 2 Double 1.2MB 5 1 Single 160KB 5 1 Double 360KB 5 2 Double 640KB 5 2 Double 720KB 5 2 High 1.2MB 5 2 High 1.44MB 3.5 2 Double 640KB 3.5 2 Double 720KB 3.5 2 High 1.2MB 3.5 2 High 1.44MB Sectors/Track 26 26 8 9 8 9 8 9 8 9 8 9 (4) Optical media for recording program/data CD became popular in ‘80s for delivering recoded sounds, replacing the record discs (SP, then LP). This was also used to deliver programs and data in mass scale, named CD-ROM. In ‘90s, recordable CD-R was introduced to record and exchange large data up to c.a. 650MB. This media is still widely used to keep the CAD data for public buildings and infra. The expected length of life is assumed as 10 years. Production of CR-R media was out-reached to developing countries, resulting cheap and uncertain (no-brand) products. MO was also developed in Japan (5 inch, 3.5inch). 3.5 inch type is still used for recording building statistics. MO was developed in Japan and did not become popular outside Japan; however the quality is still stable. This media is mostly used in governmental offices and private firms today in small amount. The recording density progressed from 160, 230, 640 MB etc., however, the latest drive can still read/write legacy media with lower density. The round recording media is protected in plastic square housing, provided with metal shutter. The recorded data is sensed without physical contact, similar to CD-ROM. Figure-7 MO disk containing 3D data elaborated in 2000 (3.5 inch, 230MB format) (5) Electronic media for recording program/data In the beginning of IC memory chips, they were installed on the circuit boards of computers. SRAM and DRAM were memories which forget all after power off. ROM (read only memory, that can contain data without power supply) was also used for keep program for startup the system. UVROM was re-writable, but need to be taken out from the circuit board for programming while power is off. EEPROM can keep the memory during the power is off and was the first re-writable memory on board. Recordable ‘compact flash’ memory, which can keep the memory without power supply and inserted into a standardized slot of machines, was introduced in ‘90s. It was widely used to keep the image data obtained by digital camera. Interface slot/adapters for PC became popular, in order to read the photos taken in fields. The device was also used to exchange the digital data between PC’s. Soon, several alternative forms with similar chips inside were in sale (SD etc.). Since 2000, similar memory devices provided with USB interface appeared in 2000’s. The capacity of those devices has seemingly extended (20GB, etc), to keep digital photo images with high resolution. However, the durability of the data is not secured. Especially, this type is vulnerable against frequent re-writing access. (6) Data center Density of HDD (hard disk drives) rapidly developed from 20MB in 1990 to more than 1TB (1,000,000MB) today. The loss of data caused by accidental crush was mitigated by RAID technology through which broken drive can be replaced by new drive while data stored on them is continuously kept and used. Such hard disks connected to local network is widely used to store the data in offices today (file server, NAS etc.). Since 2001, communal remote data center connected through internet provide service to keep data. Physical device keeping data is hard disc drives, which rapidly enhanced the speed and density. Data center is a kind of out-sourcing of maintenance works for file servers. Data center connected to network is also useful to share the common data among remote users. However, in case of archiving data of houses and buildings, the demand for access is not frequent. The frequency of access might be only once in ten years. Consumption of electricity is useless to keep such static data. According to a survey performed by a Japanese think tank, total floor area of data center is 1.64 million square meters, and the total power consumption was 7.6 billion kWh in 2010. They also estimated that it will reach 25% of total national power consumption in 20254). 3. Currently used data formats for modeling 3D objects According to our survey, c.a. 300 data formats are used to record 3D objects. Documents on the specification of major representative data formats are collected with sample data and images of the sample objects, as listed in the table 2. Some of them (e.g. VRML) were authorized by ISO; however, other data formats still followed and replaced/added. 