Capturing HV20 Footage
 

Hey ya'll,

What have you been using to capture HV20 DV and HDV footage? Right now I'm using Premiere Pro 2, but sometimes it has trouble capturing in/out points and it does not use scene detect with HDV.

Any recommendations?

Try HDVSplit:
http://strony.aster.pl/paviko/hdvsplit.htm

It detects scene changes and creates seperate files with configurable names.

I just started using HDVsplit and it works perfectly with the HV20. Splitting scenes properly is critical for 2:3 pull down removal from 24p footage if you're doing this in post. Using After Effects, I had to click the "guess 2:3 pulldown" button when interpreting footage as each clip had a different sequence. HDVsplit will log your clips unattended...very handy.

yeah i heard hdvsplit is great!

1.) does it have a preview window to view the footage? if not, how do u know what you're capturing?

2.) so hdvsplit removes pull down? what exactly is pulldown and why do you want to remove it?

1. Yes it has one.

2. No HDVSplit doesn't remove pulldown it cuts the clips accurately so that you can remove the 24p pulldown in programs such as After Effects.

When you record 24p with the HV20 you're actually recording 60i because the HDV standard doesn't allow 24p. This is called 2:3 pulldown.

http://en.wikipedia.org/wiki/Telecine

Have any of you guys tried viewing the .m2t files from HDVSplit in Windows Media Player? My aspect ratio ended up all weird. I recorded HDV on my Canon HV20 and the previews looked perfect in HDVSPlit, but when I play the files in WMP 11, the video is not widescreen (looks like 4:3 instead of 16:9) and the people are stretched.

Any ideas? I checked HDVSplit site and no one has reported similar problems.

Terry,

I too have noticed this when playing either M2T files or the Cineform Intermediate. HDV compression uses 1440 x 1080 pixel count (4:3) instead of 1920 x 1080 (16:9). It is probably playing in WMP with a 4:3 aspect ratio, therefore not decoding correctly, I think. Somebody check me on that.

A question for the thread:

I plan on using HDVSplit for capturing and I have VirtualDub for pulldown removal, but I think I read that it CANNOT read M2T to remove pulldown. I use Vegas and the Cineform that comes with, but I don't want to have to convert each clip to Cineform AVI one by one. Is there an open source program that can remove pulldown from M2t clips?

Thanks!

Canon advertising says "The HV20 captures true 1920 x 1080 High Definition resolution video using miniDV cassette tapes." This is not strictly true, is it? I have the PAL HV20. I have captured my files off tape using Vegas 7 or Cineform Connect HD HDLink. My files are reported as 1440 x 1080 (with a pixel aspect ration of 1.333 to create the 16x9 aspect ratio).
Presumably, it cannot record better than 1440 x 1080, if it is using the HDV specification?

Would it be possible to capture true HD 1920x1080 using the HDMI output? Is there any current hardware/software solution to do this? Has anyone done it, and, is the quality better than off tape?

Patrick

Yes, when you record footage in any HDV camera, it's recorded as 1440x1080 - this is how the codec stores the image.

With the HV20, Canon is pointing out that their sensor is a true 16:9, big CMOS sensor - most other 3chip camearas have narrower sensors at the same size and use pixel-shifting voodoo to achieve the wide-look. In all fairness, pixel-shifting doesn't look bad - most cameras do it very well. The HV20 just happens to be one that does it the way you'd expect most cameras to do it - and this may be why the image is so good for the amount we're paying: the big 16:9 sensor.

If you Capture live footage straight from the camera's HDMI or Component out ports, you can record the image at 1920x1080. The point here is to bypass the recording of the footage to HDV - playing back HDV and capturing this way will still give HDV quality video in the end.

Thanks, Robert.
I see on another thread that you have been capturing full HD from the component outputs. Presumably you can record to HDV on tape at the same time? Is the component HD output noticeably better than HDV tape?
Patrick

3:2 Pulldown?
 

All. I shot HV20 at 24p with Cine Mode at the Highest Resolution (1440X1080?). When I capture, I have been getting a "double image" in Video Vegas. I tried HDVSplit. I brought the footage into After Effects and tried many settings. I finally got it to look great, but every 3rd frame is a duplicate. I want to have an output of true 24 frames per second, like film. Finally, I got the image to look great without interlacing, banding, ghost images, of what I can see, but there is still every 3rd frame is a duplicate. When I tried different 3:2 settings the image got messed up again.
Another question, just to confirm: IS there a way to capture true 24p 1920X1080 off of the HV20, instead of the 1440X1080? If so, can someone point me to the specific sites with capture cards/info, etc?
Help,
Thank you!
Thomas

AE is probably missing it
 

AE must not be doing as good a job in doing the 3:2 pulldown if you're still getting the interlaced frame. We've been working on a "free" windows process to achieve good pulldown. Using the farnsworth method we've got nice clear 24p in 1440x1080. HDV is non-square pixels, so for final output it gets unsquished to 1920x1080 for players that recognize the bit. Unfortunately most players like WMP seem to ignore this for AVI. For WMV it seems to have it. You can fix the aspect ratio in your NLE before final output then choose a format that supports it (e.g. MP4, MOV, WMV, MPG).

