Is This Even Do-able??

[Wannago]

Hi All,
Relative video "N00b" here...I have a question regarding a proposal I made to my boss and have been told to cost it out.

I guess the first thing I need to do is determine if my concept is even feasible.

I work for a major railroad. Our rail network (tracks) span thousands of miles.

We have videotape footage of almost every mile of this track and as you can appreciate, these are recorded on SEVERAL videotapes.

Head Office (Senior Management) often requests to view the footage of a particular location/area so they can better understand and appreciate what might be going on out in the field. This is very useful in the event of a mudslide, avalanche, accident, etc.

Using the current videotapes means searching through box after box to find the right tape, then searching that specific tape to find the desired footage.

I have proposed a project in which I will capture, compress and store this footage on one(or more) of our IT servers. The plan being that once this is done, we can develop a front end or even integrate into one our many exsisting systems which would allow us to quickly and easily access specific video footage for a particular location.

I know that the footage can easily be captured and compressed. Should this project progress, I will probably come back here for suggestions on the best compression method to use, hardware suggestions etc., but for now my main concern is IF this can even be done.

My biggest issue is trying to figure out if/how I can capture and store this footage and be able to directly access specific areas of this footage. Specifically:

As I mentioned earlier, we have thousands of miles of track in our rail network.

Our entire network is further broken down into smaller subdivisions consisting of approx 100 miles worth of track each.

Each subdivision has posted mileage signs at every mile and at other significant locations. Consequently, any time we refer to a specific location, we refer to an exact mileage on a particular subdivision.

For my project to work, I would like to be able to quickly access the compressed video and go to a specific mileage (mile sign) on a specific subdivision. I guess it would be like a DVD with 100 or so chapters?

Does this sound like it is even doable? I know my company will allow me to do this (as long as the price doesn't get out of hand) and I know all the Senior Managers / Executives would appreciate (read = bonus to me??) the ability to sit at their desk and view the footage of a specific area, on their computer through our existing IT network, quickly and easily any time they so desire, without having to wait for a specific videotape to be found then have to congregate in front of a TV /VCR in some conference room.

Sorry for the long post. Thanks in advance for your feedback.

[ricoman]

I don't know the particulars of compression mode etc. but what you ask is feasible. I work for the CT DOT and we have a server based system called Digital Highway that has video of all State roads. It is updated once a year. Anyone in the building (100's of computers) can access any traffic signal in the State or travel any State road from their desktop. It replaced a similiar CD base system, but everyone had to have their own set of CDs (approx. 50-60 per set for each computer). Now it is all on a server that anyone can access. We use it all the time, it saves us time and many road trips. Our index is set up by mileage or crossroad. Is this the type of thing you are looking for?

[SLK001]

Sounds like a fun project. If you get the option, re-shoot everything on a good quality camcorder. If not, then it might be possible to convert the video tapes (provided they haven't degraded too much). You can use Window's Movie Maker to convert each individual segment to .WMV (which does a good job of compressing video). Also, .WMV is a native Windows format (no additional codecs needed) and is included with all Windows installations.

[Wannago]

I don't know the particulars of compression mode etc. but what you ask is feasible. I work for the CT DOT and we have a server based system called Digital Highway that has video of all State roads. It is updated once a year. Anyone in the building (100's of computers) can access any traffic signal in the State or travel any State road from their desktop. It replaced a similiar CD base system, but everyone had to have their own set of CDs (approx. 50-60 per set for each computer). Now it is all on a server that anyone can access. We use it all the time, it saves us time and many road trips. Our index is set up by mileage or crossroad. Is this the type of thing you are looking for?

Yes - this is very much in lines with what I am thinking - we're even in the same basic 'biz (transportation). Actually, it's probably EXACTLY what I am thinking - only a Digital Railroad.

I assume that your Digital Highway data is compressed on a network server? Not knowing too much about this vid stuff, I'm having a hard time trying to figure out how I would be able to setup an index to access the landmarks in the compressed footage (like yours does with mileage and crossrads). I can understand how - if I was making a DVD - I could setup a bunch of "chapters", say one for each mile sign - but I don't know how to do something like this with compressed video. Without being too much of a pain in the @$$ - how is this done? Do you administer this or were you involved in it's production? I'd love to learn more about how this was done as it could very likely save me a great deal of time learning from someone who has already produced something like this.