3D data is useful to describe the shapes of parts and structure of machines, widely produced by CAD systems and used for CAM production lines. Table-2 Major 3D data formats surveyed in fy.2010 Extension 3dm 3dmf 3ds 3dxml ac arc ase asm atr bdl blend blk br4 c4d cab cadds catdrawing, catshape catpart, catproduct cgr chr dae ddf dte dem df dlv drf dwf dwg dws dwt Name Rhino Quickdraw 3D 3D Studio 3D XML AC3D I-DEAS ASCII Scene Export Pro/Engineer, Solid Edge, SolidWorks Lightscape Material OneSpace Designer Blender Lightscape Blocks Bryce Cinema 4D TrueSpace CADDS CATIA V5 CATIA V5 CATIA Drawing 3Ds Max Characters AutoDesk Collada Data Descriptive File DTED, National Geospatial-Intelligence Agency (NGA)'s Digital Terrain Elevation Data Digital Elevation Models LightScape Parameter CATIA V4 VIZ Reader AutoDesk Composer Design Web Format Legacy AutoCAD Drawing AutoCAD Standards AutoCAD Drawing Template dxf e00 eim exp fac fbx fbl fig flt fmz ft#, rt# gmax geo, scn gml gts hrc htr ifc ige, igs, iges ini iob ipt, iam iv jt k3d kmz lay lp ls lw lwo lws lxo m3g ma max mb map md2 md3 mdd mel mf1 model mot mp ms3d mtx ndo neu Ntf obj obp off par, psm, pwd pdb pd, pd pkg plt ply pov pp2 prc, prd prw prj prt pwc pz3 raw rib rif rvt, rte, rfa rwx s3d sab, sat scn sda, sdp, sds, sdw, ses sdac, sdpc, sdsc, sdwc session sfc shp, shx, dfb, prj skp slc, sl, slo sldasm, sldlfp, sldprt slp stl stp, step svg tab, dat, id, map u3d unv vrml vue vw w3d wings wire wrl, wrz x3d, x3dv AutoCAD Drawing Exchange Format TIGER/Line Arc Interchange format Electric Image CATIA V4 Electric Image AutoDesk Kaydara FBX CADfix Log File xfig Flight Studio OpenFlight FormZ Project File TIGER, Topologically Integrated Geographic Encoding and Referencing AutoDesk Game Creator LSS OpenGIS Geography Markup Language GNU Triangulated Surface SoftImage Motion Analysis HTR file Industry Foundation Classes Initial 2D/3D Graphics Exchange Specification POV-Ray animation script 3D Object TDDDB Format AutoDesk Inventor Open Inventor JT K-3D Native Google Earth Model LightScape Layers LightScape Presentation LightScape LightWave 3D LightWave 3D 5.0 Object LightWave 3D Scene Luxology Modo JSR-184 Maya Scene ASCII 3Ds Max Maya Scene binary Quake 3 Quake 2 Player Model Quake 3 Vertex Key Frame Animation Maya Embedded Language Script I-DEAS CATIA V4 LightWave 3D Motion Maya Scene PLE MilkShape 3D OpenFX Model Nendo Pro/Engineer National Transfer Format Wavefront Bryce DEC Object file Solid Edge PDB Reader v3 CADDS I-DEAS, OneSpace Designer HP-GL Stanford PLY POV-Ray Rendering Instructions Poser PRC Adobe 3D Reviewer Adobe 3D Reviewer 3Ds Studio Mesh I-DEAS, NX (Unigraphics), Pro/Engineer Pulse Poser Raw Faces Renderman Radiance Rent Renderware Strata 3D ACIS TrueSpace OneSpace Designer CATIA V4 OpenGIS Simple Features for CORBA format Shapefile, Esri's open hybrid vector data format Google SketchUp Renderman SolidWorks Pro Engineering Stereo Lithography Standard for the Exchange for Product Data Scalable Vector Graphics MapInfo's vector data format Universal 3D I-DEAS Virtual Reality Modeling Language AutoDesk 3D Studio Animation LightScape View Shockwave 3D Scene Wings 3D Alias Wire VRML 1.0, VRML 2.0, VRML 97 Extensible 3D (VRML 3.0, uses XML, 2004) x x b, x t, xmt, xmt txt xas, xpr Xyz Direct X Parasolid Pro/Engineer Simple point cloud formart BRI also started to utilize 3D data of houses and settlements in 1993 for the purpose of landscape designing. Initial core part of the system ‘Landscape Simulator’ was developed during 1993-96 and started to be delivered through internet as ‘free ware’ since 1996. For this system, in 1995, a native data format (LSS-G(1)) was defined and programs were coded in C language to input and output in the format, which was packaged in ‘IP’ library (a kind of interpreter for LSS-G commands). The latest version 2.09 was released in March 2011, and modeling functions for creating flat site for housing complexes on the hilly land are added in 2012. However, the data format has not been changed since the original version for 17 years. In the first stage before 1996, modeling functions of the system were very limited, therefore, 3D data of houses and settlements were provided by professional draft man operating commercial CAD system, and then imported to the scene through file converters. Several file converters were elaborated to import external 3D data (Dxf, Microstation, Minicad, Vector Works, VRML, SHP etc.), coded in C language. Modeling of existing houses from photos was also often performed (.skv format). Land shape was also imported through file converter from DEM style data provided by Geography Survey Institute in 2 formats (.mem:50m mesh, .lem:5m mesh). Since beginning of 21th century, mushrooming of new formats occurred. Many of them are XML based. Usually, attributes are defined on 3D objects in various styles. Exporting/importing data of houses and settlements between different systems became inevitable for simulating various issues. Examples are urban scale fire simulation and evacuation in case of emergency, etc. Evaluation of settlements from viewpoint of global environmental issues was also required. In developing countries, CAD software without strict license control was affordable for students, and widely used to produce 3D data. In 2004, in the field of civil engineering, standard data format (SXF) for digital submission of designed data (CAD) was established by NILIM. Result of designing social infrastructure (bridges, tunnels, pavement, banks etc.) are converted into this format, and handed to the branch offices of the ministry. The SXF format specifies only 2D vector data, and 3D format is