Picked up the HV-20 yesterday from Sixth Avenue Electronics in northern NJ yesterday.

Took some 24p video of my son playing ice hockey - wanted to experiment with 24p editing on my Mac (quad-core Mac Pro, 4 GB RAM, 10.4.9, Final Cut Pro 5.1.2)

To be frank, I wasn't understanding some of workflows being discussed here (DHVSCap to MPEG Streamclip? Why and how?)

So I played, and did this (in part based on another post here - by flyingscott I think)

- Footage capture in 24p mode on HV-20

- Firewire connection to my Mac

- Set up a HDV 1080i/60 sequence in FCP (preset). Set the editing timebase to 23.98 in the Sequence settings menu

- Utilized the HDV-Apple Intermediate Coded capture preset

- Batch captured (it seems you cannot log and capture specific clips with the HDV/AIC setting - selecting it still jumps right to the capture now screen - it does auto-detect stop/start, so separate clips are saved.

- Upon completion of capture, file format is 1440 x 1080, 29.97 fps, Apple Intermediate Codec, pixel aspect 1440 x 1080

- In FCP, selected all the capture clips, selected Tools > Cinema Tools Reverse Telecine.

- Cinema Tools launches in the background, and clips are batch processed (what default settings??? Hmmm...)

- After roughly 10-15 minutes, the 20 minutes of footage conversion is complete - the format of the clips in the Fcp browser are now: 1440 x 1080, 23.98 fps, Apple Intermediate Codec, pixel aspect 1440 x 1080

I think the results are decent so far, but I can't say I'm getting the same quality yet as the NYC street scene (hv20 raw.mov) posted elsewhere in this forum. The "raw" file posted contains clip properties that state:

Apple Intermediate Codec, 1920 x 1080, Millions
16-bit Integer (Big Endian), Stereo, 48.000 kHz

My "Raw" captures have clip properites that state:

Apple Intermediate Codec, 1440 x 1080 (1888 x 1062), Millions
16-bit Integer (Big Endian), Stereo, 48.000 kHz

Anyone know why? Was the original file capture via HDMI (Black Magic?) Unlike most other capture presets, which you can duplicate and edit in FCP, the HDV/Apple Intermediate Codec cannot be duped and edited.

Nonethless, the handoff from FCP to Cinema Tools worked very well, it "chose" the correct reverse telecine option it seems (for me, CC or DD looks best, and when I manually tested and compared, the auto-converted files appeared to use one of these settings.

Other issues I need to look at seperately from the above workflow:

- I didn't have the Cinema Mode setting on while shooting 24p (have others, does it matter?)

- Need to shoot a better subject and in a better environment. Handheld, fast subject motion and panning, poor lighting in a rink - I think this is a factor too.

If whoever shot that NYC street clip can post detailed as to how the file was captured and converted, that would be much appreciated.


Greg

So digging somemore, it seems the NYC street scene was likely captured using DVHSCap from Apple's Firewire SDK. I downloaded and tried some new footage (raining all day here in NJ, but stood on my porch and took a minute of footage, this time with Cinema mode on as well)

I captured two ways - in the manner I posted above, and as follows:

- Capture with DVHSCap (.m2t file). This manner does show a clip property of 1920 x 1080

- Imported to MPEG Streamclip 1.8. Exported Quicktime Movie to Apple Intermediate Codec at native resultion.

- Processed resulting file in Cinema Tools to remove pull down (DD or CC looks best)

In comparing the results, there is very little difference between the two, and it brings up the question of why use DVHSCap to capture a "1920 x 1080" frame, when, unless I don't understand this correctly, the maximum output of HDV/Firewire is 1440...It would seem that I'm taking a lot more time and creating much large files for the most modest of quality improvements (I would say that the DVHSCap/MPEG Streamclip method yielded arguably slightly more contrast than the first method)

Now for pros, this may matter - but for prosumers, perhaps not.

I will post screen grabs from both processes shortly.

I could be missing some obvious things here, so any input would be appreciated.

Thanks

I tried posting images to the picture gallery here, but I kept getting errors, so

http://rapidshare.com/files/26157859/Canon_HV_24p_Compressions_Screen_Shots.zip

Let me know what you think...

How cathartic talking to myself here.... :)

So in studying things more, I noted that there was strobing when using the first method (FCP HDV/AIC capture - Tools>Cinema Tools Reverse Telecine) It seemed it may have been choosing the wrong cadence mode.