SLK0001 - I won't be able to reshoot it. It is taken over many months as a certain piece of evaluation equipment travels and examines our complete network. At this point, I don't even know what type of camera was used to record it - all I know is we have BOXES of video tape.

Thanks for the responses.

[lordsmurf]

I don't know that you'll save any money per se, but it will surely increase productivity (better organization, easier seek times for materials). I guess that could be productivity=costs, just depends on how the boss looks at it.

[One808]

Depending on your budget, you could check out Cisco's Digital Media System. Allows you to digitize all the video, and then make it available as Video on Demand across the entire network...

http://www.cisco.com/en/US/netsol/ns620/networking_solutions_package.html

[kschang]

It may be easier to upgrade to digital video cameras first, THEN slowly upgrade the existing archive (if you even want to). Existing video cameras are easy, as there's a bazillion different IP-ready cameras, some of then even run on WiFi, and many will even rotate/tilt/zoom via remote.

[ricoman]

Unfortunately, I'm just a user not a developer. I'll try to get more info for you, hang tight, it might take a few days to get a response.

[Richard_G]

Virtually anything is do-able, it just depends how much time you have to put into it. There are probably better ways of doing it but I'll give you my thoughts. If you get the video into the digital domain by whatever means you can then do just about anything with it. You could create a number of virtual DVDs, one for each of your sub-divisions. Each could have up to 99 chapters (the maximum the DVD spec allows). If you need more, maybe you could start each chapter at every 2 mile marker for instance.

By doing it this way, using a simple menu system a user could access a sub-division and go straight to a particular section of track. This would require every computer that needed to access the footage to be installed with a software DVD player such as PowerDVD or similar, but you could use a DVD compliant resolution lower than full D1. You could, for instance, compress to mpeg1 at VCD resolution, which is still a legal DVD format.

Just my two penneies worth.....

[Nelson37]

Making the video network accessible is the easy part. Just digitize and copy the files on the server. Format should be something already present on all PC's, this eliminates MPG2 and conventional DVD structure, unless you want to pay for and install DVD player software on all your PC's. This leaves either WMV or MPG1. Note that MPG1 can be full 720x480 for PC playback. Do some test encodes and math to determine storage space and feasibility for each codec, WMV is smaller but does not seek as well.

The tricky part is the "indexing". I would use the tool already present on all PC's, the Directory structure. Make a Video directory, within that directories for 100, 200, 300, etc, within the "100" directory seperate files for 10, 20, 30, etc, each one covering 10 miles of track.

How you break this up would depend on how your track is organized, I would keep the files relatively short for fast searching. Maybe name the directories Albuquerque and Santa Fe, whatever fits your structure.

Also using a prog which allows fast seeking, such as VirtualDub, rather than a conventional movie player.

[Scorpion King]

I like the dvd folders idea because the user has the choice to 'continue' to play the video where the individual files idea is going to require the user to be constantly browsing for the next mile post section of track. The dvd folders would allow the user to view the entire sector (like 100 miles) without doing anything but watching the video.

What would be nice is a program that the users would be in instead of being in folder view in Explorer trying to figure out how to manually find and play a dvd file. The program would give them the ability to browse for a specific section of RR track then play the desired dvd menu where the user could then select the mile marker to start playing. Get a programmer to write a custom program for you, probably be worth it.

Problems: Getting all those boxes of video tape onto a hard drive. Don't call me. ok, I'm too busy.

[ZAPPER]

Whoa! Sounds like a big project! But kind of cool and doable.

Evryone here knows I can't keep my mouth shut, so here are some thoughts.

First off, the access method should use a simular terminology to the way you normally decsribe a track location. I don't know but maybe something like

Region #5
Chicago-St.Louis
Mile Marker 35
West Bound Veiw

Each line would drill down to the folder/sub folder/file/clip that was wanted.