still under active discussion. However 2D digital data is far easier than the conventional drawings printed on papers for modeling designed 3D objects. Very recently 3D printers became cheap and popular. They require specific data formats. Also, displaying 3D objects including houses as augmented reality (AR) or mixed reality (MR) using popular tablet devices is useful. Usually, adaptation to this situation is performed by applications through providing file I/O functions in the form of ‘plug-in’, or through provision of external files converters among major contemporary file formats. However, from viewpoint of diachronic longevity of data, when a format will become obsolete in future, re-use the legacy data will be difficult. File conversion to contemporary file formats should have been done before the data format is forgotten. If once a file format is forgotten, the data describing an important building cannot be opened. Even an application executable is preserved; the environment to execute the program might be already lost. For example, if we have to re-use a file created by legacy CAD system and saved in an 8 inch floppy disc, double sided high density (2HD), formatted by MS-DOS, then we need following items and steps: (1) Disc drive which can read the media (2) PC to which the drive can be connected (3) OS accompanied by device driver to operate the disc drive (Windows 95 etc.) (4) Historical CAD system that can open the data file In order to perform all of these, we need a kind of zoo rather than museum to retain antique machines and software at live status. Then we can open the file which is recorded in the diskette (if it is not broken), and observe the shape of building described in the file. However difficulties still remains for us to take out the data in some contemporary file format, and re-use it. If the hardware (PC and disc drive) and software (DOS and CAD system) are not available, then following alternative ways should be taken. (1) Read all the physical tracks and sectors of the floppy disc, neglecting the read-error that may occur, and save it in a long one file in binary format. (2) Analyze the organization of file system (DOS), if file allocation table is not broken. (3) Take out the file of data which describe the desired houses. (4) Provide file converter, based on some historical document on the CAD system at that time, or directly analyze the file format Some of these problems will be solved, if we could describe (2)-(4) in a virtual environment, in which functions of disc drive, dos/OS, cad, display etc could be described in scripts. Virtual zoo of antique machines on virtual machines will be cheaper to maintain than physical zoo of machines and programs. However, if all the information on the format of the data is missing, we need analysis with try and errors. That often occurs also, if the data is given in a format which is popular but not disclosed in an available document (kept secret). In such case, we have to look into the inside of the data. Fortunately, in case of 3D data, the result of the analysis will be proved through the successful 3D presentation on the display by using some contemporary software. The achievement of such archaeological research will be recorded in the form of ‘source code’ of file converters. It suggests that alternative technologies are needed, which enables keeping data for a long term with intension. The expected term will be longer than the lifetime of houses, with less consumption of energy (electricity), and less risk of unexpected deletion/loss/alteration,. The demand for accessing speed is not the issue, in comparison to the contemporary ‘big data’ on ‘crowd storage’. Also, the risk of data leak or illegal alteration remains if the data are kept in hard disk drives which are always connected to network. 4. Development of alternative technologies (1) Static and sustainable media for recording In the history of architecture and buildings, record of construction attached to a building played important role to identify the absolute year of construction. Usually, name of carpenters are also recorded on the pentagonal timber plate. Figure-8 Historical ‘MUNAFUDA’ Figure-9 Electronic recording board, we propose Even though the amount of data saved on such device is very limited (<1kb), the length of the life of documents written on the timber plate with traditional black ink is quite long. This traditional recording media is called ‘Munafuda’ (a timber plate attached under highest beam of a building recording the date of construction and name of client and carpenters), can last longer than documents on papers. The oldest one dates in 1122 kept in Chuson-ji temple (world heritage) is estimated to have been attached to a lost library building for storing literature of Buddhism5). Documents written on timber plate are called ‘Mokkan’ in general, which are often excavated from ruins after 7th century. Conditions required for archiving 3D data describing houses and settlements are significantly different from that for synchronic transaction/exchange