So I did some more tests, and found that if I took the clips captured in FCP (HDV/AIC), and manually launched Cinema Tools, selecting BC Reverse Telecine mode to Batch convert the clips - worked! No more strobing, making the clips identical to the DVHSCap > MPEGStreamclip process, sans the fine contrast issue.

I've made a comparison movie - it's compressed as an MPEG-4 file (done using Apple's Quicktime to Apple TV conversion), but it clearly illustrates the difference.

Now if there were a way to have Cinema Tools default to the BC mode, the auto process would work as well, I presume.

Here's the comparison m4v file...

http://rapidshare.com/files/26180053/Auto_Manual_Reverse_Telecine_Test.m4v

Curious what others are discovering....


Greg

Interesting stuff... but I think I am even more confused. :confused:

Our work flows are very similar. When I run a rev telecine, then open the .rev file with quicktime and look at the movie properties, I get what you see below. What is weird is that it shows the "normal size" as 1920x1080, how did it come up with that, seeing as how the format is 1440...? and what is with the 1888x1062?

Also, my video looks excellent in Final Cut... but as soon as I export it looks washed out. Any ideas?

http://www.sacreativeservices.com/forum_stuff/rev.jpg

Good news... Apple just announced Final Cut Studio 2, looks like this will address a lot of these issues:

http://www.apple.com/finalcutstudio/finalcutpro/

Good news... Apple just announced Final Cut Studio 2, looks like this will address a lot of these issues:

http://www.apple.com/finalcutstudio/finalcutpro/

I've already ordered.... :rolleyes:

Still, it see confusing - on one hand, it talks about rendered-free placement of clips, regardless of codec or FPS..yet,in reading the info on Cinema Tools Pro, it seems you still need to pay a visit there to work with 24p video...hopefully it will be a lot more intuitive and automatic than it is now...however, I suspect that Apple envisions most users of FCP Studio will NOT be using consumer cameras (one, in this case, that doesn't provide a pulldown flag...)

But the BUMMER is DVDSP is still at version 4! No upgrade? No increased support for true HD-DVD and Blu Ray authoring? That's a big hole in the package...

good points, but I think it will be worth the upgrade...

I was under the same impression that all hdv footage was 1440x1080 as well.
I wonder what's going on here...

I was under the same impression that all hdv footage was 1440x1080 as well.
I wonder what's going on here...

It is 1440x1080... and that's what natively comes out of the cam. You can however upconvert through the capture/edit process, and any competent editor can do... which I suspect is what's happening here somewhere.

It is 1440x1080... and that's what natively comes out of the cam. You can however upconvert through the capture/edit process, and any competent editor can do... which I suspect is what's happening here somewhere.

Yeah, it seems the DVHSCap creates a 1920 frame - but why would you want to, unless you were trying to use the footage with other native 1920 material.

I would think working at the native res would be best, but perhaps I'm missing something here...

Yeah, it seems the DVHSCap creates a 1920 frame - but why would you want to, unless you were trying to use the footage with other native 1920 material.

I would think working at the native res would be best, but perhaps I'm missing something here...

LOL... you and me both... I'm sort of biting my tongue though... don't know too much about FCP or DVHScap

Stumbled across JES Deinterlacer, a free utility for OS X available here:

http://www.xs4all.nl/~jeschot/home.html

There is an option to detect cadence, so there is no need to manually locate and snip the beginning of each clip (to identify the A frame)...JES will analyyze and do this (and it seems Cinema Tools will not)

You can also easily batch process in JES, which I cannot get CT to do (and at least use the right cadence settings) There are also some nice time remapping features and other image enchancement tools.

And did I say it's free?

Exported files can be in any format your system already supports, so my test workflow was to capture in FCP as HDV/Apple Intermediate Codec - batch process the files in JES with a custom size setting of 1440 x 1080 at 23.98, and used the Quicktime Export option to save HDV 1080 using the Apple Intermediate Codec.

Sequence setting in FCP is AIC, HDV, 23.98 - clips drop in with no rendering required and you're on your merry way...

Stumbled across JES Deinterlacer, a free utility for OS X available here:

http://www.xs4all.nl/~jeschot/home.html

There is an option to detect cadence, so there is no need to manually locate and snip the beginning of each clip (to identify the A frame)...JES will analyyze and do this (and it seems Cinema Tools will not)

You can also easily batch process in JES, which I cannot get CT to do (and at least use the right cadence settings) There are also some nice time remapping features and other image enchancement tools.

