FORUMS: list search recent posts

Neurosurgery video

COW Forums : Compression Techniques

<< PREVIOUS   •   VIEW ALL   •   PRINT   •   NEXT >>
nickeljohnson
Neurosurgery video
on Jan 19, 2007 at 5:02:26 pm

I am the media specialist for the Department of Neurosurgery at the University of Utah. Part of my job entails going into the OR and taking pictures of the case setups and then taking intraoperative video captured from the camera on the surgical microscope.

Currently we are using a win2000 computer that upon capture automatically encodes the video as WMV. This really erks me because my editing station and storage database are on macs. I can edit the WMV using 3rd party flip4mac but it is really bad. I am currently working on replacing all of this but that is a different forum.

Here is my question. I want to trans code these videos to a better codec for storage on my database. Something that will allow me to easily edit it but also keep my videos files as small as possible. Any help would be greatly appreciated.


Return to posts index

mike velte
Re: Neurosurgery video
on Jan 19, 2007 at 9:13:40 pm

Editing and recompressing already highly compressed video is waste of time. Windows Media Video and most other web formats are lossy compression codecs.
To get decent quality video at 1 mbps, one should start with at least 25 mbps digital video. DV is lossy, but ever so slightly and can be re-edited and recompressed several times without noticable loss of quality.
Now if you are just doing straight cuts only...no titles or effects, Windows Media Encoder 9 comes with a File Editor utility that can cut clips. Then there are a few "joiner" apps that will join the pieces WITHOUT recompression. I use Rejump Video Joiner on a rare occasion for this.

http://www.video2stream.com


Return to posts index

Charles Simonson
Re: Neurosurgery video
on Jan 19, 2007 at 10:35:54 pm

If you are looking for an intermediary less-lossy codec to store source files and want to maintain QT compatibility with accurate editing capabilities, I would recommend either MJPEG for interlaced sources or Photo-JPEG for progressive sources. If you have iMovie HD installed on the system, I have also found the Apple Intermediate Codec (originally intended for HDV captures, but works well for other formats too) to be a good choice. If maintaining Alpha channels is important as well, then the PNG codec is an option, but it is slower than any of the above.

Overall, I would say that ANY of the above options is a much better way to go about this than your current solution.



Return to posts index


Ed
Re: Neurosurgery video
by
on Jan 20, 2007 at 10:13:53 pm

Yeah, c'mon, it isn't brain surgery! Sorry, I've never had a chance to say that
to someone around brain surgeons before. :-)
How about the Sheer Video codec for archiving too? And doesn't saving as
PhotoJPEG convert the YUV video into RGB? Not sure of that, but I think so.
Then there's the dreaded random QT gamma change when playing PhotoJPEG.
Ed


Return to posts index

Charles Simonson
Re: Neurosurgery video
on Jan 21, 2007 at 1:52:19 am

Ed, Photo-JPEG can handle both RGB and YUV at various color sampling modes. At 76 - 100%, P-JPEG is 4:4:4, and below 75%, it is 4:2:2.

Here is a great resource for learning about editing codecs that I have referenced for a long time now:
http://www.onerivermedia.com/codecs/

Also, as mentioned above, Sheer is a good codec I have included a link that compares it with the codecs I noted earlier to it. The only issue I have with Sheer is that it is not free and you will have to install it on all systems you intend to play back your clips on:
http://www.bitjazz.com/en/products/sheervideo/speed/broadcast_codecs.php


Return to posts index

Ed
Re: Neurosurgery video
by
on Jan 22, 2007 at 5:45:22 pm

What does it make it *at* 75% then? :-)

Photo JPEG is RGB-only at 100%, and I don't think you're right about the 76-100%. There's a lot of conflicting
information on the web about it. Do a test. Take a 10 second video clip and export it at best (100%), high (75%)
and then export one at just a touch below 100%, as little as you can move the slider. The 75% file will be significantly smaller
than the 100%, of course, but so will the "almost" 100% clip. My clips were 32MB, 48MB, and 117MB. The clip at (approximately)
97-99% is less than half the 100% size. 4:2:2 at 99% or less? I think so.
A lot of the "75% setting" opinions talk more about how you're not gaining much quality by going higher, not whether it switches to 4:4:4
mode that low.
Ed



[Charles Simonson] "Ed, Photo-JPEG can handle both RGB and YUV at various color sampling modes. At 76 - 100%, P-JPEG is 4:4:4, and below 75%, it is 4:2:2.