of 3D data for business. - Accessing speed is not essential (several days are allowed to access to the old data) - Frequency of access is very low (maybe once in ten years) - Cost/energy for keeping is important (hopefully free from power supply) - Security against illegal alteration/deletion is important In order to meet these requirements, we chose and proposed ‘curving on long-life material’. Recently fine laser cutters are available, enabling fine (middle dense) recording on lasting materials, like stones. The curving can be performed in small firms. The curved data will be scanned, and the current/future OCR technologies can easily pick up the scanned data. The curved data contains (a) 3D data itself, (b) a metafile to describe the format (for virtual converter), and (c) some additional small images and short literal explanation. (2) Metafile handling to assure readability Since 1980s, efforts to standardize digital data describing buildings and built-up environment have been made. Exchange of data between different CAD systems was desired. Fortunately these problems have been seemingly improved. Useful software that can create 3D models appeared, like SHADE, ArchiCAD, Myhome Designer, Studio 3D Max etc. have been developed. Delivery of 3D data through internet was also enabled through introduction of common DXF, VRML etc. All those efforts were directed to the contemporary data exchange between different systems. If a final ultimate permanent common data format and recording media could be defined and agreed globally, then it would be enough for us to convert any data into it, and store it forever. If it is difficult, then the second choice is to continue efforts to find more useful technologies, and ever converting the data into the contemporary most popular container at that time. This research tried to find another, the third approach. 3D data is saved in the form of a file, which is equivalent to a bit stream (0110110000), and the preservation of any 3D data is subdivided into 2 issues, namely: a. Media to keep the bit stream b. Format of the bit stream A metafile attached to data is useful to describe how the data describe a house or a settlement. We developed a compiler system, named ‘virtual converter’, which compiles the metafile and create an executable which will input the data file, parsing the grammar, and get elements that describe the house or the settlement. <Recording phase today> [1] When we preserve a file that record a house, we attach a metafile. [2] We record the set of data file and attached metafile on durable material <Replaying phase in future> (maybe 100 years later) [3] The bit stream containing data file and metafile is obtained from the recording material [4] The metafile is compiled by virtual converter, creating an executable [5] The executable will open the data file and do actions for usage The metafile in this system looks like a kind of program for replaying the data file. The metafile need not cover all the specifications of a format (e.g. whole features of dxf) suggested by the extension of data (.dxf) and omit the features that are not referred by the data. However, if some additional dialect or misinterpretation of the specification is performed in the data, it must be followed in the description of metafile, if the feature is significant to preserve. The grammar of the metafile we developed and proposed is a kind of programming language(2) to be processed by a compiler system. The grammar has following features: [1] A metafile is described by ASCII characters. [2] The grammar of metafile is similar to ANSI-C language, but simpler [3] The metafile can refer to library functions to define 3D data [4] The library functions contains LSS-G commands as subset [5] The library also provides commands for scanning file, and accessing to 3D data in memory Based on this grammar, we developed a new system of ‘virtual converter’, a kind of compiler system, which accepts and processes a pair of 3D data and attached metafile. This compiler will read the metafile and creates a run-time program, which can decode the 3D data in future. The sample applications we provided with the ‘virtual converter’ are (a) file-file converter, (b) virtual reality display, (c) tablet AR viewer (android based) and (d) web based database entry (server based), described in the (3). These are not needed directly when the data is saved. However they are useful to check and debug the metafile to make sure the future readability. The definition of the grammar (version ID) for creating the metafile must be also attached to the preserved data of each house. However, it is enough to keep the grammar book in e.g. national library, and data for each house refers to the ID of the book. Even if some of the rules for creating metafiles are changed in future, the revised version of definitions will be stored with different ID, and the change will not affect the ‘readability’ of previous data file and metafile. Theoretically, the developed method is inspired by the concept of compiler-compiler proposed by Futamura (1971) (3). (3) Prospect of usage in future Of course, we cannot clearly predict the needs and usage of preserved 3D data in future; however, it is worth while assuming some typical usages which are already practical today. For this reason, we developed 4 trial utilizing systems that can input pair of data file and metafile and execute something. All these programs/systems contain the same compilers that handle the pair of data file and metafile, while different linked libraries perform different way of utilization, providing different actions triggered by the same library commands (COORD etc.). (3-1) File-file converter The compiler is triggered by the command line console with three string parameters, which compiles the metafile (first parameter), and the executable (running on the virtual machine) input the data file (second parameter), and finally output a file with the name given as the third parameter. The current version (vc-1c.exe) is developed on VS2005 as console application, and executed on cmd.exe on Windows versions. This version processes any file format defined in the metafile, and library functions write a file in LSS-G format. This simple application was used to develop, test and debug the prototype kernel of the compiler. Figure-10 Command line (to convert a metafile In_vrml.cmm and a data file house1.wrl into house1.geo) (3-2) Presentation by virtual reality devices This system receives a pair of metafile and data file, and the obtained 3D building is presented as virtual reality (stereoscopic view). The system compiles the metafile, and the resulted executable input the data file and synthesize the result as 3D objects in the memory of the system, which are presented on the stereoscopic display. 1. Title: Data entry to virtual reality presenter 2. Selection of metafile 3. Selection of data file 4. Start operation Figure-11 Converting a pair of metafile skv_i.cmm and data file Z(C6_4).geo Figure-12 Presenting by the VR projector (3-3) AR viewing in the field The way of usage of preserved 3D data is almost same as (3-2). However this system runs on tablet devices and obtained 3D objects was synthesized on the photo image obtained by an image censor equipped on the back of the device, and the synthesized image is displayed on the LCD of the device. The tested device was ‘Android’ based, instead of ‘Windows’ based (3-2). Therefore, for this system, the source code of ‘virtual converter’, which was developed on VS2005 for Windows, was re-compiled on Android OS by using ‘ndk’ compiler system. The compiled ‘virtual converter’ linked with wrapper application written in Java language run immediately on Android based tablet and almost compatible to that on Windows system(4). Figure-13, 14 AR view of a pair of metafile and data file by tablet (3-4) Data storage and conversion on web-server with SQL database This trial system runs on the web-server, with large data storage in SQL server, which contains two different dialogs accessible through network. 1. Title: 3D data storage 2. History of past operation (upload/download) 3. Entry to upload page 4. Entry to download page Figure-15 Entry point of 3D data storage – web site A. Upload function Users can assign two local files, namely data file and metafile, and request ‘upload’ function. The server receives two files, and compiles the metafile. The created executable processes the data file and the obtained elements (coordinates, vertexes, lines, faces, solids, etc.) will be stored in the SQL server in the form of tables. This process need far more time than system (3-1)-(3-3). Therefore, the process is performed in the backyard, as a process separate from data accepting procedure. The progress of the process is shown to the client through individual web page created for each upload event. The contents visible at client is automatically updated when the status. 1. Title: 3D data storage 2. Subtitle: Upload 3. Selection of local metafile 4. Selection of local data file 5. Start uploading Figure-16 Uploading a pair of metafile and data file through web page B. Download function Users can select uploaded 3D data from the list, and assign a metafile that defines the data format to be downloaded. The executable will select and pick up elements from SQL tables, and synthesize the new 3D file defined by the metafile. This process also take more time equivalent to process A. Therefore, the progress is reported to the users until download is completed. The users can quit the contact to the web server and shutdown the client terminal during the process. Accessing to the page again lets download the result after the completion. Process A and B are taken at the same time, when a user desire to utilize some preserved old data file and metafile through conversion to contemporary format. However, if desired, the SQL tables created by the upload process can be maintained before the future download process, and SQL tables (physically files on HDD) can be utilized for data storage for a short time (<10 years). 