And did I say it's free?

Exported files can be in any format your system already supports, so my test workflow was to capture in FCP as HDV/Apple Intermediate Codec - batch process the files in JES with a custom size setting of 1440 x 1080 at 23.98, and used the Quicktime Export option to save HDV 1080 using the Apple Intermediate Codec.

Sequence setting in FCP is AIC, HDV, 23.98 - clips drop in with no rendering required and you're on your merry way...



I wish (s)he could deinterlace his/her (the name Jan is androgynous) home page with that program! :)

http://www.xs4all.nl/

(To be fair, I bet this is just his/her ISP's home page)

Very interesting... I wonder how it's able to figure it out. I was under the impression that the reason Cinema Tools had a problem was because the HV20 doesn't set any flags, like the more expensive cameras do... I could be wrong.

Anyhow, good stuff!

Very interesting... I wonder how it's able to figure it out. I was under the impression that the reason Cinema Tools had a problem was because the HV20 doesn't set any flags, like the more expensive cameras do... I could be wrong.

Anyhow, good stuff!

I thought similarly - the cadence check in JES is a two pass process, so presumably it's analyzing the sequence of frames and picking the first progessive frame based on that analysis. When you think about it, software should be able to analyze and detect the sequence - why CT does not is unclear.

Let me know if you have similar success... I tried four or five clips, and it detected each one correctly.

people:

the video coming out of the camera is indeed 1440x1080i. its just that the pixels are not square. or rather, the pixels are square, but there is aspect ratio information in the mpeg headers which say, "this video should be displayed as 16:9, so scale it." that's where the 1920x1080i comes from.

the .m2t file that's coming out of DVHScap is indeed 1440x1080i. its just that quicktime sees the 16:9 flag in the header and scales it appropriately.

people:

the video coming out of the camera is indeed 1440x1080i. its just that the pixels are not square. or rather, the pixels are square, but there is aspect ratio information in the mpeg headers which say, "this video should be displayed as 16:9, so scale it." that's where the 1920x1080i comes from.

the .m2t file that's coming out of DVHScap is indeed 1440x1080i. its just that quicktime sees the 16:9 flag in the header and scales it appropriately.

So essentially it is creating larger files because of scaling, but obviously, there is no more "real" resolution. And since it's already HDV on the tape, the data is compressed at the source, and DVHSCap woudn't offer an qualitative improvement over other methods, it would seem.

So it would appear an FCP HDV/AIC capture would give you maximum resolution/smaller file sizes, and ability to use the workflow above for 24p to complete the process.

Now, the question I have left is why does FCP NOT present the normal Log and Capture window when selecting HDV/Apple Intermediate Codec as the Capture preset? - you immediately are prompted for a file name, and then go immediately to the Capture Now process which you end with the escape key. Does anyone know why this is the behavior? Why can't I scrub through the tape and set my in and out points and then do a batch capture? If you choose the standard HDV Capture preset, you CAN access the normal log and capture features of FCP.

I'm not 100% certain, but I believe it is because AIC does not utilize the timecode on the tape and generates a new one... or something along those lines.

In comparing the output between JES Deinterlacer and Cinema Tools, I noted a subtle degredation in quality...exploring, I discovered that JES only has a hard-wired "Medium" quality option for the Apple Intermediate Codec Quicktime output - no way to change to high.

Indeed, the resulting file is smaller and has a lower average frame rate the the Cinema Tools-processed version of the clip, which presumably maintains the settings of the original clip.

I emailed the developer of JES Deinlterlacer, and he responded in minutes to say that a new version, "expected soon", will support direct output of the AIC format with a High quality setting (whether he meant an upgrade to the AIC codec from Apple or an upgrade of his application wasn't clear...)

Even so, the quality difference is fairly subtle, and in the interim, you could output in an M-JPEG format with a quality akin to the AIC/High setting....

Thrang, as I'm poking around I see your comment that fcp cannot accurately log and capture and that I will have to just use capture now or batch command. Is that still the case? My experience so far is with my new HV20 and FCP 6.02. My understanding is that the GOP length is too long for accuracy? Please let me know your conclusions. My old sony digital8 has had no problems... thanks

Ok.. shot some video last night.. discovered that my output was set to DV and not HDV.. successfully logged and batch captured.. Progress! First time I've been able to do that.