Here is a great resource for learning about editing codecs that I have referenced for a long time now:
" target="_blank">http://www.onerivermedia.com/codecs/"




Return to posts index


Ed
Re: Neurosurgery video
by
on Jan 22, 2007 at 5:50:02 pm

I forgot to include a link.
http://www.revostock.com/phpBB2/viewtopic.php?t=292&sid=7b906ffb095dde15a00...
Marco from OneRiverMedia even tells a poster that it's 4:2:2 until 100%.
Ed


Return to posts index

Charles Simonson
Re: Neurosurgery video
on Jan 22, 2007 at 10:07:21 pm

Photo-JPEG definitely has compatibility with YUV at 4:4:4. Even the BitJazz link I provided contains comparisons of Sheer to PJPEG in both RGB and YUV environments at these sampling rates. As for where the 4:4:4 and 4:2:2 modes kick in/out and where RGB takes over from YUV, I don't know. You're right in that there is a lot of conflicting information on this out there. From the results you noted in your tests, I would say that the 100% file was likely RGB, the 99% clip was 4:4:4 and the 75% clip was 4:2:2. If I have some time this week, I can do a real in-depth analysis on this.



Return to posts index

Ed
Re: Neurosurgery video
by
on Jan 22, 2007 at 10:20:53 pm

You're right about the compatibility. I'm thinking more about what products support it.
The Blackmagic and Aja products only output 4:4:4 RGB (although they support YUV in 4:4:4).
The one exception I know from them is the Decklink Multibridge which *does* support 4:4:4 YUV.
Bluefish only does RGB too. Not sure about the others.
It's weird that there isn't a more definite source on-line for all of this. I expect that with marketing
hype, but even the geeky sites say different things. I'll look into the 99% is 4:4:4 YUV and 100% is RGB thing.
I have my doubts, but I'm wrong often enough, to be ready for it again. :-)
Ed


Return to posts index


nickeljohnson
Re: Neurosurgery video
on Jan 22, 2007 at 7:37:16 pm

Well I appreciate all of the information I have received. It sounds like I will trans code them to Photo-jpeg at 75%, so as to reach the smallest size with quality. With my new setup I want to capture to DVCPRO HD tape or right to the hard drive DVCPRO HD. The I can do any thing with the video and keep my quality.

One question concerning YUV, does WMV change it to RGB? I looked and didn't find any information.


Return to posts index

mike velte
Re: Neurosurgery video
on Jan 22, 2007 at 8:16:58 pm

[nickeljohnson] "One question concerning YUV, does WMV change it to RGB? I looked and didn't find any information."

Good question! I found this;
http://en.wikipedia.org/wiki/VC-1


http://www.video2stream.com


Return to posts index

Charles Simonson
Re: Neurosurgery video
on Jan 22, 2007 at 10:29:48 pm

Windows Media at this time is only a 4:2:0 codec. This is one of the main reasons it is not generally suitable for editing. There is talk about releasing a better sampling version for a later implementation of VC-1, but nothing is publicly available yet.

Per your capturing, if you are shooting on DVCPRO HD, then the best option would be to keep it DVCPRO HD (if your DVCPRO HD setup has firewire connection capabilities). No reason to convert it from DVCPRO HD to something else if you are using a Final Cut system workflow. The only benefit to capturing to PhotoJPEG (or MJPEG is shooting interlaced) is if you need to capture with an SDI card, which would normally require very large storage disks for capturing as uncompressed files.



Return to posts index


Ed
Re: Neurosurgery video
by
on Jan 23, 2007 at 6:27:03 pm

This from the Microsoft site:
http://www.microsoft.com/windows/windowsmedia/howto/articles/UnderstandingH...

4:1:1. Y is sampled at every pixel, but the Cb and Cr color information is only sampled at every fourth pixel, saving even more bandwidth. This sampling rate is used in consumer DV cameras and is currently the default sampling rate for the interlaced mode in the Microsoft Windows Media


Return to posts index

Charles Simonson
Re: Neurosurgery video
on Jan 23, 2007 at 7:13:12 pm

Hmmm. Looks like they haven't updated that page in a long while (page has a date of Jan 2004, well before the final VC-1 spec was ever submitted). It is true that for MS' "original" spec of VC-1 (WMVA), 4:1:1 was the default for interlaced encoding. (*Note that WMVA is no longer supported, but if you have just Windows Media Encoder installed on your system, then you probably have some version of the WMVA encoder installed.) Windows Media 9 Standard (WMV3) doesn't even support an interlaced mode, and thus all it ever supported is 4:2:0. The SMPTE "approved" spec of VC-1 only supports 4:2:0 as well, in both interlaced and progressive modes.

I know this because myself and my company is heavily invested in the development of VC-1 encoders in both the post production and broadcast spaces. Obviously for our broadcast products, understanding and getting interlaced encoding right is paramount. While I could certainly see benefits to using a 4:1:1 color space over 4:2:0 for interlaced encoding, this just isn't possible as of today.



Return to posts index

Ed
Re: Neurosurgery video
by
on Jan 23, 2007 at 7:48:44 pm

Leave it to Microsoft to leave an old document on their site (and for me not
to catch it). :-)
Ed


Return to posts index

<< PREVIOUS   •   VIEW ALL   •   PRINT   •   NEXT >>
© 2017 CreativeCOW.net All Rights Reserved
[TOP]