Perhaps if you gave us a sample of like how you would send a repair crew out to a specific spot or something we could work out the details to fit your companies terminology. If no such complete system exist, it might be time to invent one.

Directional veiws would be nice to add if they exist, and even if they don't yet, plan on them for the future.

I think that a spreadsheet type database search, drilling down to the file could be the better option because then you could add cross refference data/querries like crossing gate number/locations or switches or whatever other elements you have and possible refferences to them. Like if you knew that "Switch #1032" had been problematic or a civilan called in and said that your crossing gate at Iowa State Hwy 105 was damaged you could also input that search information to get to the clip.

I am not a fan of compression, and it also depends on what format the "Tapes" are right now. Another of the first steps is to catalog and index the tapes that you have and decide just how much information(footage/quality) you want/need to file. Do you need 30 frames per minute of every inch of rail? Would cropping/editing them down to a single frame every 20' of rail be OK? Keep in mind that in real time a train filming every inch at 10 MPH V/S a train filming every inch at 70 MPH is going to give you different amounts/quality of footage for the same mile of track.

In all reality the system should be in place before the footage was shot. With onboard sensors stamping the tape/hard drive with all sorts of information. DD/MM/YY, GPS position, odometer reading from various benchmarks, train/camera travel speed and so on. I would bet that such multimillion dollar systems already exist and sends it out to the main office by satellite.

OK I got off on a tangent.

Create a catalog of the tapes/footage

Digatize/capture the tapes(depending on the format, either on the fly analog to DV or if you are lucky enough, straight from a DV cam) to the computer

Devise a search method/database for the track location and link it to the footage

Publish simple but good directions on how to use the datadase to find and play the file.

Start looking into upgrading the filming and indexing of the newer versions of tapes. The amounts of information that could get into this database are limited only by the imagination. Telemetry of the trains systems and track is I think what I am working up to



Good luck and keep us posted! This sounds like a cool project.

[Wannago]

Thanks for all the feedback everyone. Great forum!

Time is not an issue - in fact, if I am given the green light it will get me off a shift-work rotation and onto a regular schdule while I work on it. So the time frame is not a big deal - 6 months - maybe even a year to do all this will probably be fine, as long as i keep it "cheap".

Agreed - having this done with a digital cam would have been convenient, but that is definitely not an option. The videotapes should be in reasonable condition, as I believe they are maybe 2-3 years old at the most.

This footage will need to be available to potentially thousands of users via our network. Like most big organizations, our computer workstations are quite "locked down" and are running Win 2K. So - the format will likely need to be WMA or MPG1 etc. - something that can be played back on a standard windows PC. I definitely won't be allowed to go ahead if I will require any third party software for the playback.

We have a couple existing applications that we use to access data pertaining to a specific location. I was thinking that I could piggy-back on one of these and use it to also access the video footage of that area. Also, I'm sure I could get a very crude web page added to our intranet site (or even a custom built search app) that would allow users to (via dropdown menus for example) select a subdivision, then drill down to select a specific mileage or landmark location. I do think I want the ability to keep the video rolling from this point ie. if i select mile 50, i can watch the footage beginning at this point and unless I stop, it will continue on until the end of that particular subdivision. This is not a show-stopper, so to speak, but it would be a nice feature.

For example, many of our train dispatchers control trains on territories they have never seen, often hundreds of miles away. A big feature to having this footage available is that we would be able to have our dispatchers take a virtual road trip - actually see the territory they are controlling without ever leaving the office. This would have a big benefit in putting into perspective some of the factors that need to be considered when they are controlling the movements of a 10,000 ft - 17, 000 ton freight train. If we had to begin a new "search" after every 10 miles or so, this functionality would not likely be utilized.

Ultimately, if it was possible, I would like to take the footage of a complete subdivision and compress it (as a single large file) and manually add index marks for each mile sign and/or landmark. This single file would be able to be accessed directly via our network with the user going directly to the specific index mark and then being able to play / ffwd / rew etc. from that point. I know it won't be this simple, but no matter how I do it, this is how I envision the end result (at least from the end-user's perspective).

Thanks again everyone for your input.

[Wannago]

Whoa! Sounds like a big project! But kind of cool and doable.