1. Title: 3D data storage 2. Subtitle: Download 3. Selection of 3D data in storage 4. Selection of local metafile 5. Start dowloading Figure-17 Downloading a data file in any format defined by respective metafile (4) Possibilities of other alternatives We realized a system of handling data file and metafile, in order to keep the readability of data for a long time. We neither protest that the system we proposed is the only one system, nor the best system to achieve the purpose. Various other systems for the same purpose can be developed. In order to preserve a data, we attach a metafile (1) based on the compiler (1). The definition of the compiler is printed on a book and kept in a library in charge of lasting data preservation, and the future user can compile the metafile and then use the data. Assume that another compiler (2) is developed and another metafile (2) will fulfill the same purpose. We can attach both metafile (1) and metafile (2) to the original data, the future user can use both compiler (1) and compiler (2) as alternatives. If the compiler (2) performs better than the compiler (1), then those who wish to preserve data will choose compiler (2) to provide metafile, and the users of the compile (1) will decrease. Even in such case, the readability of previous data attached with metafile (1) will be assured in future. It means that the two compilers (1) and (2) are not competitive, but co-operative. We expect that a metafile is provided and attached to the data file when it is saved and preserved in the form of durable media. However, later generations can also analyze a data file described in an unknown format and describe the result in the form of a metafile. Since then, the data file will become readable as well as a data file which was attached metafile since beginning. It means that this compiler-metafile system can play a role of tools for archeological studies for digital data. The most important thing is that at least one compiler exists and is available, to provide a metafile to be attached to the data file. 5. Next steps Technical approaches toward practical production of the developed electronic recording board at appropriate cost will be the main issue. During this experimental stage, sample data will be obtained from past houses that are already lost but requested to be recorded forever and ready to be disclosed. (1) Indexing past/conventional 2D data into 3D archives Modeling 3D data from old 2D analogue records (mostly papers) and utilize 3D model for searching and indexing original historical records are being challenged. Data of individual houses are collected into settlement model, and the source data for modeling (photos, maps and drawings etc.) will be connected to the 3D model of the lost settlements. (2) Preparing for expected usage in future Simulations will be performed on settlements recalled from archives in desired formats. For that purposes, various metafiles will be elaborated based on the grammar, so that downloading from the server with any desired format for simulating will be obtained. Conclusion Today, hard disk drive (HDD) is the major device to keep the massive digital data. In order to secure the risk of crush, RAID technologies are utilized, and the out sourcing of data center management was started in the beginning of 21c (called ASP, then ‘cloud’). However, huge consumption of electricity (30Giga Watts to operate the world’s data centers, equivalent to 30 atomic plants) for drives to keep the ever increasing digital data seems to be unsustainable. In order to conserve the data which record past houses and settlements, for future generations with small energy consumption, infrequent access and high security, the method which we developed and proposed in this paper will provide alternatives. Notes: (1) LSS-G format (1995) is defined in the NILIM’s web site6), and also explained in a printed report7). Library functions available for the Metafile include LSS-G commands as a subset. (2) Summary of the grammar for the metafile is as follows: ・Numerical types: 32 bit integer, 32 bit float, 256 bit quaternion Used for definition of valuables, functions. ・Functions Called from other functions The main function can be omitted, when no sub-functions are used. ・Program flow control if(condition){・・・}[else{・・・}] do{・・・}while(condition) while(condition){・・・} for(initialize; condition; incremental{・・・} switch (condition){case } ・Library (built-in) functions data scanning from data file data acquisition from SQL server or Memory Data (same command) definition of coordinate, color, vertex definition of line, face definition of significant group of lines or faces (solid body, group of parts) (3) Futamura8) argued concept of compiler-compiler from viewpoint of partial evaluation in 1971. program A π(C,R) processes constant value C and runtime value R, compiler α converts the program π into α(π,C) (R) which process only the runtime value R, and the process becomes simpler. For example, if π is ‘Y=1+C+R’ where C = 2, then α(π,C) is ‘Y=3+R’. If πis an interpreter int, that executes source code S with runtime data R, the program