Tip on transferring MiniDV and Digital8 tapes?
 

Hi everyone.
For years I have coped with VHS, standard 8 and MiniDV tapes and how to bring everything into the digital era. While I still struggle with building a workflow for VHS and standard 8 I came across a project on GitHub that turned out to be (or at least what I guess is) the best way to losslessly transfer DV tapes, in my case MiniDV to my pc. I think that WinDV and Scenalyzer is recommended here on the forum, but the thing with those is (correct me if I’m wrong) that you can not specify whether you want to save the transferred video as a raw DV stream or in a container. More important is that these do not transfer bad and corrupt frames, maybe they give you a dropped frame counter or just a dialogue saying that frames were dropped. Fast forward to a year ago when in found DVRescue on GitHub which appears to be a program created be archivists. I’m not very familiar with DVRescue but when I tried to learn it I found some neat features to help me with what I wanted to achieve with the transfer. The program is cross platform and comes with a GUI which I am not to found of as I never got it to work any OS, so I decided to use the CLI. DVRescue is made up of some modules:

dvrescue – which I used to transfer the miniDV tapes (it captures bad frames to)

dvmerge – “A script that takes multiple transfers of the same tape containing errors and combines them to create one file with the best information available for each problematic frame.”

dvsplit – which I used to split my raw transfers. It even splits streams with time code gaps!!

I know for a fact that this runs on Windows and Linux too but I used a Mac. So what I did was I connected my consumer camcorder via FireWire to my Mac. After that I started up DVRescue to transfer my my tapes. I could then inspect it in the GUI version of DVRescue and see if there were any errors. After that I transferred all the tapes with errors ones more. Then I used dvmerge (back to CLI) to combine the two ctransfers of the same tape which for the most part removed some of the errors. I repeated the process until dvmerge didn’t improve the transfers any more. I know that there is a future built into DVRescue that rewinds and rereads the part with errors when transferring so you won’t have to use dvmerge, but I never got it to work so I suppose it won’t work with consumer camcorders. At last I used dvsplit to create clips of my transfers. I was really impressed with dvsplit which all previous programs I used failed to split transfers with gaps in the timecode (except for Scenalyzer 3.5, but Scenalyzer adds unnecessary metadata to the DV stream when inspected in MediaInfo). For those of you that don’t want the raw dv streams. I think that dvpackager can be used to rewrap it.

Well, sorry for my messy post but I hope that it could perhaps be helpful to someone else…

Last but not least, here is the link to their GitHub page. I would truly recommend watching their YouTube seminars to understand a bit more about how DVrescue works, for exemple how a DV tape can be read backwards to try to correct as many errors as possible during transfer.

https://github.com/mipops/dvrescue

Thanks for sharing the project. I don't do a lot of DV tapes but the ones I've done occasionally I was fine with Sclive and a Sony HVR-M15AU deck, Maybe consumer camcorders don't have consistent results like a pro deck. Also for home videos maybe it is not that important to milk every frame from the footage, It is usually the start and end of the footage that has few problematic frames or if the tape has severe dropouts.

If a frame is not delivered by the reading circuit of the player becase an error in the tape, it will not be captured by any software. The dropped frames are there because they do not arrive in time (based on the framerate information), and are skipped byt the software; it can be a temporary problem solved by a further capture or a permanent problem because a tape drops and again, if the player does not deliver the frame, no software can capture it.

To be honest, I do not trust any automatic engine to rebuild the frames across multiple captures. In the analog capture, when you try to do a median of multiple captures and one of the frame is damaged the result is a bad, and manual intervention is necessary to exclude/duplicate/repair/merge a bad frame.
If a concealment is present because tape drops, I just recapture many times the concerned segment and decide after a visual inspection how to deal with it.

My DV archive flow is the following:

- use "ScenalyzerLive" or "WinDV" to transfer data to DV type-1 files
- run "AVPS DV Analyzer" on the transfered file for video error concealment detection to validate that there are no transfer errors
- transfer again the segments with concealment, trying to get as many correct blocks/frames as possible and choose/splice the frames together if necessary
- run ICEECC on the DV files and create 10% parity files to recover future data corruption

But in any case DVrescue looks promizing, and I will give it a try, thanks for sharing the project.