Steve is making a nicer version with Vegas:
http://dvinfo.net/conf/showthread.php?t=92893

My current ghetto method:
http://yousillyman.blogspot.com/2007...-hv20-m2t.html
Needs to be updated with vdub 24p scripting.


The only way to get the 1920x1080 direct from the sensor is to get it from live HDMI but it comes with its own issues (look for threads on BlackMagic Intensity). You can get an HDV unsquished version from playing back from tape too but that's not too different from taking the m2t then upscaling it to 1920x1080 in software. In my case I prefer to downscale to 1280x720 but that's just me.

Hope that helped rather than confused.

Worked... Almost...
 

Mike,

I followed your instructions to a "T", (I think). Everything seemed to work until the: E) on your instructions (http://yousillyman.blogspot.com/2007...hv20-m2t.html). I only got two fixedaudio.avs files from my four .m2t files that were in the directory. One fixedaudio.avs file was only 1 kb, and the other 0 kb in size.

There were no regenerated video files whatsoever (should there be?)

I went to go to F) on your instructions, running VirtualDubMod. I dragged the fixedaudio.avs file onto the opened window and got the error:

Avisynth open failure:
TFM: d2v file is not a d2v file or is of unsupported format!
(C:\RedRocksDay11\HV20day11a.m2t_fixedaudio.avs, line 4)

So... without the VirtualDubMod being able to open my files, and without all the fixedaudio.avs files, and with no new video files I could not go on to G), etc., on your tutorial.

Can you help me please?

Hmmm ... you're the second one to have this sort of problem. The original template.avs must not be created correctly for you. Wonder which part isn't clear. I have a hunch ...

for now let's rename your directories to the following:

c:\capture = the directory where you have you m2t files
c:\capture\24p = the directory where you installed go.bat, dgindex directory, virtualdub directory, template.avs)
c:\program files\avisynth 2.5 = the directory where avisynth 2.57 was installed

a) Please make sure that dgindex.exe and its other files are in 24p\dgindex\ ( C:\capture\24p\dgindex\dgindex.exe, dgdecode.dll, etc.).

b) Please go to you plugins directory for avisynth (c:\program files\avisynth 2.5\plugins) and make sure there is a dgdecode.dll in there. If there isn't go ahead and copy it from your "c:\capture\24p\dgindex" directory.

[as is, dump your 4 m2t files into c:\capture, make sure there are no other files mixed in, then run:
cmd.exe -> c: -> cd \capture\24p -> go.bat c:\capture]

If the above worked then you were either missing the .dll files or specifiying trailing \ in your path. When calling 'go.bat', do not put a trailing '\' in your path. In your path the correct way to call it is "go.bat c:\redrocksday11" (good) not "go.bat c:\redrocksday11\" (not good)

c) Go to your 24p directory open up go.bat. Change the contents to this:
d) Make sure that template.avs is inside the 24p directory. (e.g. c:\capture\24p\template.avs).

[as is, dump your 4 m2t files into c:\capture, make sure there are no other files mixed in, then run:
cmd.exe -> c: -> cd \capture\24p -> go.bat c:\capture]

If the above worked, look at the name of your original "capture" directory. If it had any spaces in the name (e.g. "capture me"), it causes problems for go.bat when you use double-quotes. Dgindex fails to create the original xxx_m2t.avs file and thus fixavsdelay.avs has nothing to work on and does a dumb create of a blank _fixedaudio.avs.

doublepost ...

Also, you may want to consider steve's new process ... http://www.dvinfo.net/conf/showthread.php?t=94235

informative thread. thanks!

Same exact thing...
 

Didn't work. Came up with 0KB sized AviSynth Script. It won't go into VirtualDubMod. Just gets an error.

Back to the drawing board

Thanks,

Argh,

Thomas

I "think" I know the problem. I have been putting my m2t file in the c:\capture\24p directory versus just the c:\capture directory.

When I get home tonight I will move the m2t file up a directory and see if that fixes things ...

Mike, did you have a chance to try the plugin? I saw about 190 views to the thread but no responses, so was wondering a bit.