Evryone here knows I can't keep my mouth shut, so here are some thoughts.

First off, the access method should use a simular terminology to the way you normally decsribe a track location. I don't know but maybe something like

Region #5
Chicago-St.Louis
Mile Marker 35
West Bound Veiw

Each line would drill down to the folder/sub folder/file/clip that was wanted.

Perhaps if you gave us a sample of like how you would send a repair crew out to a specific spot or something we could work out the details to fit your companies terminology. If no such complete system exist, it might be time to invent one.

Directional veiws would be nice to add if they exist, and even if they don't yet, plan on them for the future.

I think that a spreadsheet type database search, drilling down to the file could be the better option because then you could add cross refference data/querries like crossing gate number/locations or switches or whatever other elements you have and possible refferences to them. Like if you knew that "Switch #1032" had been problematic or a civilan called in and said that your crossing gate at Iowa State Hwy 105 was damaged you could also input that search information to get to the clip.

I am not a fan of compression, and it also depends on what format the "Tapes" are right now. Another of the first steps is to catalog and index the tapes that you have and decide just how much information(footage/quality) you want/need to file. Do you need 30 frames per minute of every inch of rail? Would cropping/editing them down to a single frame every 20' of rail be OK? Keep in mind that in real time a train filming every inch at 10 MPH V/S a train filming every inch at 70 MPH is going to give you different amounts/quality of footage for the same mile of track.

In all reality the system should be in place before the footage was shot. With onboard sensors stamping the tape/hard drive with all sorts of information. DD/MM/YY, GPS position, odometer reading from various benchmarks, train/camera travel speed and so on. I would bet that such multimillion dollar systems already exist and sends it out to the main office by satellite.

OK I got off on a tangent.

Create a catalog of the tapes/footage

Digatize/capture the tapes(depending on the format, either on the fly analog to DV or if you are lucky enough, straight from a DV cam) to the computer

Devise a search method/database for the track location and link it to the footage

Publish simple but good directions on how to use the datadase to find and play the file.

Start looking into upgrading the filming and indexing of the newer versions of tapes. The amounts of information that could get into this database are limited only by the imagination. Telemetry of the trains systems and track is I think what I am working up to



Good luck and keep us posted! This sounds like a cool project.

Thanks Zapper - sounds like you might have some basic railroad knowledge??

Actually, you are pretty close with your description of the location. For ours, it would be something like:

District (usually comprised of 2 or 3 subdivisions)
Subdivision (approx 100 miles worth of track)
Mileage (sign posts at every whole mile)
and/or
Landmarks (signals, crossings, switches, etc.)

Direction of view would be an option as well, however I think the footage might have been taken in one direction only, as our piece of track evaluation equipment made it's way across our rail network over the course of one summer.

When we dispatch our section forces (repair folks) we send them to a specific mileage on a particular subdivision.

Good comments on the spreadsheet-style search option as well as the frame rate. To be honest, I have no idea what type of camera was used to record the footage - apparently it is an onboard camera on our evaluation car. Definitely not DV though. The piece of evaluation equipment is worth SEVERAL million dollars, so I would not be surprised if the camera used is a $45.00 Wal-Mart special (LOL). The piece of equipment definitely has all the data availability you cite (even GPS) but I do not believe that data is transferred to the videotape footage.

[ZAPPER]

No thanks needed. I don't want my name on this trainwreck!


How do you play the tapes? I mean are they VHS tapes? 8mm? Other?

I think that getting the tapes to the PC is the easy part (Providing that they are not some bastard million dollar railroad service car play only in some special deck kind of tape) The tapes may actually be some very high quality tapes or from a decent cam and recording deck.

Lets assume that you are watching them from a VCR deck on a regular TV with regular RCA jacks, composite=analog

Analog means that it will have to be converted/captured to a digital format for the PC

Let the wars begin... I like straight DV. But.... end result is important to decide and MPEG editing has come a long way I have heard.