int(S,R) is partially evaluated by the compiler to become a programα(int, S)(R) that process only R. (projection 1) Similarly, the relation α(int, S)(R) = α(α, int)(S)(R) is derived, where α(α,int) is a compiler that translates the source code S to make a runtime program which process the data R. (projection 2) Further, α(α, α)(int) is derived which translates the interpreter ‘int’ into compiler where α functions as ‘compiler-compiler’ (projection 3) FILE command in the LSS-G(3) format (1995) for defining models for landscape simulation performs similar. For example, a command G=FILE(B.geo); in an LSS-G file A.geo refers to another file ‘B.geo’, and put a fixed model B as a part of the model A. If the FILE command refers to an executable E, with runtime data R G=FILE(E.exe, R); then the executable E.exe is called with runtime data R and the result of the calculation is put as a part of model A. For example, E.exe will create a door which is half open (parametric definition of a model), and R gives the rate of open, then the fixed shape of a door open 30% will be resulted. In the same way, program C is a converter to import a model F.dxf described in a foreign data format, then G=FILE(C.exe, F.dxf); means a model imported (converted) from F.dxf. (projection 1) In those cases, E.exe or C.exe are native to MS-DOS or Windows environment, and cannot be used on other OS like UNIX or Macintosh. However if VC is a compiler which compiles the source code C.src and create a runtime program whose performance is similar to C.exe, G=FILE(VC, C.src, F.dxf); (projection 2) performs equivalent to G=FILE(C.exe, F.dxf); on MS-DOS system, and if VC for another operating system is provided, C.src is commonly used on it. If a common compiler-compiler CC is provided for various operating systems, and the source code of the VC.src is provided, the executable VC is commonly used among those operating systems, and the command will be equivalent to: G=FILE(CC, VC.src, C.src, F.dxf); If we use a multi-platform compiler system CC which is compatible among various operation systems, and define VC.src for that compiler as a permanent agreement as a grammar for metafile C.src, the definition of a converter C.src will be useful forever in order to process F.dxf. G=FILE(CC, GRAMMAR, METAFILE, DATAFILE); (projection 3) Once the compiler-compilers CC native to various platforms process that can generate an executable of virtual converter defined by GRAMMAR, then the executable process METAFILE to create the converter’s executable that input the DATAFILE. However, the purpose of this system is the readability and longevity of digital data, instead of the execution speed which was the basic concept of Futamura. (4) Minor difference of compilers between VS2005 for Windows and NDK for Android was identified in the handling of address calculation of arrays, minor difference between Windows (VS2005) and Android (ndk) was identified. For example, Integer array and variables int stack[1000]; int stp; are defined as global variables, and following command is executed. stack [stp] = f1(stack[stp]); If the value of stp changes in the function f1(), executable created by VS2005 save the return value into the address, using the new stp value after the execution of f1 function, while Android version save it to the address pointed by the stp value before the execution of f1 function. This kind of problems (compatibility) can be solved by extracting the program into two lines: int stack[1000]; int stp, return value; return value = f1(stack[stp]); stack[stp] = return value; Reference: 1) Statistics Bureau. Ministry of Internal Affairs and Communication. “Quick Summary Report of Census on Housing and Land 2008” (in Japanese). http://www.stat.go.jp/data/jyutaku/2008/10_1.htm (accessed 2013-7-1) 2) Tsutsumi et-al. A Study of Trend for Stock and Demolition of Wooden Dwelling House. Journal of Architecture and Planning No.649, Architectural Institute of Japan, 2010.5, 695~700 (in Japanese) 3) Rothernberg J. ‘Ensuring the Longevity of Digital Information’ 1999, RAND, http://www.clir.org/pubs/archives/ensuring.pdf (accessed 2013-7-1) 4) MIC Research Institute. ‘Survey on power consumption by data centers toward green IT’ 2010, in Japanese 5) IWATE NIPPO CO., LTD. (A local press company in Iwate prefecture) Memory of past glory of Hiraizumi: the world heritage (in Japanese). http://www.iwate-np.co.jp/sekai/sekaiisan/sekaiisan4.htm (accessed 2013-7-1). 6) Ministry of Construction (1998) Description of Data Structure of Landscape Simulation System (in Japanese). http://sim.nilim.go.jp/VRNCS/MANUALS/Man-STR/INDEX.HTM (accessed 2013-7-1) 7) Kobayashi, H. ‘The Architecture of Landscape Simulation System Ver.2.09, provided by MLIT’ Research Report of National Institute for Land and Infrastructure Management No.42, (in Japanese), 2011.3, p.53-64, http://www.nilim.go.jp/lab/bcg/siryou/rpn/rpn0042.htm (accessed 2013-7-1) 8) Futamura, Y. “Partial Evaluation of Computation Process-an Approach to a Compiler-Compiler-“, Systems, Computers, Controls" Vol.2, No.5, 1971, p45-50