I think I might create a second version that isn't a Vegas Plugin, but just a straight .exe file that folks can run on their PC's and specify a folder that contains the video files they want to convert, similar to the .bat scripts, but that automate VDub as well. It would be 95% re-use of my existing code, so shouldn't take too long to produce.

-Steve

Issues?
 

My files were in the caputre and not the 24p directory, and I still get the issue. Don't think that is it...

On another note... now there were a few "gotchas" in the installation directions on the tutorial (maybe this was a source of problems?):
1. ActivePerl was a little confusing as to exactly which one of the files on this page to download and install (I do not have a 64bit machine):
http://www.activestate.com/store/dow...5-08d58c2648ca
2. MPA Source there are two files: Cache and Float. Both contain different files for mpasource.dll
3. There is no "Left click" install on msu-ls-codec.exe, or Lagarith, or HuffYUV, only right click install.
Perhaps, it is some of these things that were somehow not done correctly, as to why this isn't working?
Could someone grab all of the final files that will be needed, exactly as is and put them in a directory and zip it up for download, with appropriate names, etc. as to exactly where to put it so there is no chance of mistake? I really want to make this work for my edit.

Please?
 

Steve,
Could you do that please? Make a stand alone .exe that would do all of this stuff?
That would be amazing.
Thank you,
Thomas

No problem Tom. I'll try to get to it tonight if I have a chance. We had to take our dog to the emergency vet last night and will probably have to spend some time there tonight as well depending on how he's recovering from his surgery.

--steve

Cool
 

Steve,

Thank you!!! I hope your dog is OK... :-)

PS: Some great things would be this:
a. I can bulk alter all of my files to get rid of doubled frames/ghosting/interlacing, and have true 24p frames from my HV20.
b. Sound synced up automatically (in the same file as the video file would be awesome)
c. Proper aspect ratio for final video file so no resizing needed.
d. converts to 1440X1080 or to 1920X1080?

Thanks again,

Thomas

Thanks Tom, the pup is having a hard year. He had a stomach contortion (or something) last night where his stomach flipped around and cut off his esophagus. They had to do surgery to flip it back around and sew it to his side to keep it from happening again. I believe they also had to remove his spleen. He's 13 and we're just hoping that he'll make it through the post-op. Won't be getting many new toys anytime soon with the $$$ it's costing me, but I can't bring myself to put him down.

For your other q's:
a. Should be able to do this in bulk without a problem. DGIndex & VDub seem to do a good job with this. I'm just batching things around those apps.
b. The sound sync is in my current plugin and would be in the exe as well. It all winds up properly sync'ed in the final single output file that VDub generates.
c. VDub maintains the proper aspect ratio, doesn't alter it from the original, at least in the testing I've done with a few codecs. The cineform one I had (which shouldn't be working for me) produced it's file but was very narrow and tall with funky colors when viewed in Win Media Player, but when I put it into the Vegas timeline it look perfect and I rendered a 24p WMV with it and that played perfectly in WMP.
d. Are you asking to scale up a 1440 to a 1920 or vice-versa? Not sure whether VDub does that offhand, but if it's scriptable in the VDub language then it should be possible.

Steve

@Steve
I'm away from my editing PC so I can't try out the dll quite yet.

@Gavin
Give that a try. I found a note at Doom9 with your error message, it happens to older releases of DGIndex.

http://forum.doom9.org/showthread.ph...382#post917382

I've tried the same process on a Windows XP SR2 and Windows Vista Home Premium box and encountered a similar problem to yours initially. Back then it had to do with the template.avs not being found by dgindex because of the pathname.

Could you post a sample of your d2v file?

@Thomas
1) You want the Windows (x86) MSI file for 32-bit version. I'll update the blog. I was also compiling my version of the script with PAR for those that don't have Vegas ... unless Steve beats me to it. 8-)
2) I use the Float version because its newer. Unless you're running 80486 or less you probably have a built-in FPU.
3) The choice of codec wouldn't have anything to do with it. That part doesn't come until dgindex is done.

Could you post a copy of your d2v file also?

BTW, would have zipped everything up into one package but the virtualdub license agreement said not to bundle it with any other application. Hmmm ... I didn't see that clause in avisynth and others ... hmmm.

for Gavin and Thomas
 

http://www.sendspace.com/file/d71x3y

I've zipped up avisynth, the 24p scripts, plugins, and lagarith codec into a zip. You still need to install HDVSplit and Virtualdubmod separately.

Decompress the file and you will find an installme.bat. Run that, it will step your through Avisynth install. Then automagically create c:\capture and install lagarith. Drop an m2t file into c:\capture. Then follow the instructions in howtotest.txt. These new instructions use a compiled version of my perl script, so that hopefully takes away the perl version question.