To get analog to DV you need either an external analog to DV box (A good handfull to choose from) or an internal card (A few of these around too) or digital cam that will do pass thru (pretty common also and my set-up when things are actually set-up)

There is also Analog to MPEG hardware out there and cutting and splicing MPEG is supposed to be pretty easy too (no first hand experience about this for me)

Recording it to a Stand alone DVD recorder and then dumping it to the PC is common too.

I guess what I am getting at is, What equipment do you have and how good do you want the end results?

I would think that you could do it for nothing out of pocket until it comes to storage and then the company may have to come off of the hip(unless you could scroung up some unused company servers or a bushel full of harddrives.

[ZAPPER]

The Evaluation car may have data linked to a timecode on those tapes. This might help you refference the position of the car to the tape and mile marker as opposed to watching all of the footage looking for milemarkers on the tape and manually entering such data.

Also you may want to do some poking around to find out what other companies are doing along these lines. Trade magazines, any counterparts you may know, so on.

[kschang]

Didn't I read somewhere that one of those Japanese-sounding names (something-domi?) programs have patched the AVI format to add external indices/chapters and stuff like that?

If so, this then becomes relatively easy... TV or capture/digitzer card, and the Video Player hooked up to it. Record as AVI, add indices automatically per time, then manually for location and perhaps, mile markers (if they have GPS coords, even better). Then you generate some sort of cross index and allow jump-to access.

[Nelson37]

I'm thinking of ways to do the "searching" using existing Windows tools, and not requiring additional software.

Sounds like conventional DVD structure is not feasible due to the requirement for several hundred copies of DVD player software, so that leaves MPG1 or WMV.

Larger files gives longer continous play, but less direct "seeking". Smaller file gives better seeking, but less continous play.

A "playlist" could combine these two goals. Whatever segment size makes sense for your needs, but using my earlier scenario, the "playlist" for "SubDivision 1" could be 10 files, mile1-10.mpg, mile11-20.mpg, mile21-30.mpg, etc. Drawback would be if the user wanted Mile 95, they would have to press Next 8 times to get to that section. Advantage would be the playlists could be created ahead of time. Whatever segments are used, somebody somewhere is going to want the last mile and then the first part of the next segment.

A simple program could be created whereby a user requests a particular subdivision and mile marker, then the program builds a playlist including a number of sections before and after the requested one. This could even be user specified, in increments dictated by the individual file size. Ie, if you make 10-mile files, then the user could specify 10, 20, 30 etc miles before and/or after, and the program would use 1, 2, or 3 files in the playlist.

Other thoughts - this could be done with an HTML page, and even made available through the Internet for field personnel not connected to your network. Drawback is this would make filesize a major concern, perhaps a second set of lower-resolution files could be created, with the primary, hi-quality set for in-house, network use only.

Audio-there may not be any present. It may not be needed, though the engineers may want to hear the "clickety-clack", this could be useful for finding problems. However, you have to at least make a short demo with "I been working on the railroad" playing in the backround! This could sell the project all by itself! Could also have commercial possibilities, maybe the Lionel people would buy a short segment.

Also, a backup set on DVD should be created, this would also be good for offsite use.

[SingSing]

Sure, go ahead and do it, and we should benefit from less accidents by Amtrak.

[JohnnyMalaria]

Just how many hours of footage do you have? I suspect 1000's.

I think this is going to be your first major hurdle. Obviously, one person with one PC isn't going to be able to digitize all the footage in the timeframe for your project.

Is the footage realtime (i.e., full frame rate NTSC video)? Do you need to maintain that? Would one frame for every 50ft of track suffice? This would dramatically reduce the amount of storage required for the digitize video etc.

I think you need to be prepared to set up parallel digitizing operations (probably by the dozen!)

[ZAPPER]

Start with baby steps

Play part of a tape, ten minutes or so, or maybe one mile of track.

Assuming that you have the equipment to get it on to a PC, do that.

Play with it a little, cut, splice, change the speed, change it to a couple of different formats, save a file or two and then try to access them from a different company PC or have someone else follow the path to the file and play it, and then critque the process.

[kschang]

The other way is do it as an HTML page or Wiki, which allows for very free-form linking... But it won't look as pretty.

Looks like you have a lot of footage to digitize. If you can to time-lapse it'll cut down on storage and such.