Best way to capture VHS to frame sequence?

[Stianbø]

Hi guys,
So, I'm going to digitalize a couple thousand VHS tapes (and later a couple thousand S-VHS tapes).

What would be the best way to do this?
These tapes are around 20 years old, and we are going to run through OCR and encoding after capture. That part is already taken care of, but I need to know what options I have for capturing the frames.

The plan is to set up 30 computers, with 30 VHS players and 30 Intensity Pro (BM) cards.

Doing tests now on 3 setyps with Intensity Shuttle, but the included capture software only supports 10-bit DPX as frame sequence. I'd like to be able to capture to another format. Maybe jpeg, png or any other compressed formats (whitout loosing too much visual quality). 10-bit DPX is way overkill for VHS tapes.

So, Any advice on hardware (Intensity any good??), or software to use?

Any help would be great! So thanks a lot in advance! =)

[jagabo]

There are many programs that can take a video file and export is as an image sequence. VirtualDub, for example, has the ability to export as BMP, JPG, PNG, or TGA.

[Stianbø]

Thanks. I'd like to grab directly to image sequence. Any free or not-so-expencive apps that will do this? And should I stick with Intensity Pro? (Works great so far, but I'd like to grab and save to frame sequence, and not one long movie)..

[themaster1]

why would you capture img seq instead of avi lossless or dv ? Do you plan to use photoshop after or something ?
Not to mention, you'll have no audio and/or sync later...what a mess. That over thousands of vhs...good luck buddy

As far as i'm concerned the safest way would be .bmp img seq but your strategy is wrong

[jagabo]

I'm not aware of any video capture program that saves as an image sequence rather than a video file. It's just one little extra step to open the video file in an editor and export as an image sequence.

[Stianbø]

Thanks for the responses =)

The reason why I'm saving as image sequence, is that we've got a custom made software we are running after the grabbing process. This software runs OCR on all the frames, and syncs up 3 videos to match (3 tapes shows the same thing. Left, center and right). This is for the oil industry. They have thousand of hours of recordings of pipes on the seabed, and they need it digitalized.

There are information we need to grab, regarding GPS position and other values, that we need to put into a seperate file to work with the grabbed footage. And each tape is 3 hrs, and we split it in 15min sequences.

So I just thought image sequence would be the best bet. I might be wrong =)
If you where to do this process, how would you have done it?

This is the process each machine is going through every 24 hours:
grab 3 tapes (left, center, right) Total of 3x3hrs = 9hrs.
When this is done, start our software which reads and stores all text information from these videos (OCR), and encodes the videos into 15 minute clips (mpeg).
Then we run all the data collected through filtering systems to even out the graphs generated by the increasing values, and checking with an estimate/prediction if the graph looks right.

Next day we backup the resulting files, and start all over on a new set of 3 tapes. This happens on 30 machines, 5 days a week.

[Stianbø]

I'm not aware of any video capture program that saves as an image sequence rather than a video file. It's just one little extra step to open the video file in an editor and export as an image sequence.

How long time does it take to open up 9 hours of footage, and save as image sequence? We cant affort loosing a couple hours on that process (and we need it automated), so thats the main reason why I need image sequence right away.

We're doing tests now, capturing to .avi and letting our software split it up into images, but that process takes 3-4 hrs.

[jagabo]

How long time does it take to open up 9 hours of footage, and save as image sequence? We cant affort loosing a couple hours on that process (and we need it automated), so thats the main reason why I need image sequence right away.

We're doing tests now, capturing to .avi and letting our software split it up into images, but that process takes 3-4 hrs.

Why don't you decimate the image sequence? Surely you don't need every single frame? You can use the Frame Rate dialog in VirtualDub to save only every 10th, 30th, or 100th frame. That would speed that step up significantly.

I just ran a test. An 18 minute DV AVI file exported as PNG took about 45 minutes when exporting every frame. When decimated by 100 (1 frame every 3.3 seconds) it took 18 seconds.

[Stianbø]

Problem there is, that I need to convert 9 hrs of movie into image sequence before I can work with the data. And in your example that process alone would take about 22 hrs.. Which obviously is too much =/ And it requires user input.

We do already have an automated sollution to split an avi up into images, but that process is too slow. So I'd like a way to grab directly to image sequence. The software included with the Intensity Pro only supports 10bit DPX as sequence. The rest are movies.

So there are no software that will work with my Intensity Pro to output image sequences, instead of movies?

Thanks in advance,

[themaster1]

maybe dscaler can do that, i can see in its advanced settings that you can record img sequences in Tiff (rgb or ycbcr),
problem is it don't work on my pc (win7 x64), only the "take one still picture"work you might have better luck so give it a try

[jagabo]

Problem there is, that I need to convert 9 hrs of movie into image sequence before I can work with the data. And in your example that process alone would take about 22 hrs.. Which obviously is too much =/ And it requires user input.

Like I said, if you used VirtualDub to create the image sequence from the recorded video you could elect to save only every 10th, 30th, 100th frame (or whatever decimation you like) and cut that time down by 1 or 2 orders of magnitude. At 1/100 your 18 hours would drop to 20 minutes. Saving as BMP rather than PNG would be several times faster still.

[Stianbø]

That is true, but we cant just cut it up like that. We need 25 fps. But I'll look into it, and I appreciate all the help I can get =)

[jagabo]

That is true, but we cant just cut it up like that. We need 25 fps.

In your original post you said you were running OCR. My decimation suggestion was based on that. If that is the only need for image files I doubt any text will be on screen less than a few seconds. You certainly won't have 25 different screens of text every second.

[maryam209]

I need "sequence slideshow" . Than what site can I download?