Don't worry if you've already had avisynth previously installed. It seems to take care of itself.

Thank you!
 

All who helped me with this. It seems to finally work. Important is to keep the name of the file with no spaces, but also, this new download, and install seems to work so far. I only tried it with one file so far though (and there is no lip sync sound in the file to really put it to the test), but I will test more files soon, and hope for the best. As for the one file I did test, it seems to be great so far. No ghosting/interlacing/flicker, etc.
Thanks a lot!
If I don't write again for a while, it is because all is well!
Thank you!
Thomas

Hey Tom

I got a chance to update my Vegas script to a standalone executable tonight that you can use to automate and batch the workflow. You can find it at the below location:

http://www.szudzik.net/vegas/scripts/HV20Pulldown.html

It's C#, so requires the .Net Framework 2.0, just like vegas but his doesn't have any Vegas dependancies. The UI is similar to the vegas script, so that documentation will pretty well guide you through. I'll setup a better documentation page soon.

Here's the script doc page: http://www.szudzik.net/vegas/scripts/24pWorkflow.html

The difference here is that one the 2nd tab, you can select all the media files, from wherever you want and add them for processing. This version also does not require the copying of the file to the "working directory" so it's much faster.

Let me know what y'all think..

Steve

Compression Tools
 

This is great. It seems to work very well with my footage now...

Only thing...

When I save as MSU or similar (uncompressed) from Virtual Dub the footage is so huge (like 20X the size it was captured from Vegas [by the way, is there any benifits to using HDVsplit for Capturing, over Vegas?]).

I went from huge original 1.5 gig .m2t files captured out of Vegas, to now 22 gig .avi files (uncompressed) out of VirtualDubMod!

They don't play properly, and are impossible to view/edit, though I think uncompressed is the smartest way to go (hoping that this film eventually goes to theaters, or high def dvd).

Is there another compression that is much smaller that looks just as good, or is there another way I should be doing this? Perhaps making a high res and low res version and subbing in the high res for final output from the edit? I've never done that before (only having edited in NTSC before this).

Is that what you all would suggest, or a different compressor, so I can stay high def, but be able to move around.

When I went to MPEG 4 it looked OK to me, and was actually smaller than my original .m2t file, and played real time (but for some reason in Vegas it only loaded the sound!). But I assume this is a huge quality hit from uncompressed.

Any suggetions?

Thank you,

Thomas

Mike,

You are the man ... this seems to do the trick!

Now to play with all the codecs and pick one that I like ...

Thomas,

Yes, the uncompressed/huffyuv/lagarith/msu files are large because they are lossless (I'm capt obvious 8-). They're not meant for viewing but work well for final render. One trick is to create two versions, one in lossless and another in WMV9/XviD. The WMV9/XViD usually looks good enough for web output or a small LCD. But when you project it to a wall you will notice the compression by the loss of some detail (flat colors on a spackled wall). You can use the WMV9/XviD file for editing. Then when you're happy with your timeline you can rename the WMV9/XViD file to something else. And then use the Lossless version with the exact same name. Reopen your project and it will pickup the new files. Do your final render.

Another alternative is to recompress your lossless to cineform via Vegas. It's another extra hour of re-rendering and the chroma upsample makes it look weird in wmp/mpc but it will look great for final render and can be viewed real time.

From the practical perspective, I tend to render in WMV9/XViD since my output is bound for web. But if I ever want to inflict some high resolution vacation videos to my hapless visitors on the DLP projector ... I can confidently create the high quality MPEG2 or H.264 versions from the lossless versions (which I can regenerate later on).

---
BTW, you should look at Steve's vegas .dll script or his standalone exe. It even automates the virtualdub part for you.

How to get 1920x1080?
 

Is there a codec that will give me the full 1920x1080? I did XVid and it did 1440x1080, so things looked a little squished. Or, is there a setting somewhere in VDub or the codec that does this?

Thanks ...

VirtualDubMod Frame Settings
 

I want to edit my material for a hopeful release in theaters (true 24 frames per second). My special effects, done in After Effects, and Lightwave 3D, will be done at 24 true frames per second. All of these effects shots will have to be reedited into the film edit, with the proper sound sync, etc. and I do not want any ghosting/double frames/artifacts/interlacing, etc.

Of couse, there will be the need to put it to dVD or HD DVD at a later date (30fps), but that is another topic.

Since I am editing for 24 fps, when I output from VirtualDubMod the default frame rate that comes up, under the Video Frame Rate Control: is 23.976 fps. You can change it to 24 frames per second but there is a warning that Changing the framerate will cause audio/video desynchronization. There is also an option to change the frame rate "So video and audio durations match" (24.013 fps).

My question is: should I just leave it at 23.976, change it to 24, or change it to the 24.013 fps? Before I edit this entire film I want to make sure I am setting the proper settings for all the footage so nothing bites me on this in a few months or something.

Also, should my project in Vegas be set to 23.976, 24, or 24.013? Maybe I just leave it at 23.976, and do final render at 24?

Will this cause interlacing or some other issues if I don't go with 24 or 24.013 the whole way through?

I just want to do the right thing and not have any mistakes.

Thank you,

Thomas

thoughts ...
 

@Gavin
The aspect ratio and size isn't a limitation of the codec but of the file format you choose to encode in. If you choose XviD the compressor has an option to define non-square pixels if you want to retain 1440x1080 as the dimensions. But so far, I've found that MPC/WMP/etc ignore the flag. But if you encode in WMV (via Windows Movie Maker or Vegas) you can create a WMV format that is properly 16:9 on display. The way I get around it is to override the MPC settings to 16:9 or do it from the projector. Now if you're using MediaCenter then I understand your pain. It doesn't recognize non-square in AVI files so you'd need to resize the video.

Now, for a totally free solution. Yes, you can do it in Virtualdub. You need to go to the Video -> Filter -> Add -> Resize -> W; 1920, H: 1080. Choose either Bicubic or Bilinear (it's practically the same method used by HDV to unsquish) -> Save As -> Full Processing mode (needed when using filters). Most NLEs can do this for you. I prefer to do it on final render rather than as a prestep, only because I think (blind faith) the NLE will do a better job mixing other effects that way.

@Thomas
Personally I would leave the source files in their original 24p (23.796) format and then drop them into the 24p timeline. Let Vegas figure it out on the final render. I've mixed 30f+60i/24p/+60i sources on a borrowed Vegas box and it handled it well. I'm no professional though, you may want to visit the Vegas forums in DVInfo.net for a better answer.

Uncompressed AVI
 

Steve,

I am using your stand alone script. It is working wonderfully!! Thank you immensly. I wish I had this from the start. So much dragging and dropping of hundreds of files was making me crazy.

One thing... It is set to Uncomressed AVI with DOES create immense files. When, before, when I was doing it "by hand", before your .exe version, I was using the "MPA" uncompressed it also was huge, but these "Uncompressed AVIs" are immense!!! They are making the MPAs look small!

I would set it to MPA, but I do not see that on your pulldown on Compression area, like I did when I set the compression by hand in VirtualDubMod (yet I do have MPA in my AVsynth plugins). Is MPA unavailable with your tool? If not, how to you put it there?

Also, can you change sizes in your .exe? I was making low res versions at 75% scale before. I will try the WMP9 at full scale as well and see if that works.

Thanks again,

Thomas

Mort than one 24.exe at once?
 

Since the conversions take Soooo long, and I have to make high and low res conversions can I open more than one 24p Pulldown Removal Workflow Engine at the same time? It doesn't seem to take up much cpu power or ram, but I didn't know if it would mess something up to run 2 groups at the same time, such as one high res and one low res both running, or two different high res batches running on the same computer.
Thank you,
Thomas

Thomas, my guess is that it would work ok, though you probably would want to specify separate working directories, or choose not to remove the temporary DGIndex files.

--Steve

Problem capturing from HV20 using HDVSplit or Adobe Premiere Elements 3.0
 

Hi everyone,
It sounds like you all know what you're talking about when it comes to capturing 24p off the HV20. So, here's my issue: When I first got my HV20, I shot 1.5 minutes in HDV mode (I think, but it might have been DV), then grabbed it off the miniDV onto my PC using HDVSplit. I then bought Adobe Premiere Elements 3.0, shot some more footage (this time for sure with HDV mode and a lapel mic), and now can't capture it using either HDVSplit or Adobe. I set my camcorder to "Play" mode, start either HDVSplit or Adobe and try to capture, but it says there is no device connected or it's not recognized. I read somewhere that this could be due to the fact that I recorded HDV and DV on one miniDV. But, I don't see why that would do anything. Does anyone have any tips? Thanks.

Sounds like maybe an Windows XP driver issue? I'm pretty sure you need XP SP2 (service pack 2) to get the HDV device driver. If you captured DV first, then put the camera in HDV output mode, Windows gets very confused. Search this forum for specifics, but you should check the device manager under "sound, video game controllers" (not "Imaging Devices") for "AV/C tape device/subunit" or something to that effect.