Capturing 1080 60i as 720 24p or 30p at DVinfo.net

Laurence Kingston · June 21st, 2006, 11:03 PM

It seems to me that since you are losing vertical resolution anyway, and since the Sony HDV cameras interpolate some of their pixels, that converting to 1280x720 at 24p or 30p really isn't giving up much actual resolution that you haven't already lost in going to progressive. In any case, the resulting footage looks great and makes my 3.06 P4 with external firewire hard drives seem to absolutely fly. How much real world resolution am I actually giving up if I go this route?

David Newman · June 21st, 2006, 11:23 PM

For spatial resolution you are losing very little. All you are losing is temporal resolution by going from 60 fields to 30 or 24 frames. Some don't consider that a lose at all, unless you shooting reality TV or sports which use the higher temporal resolution.

Laurence Kingston · June 22nd, 2006, 09:40 AM

Just to make sure we're on the same page, what I was asking is how much actual practical resolution you are losing when you downrez to 1280x720p as you go progressive, given that you are already losing about half your vertical resolution and at least some of the horizontal resolution was interpolated anyway.

David Newman · June 22nd, 2006, 10:26 AM

yes, I'm in agreement.

Laurence Kingston · June 22nd, 2006, 11:34 AM

Boy is that cool! My current project has a lot of spontaneous zooms and pans, so it will look better in 1080 60i, but starting with my next project, I'll try to be a little more careful with the zooms and pans and edit either 24 or 30p.

Another observation: a lot of new HDTV TVs upconvert SD DVDs and movies from blockbuster seem to look pretty darned good even on the larger screens. Movies shot with SD DV cameras don't get treated quite so well. The uprezzing algorythms ignore the interlaced video and the 60i gets deinterlaced and actually loses resolution.

This has got me going on a progressive kick much more than before. None of my stuff is going to film and I've never worried so much about progressive vs interlaced because I actually preferred the extra temporal resolution of 60i.

Viewing my stuff on a relatively inexpensive LCD HDTV has changed my mind however. 60i gets deinterlaced on the fly and not only does the vertical resolution drop, but it looses the temporal resolution anyway. It seems to me that being a little more careful about pans and zooms and converting to 30 or 24p 720 on capture will give me an image that will uprez rather than deinterlace and downrez on one of these 1280x720 HDTV sets that seem to be so common these days.

By the way, does anyone know if an average 1280x720 HDTV will detect and uprez 30p properly. I know that mine at least does a pretty good job with 24p. I have a feeling though that when it sees 30p video it bypasses the uprez routines and deinterlaces just like it would with 60i. Aside from this issue I would rather work with 720 30p, but this has me worried.

Thomas Smet · June 22nd, 2006, 02:49 PM

This does work great. The only thing to think about is that interlaced video is filtered so the resolution is actually a little bit lower than if the camera was to shoot progressive. For example on the SONY Z1 in the CF modes you do get around 540 lines which is about all one filed can give you. In interlaced mode however each field may only have around 405 lines of detail. So if you shot CF and down sized to 1280x720 you would actually get a little bit more detail than if you threw out a field from normal interlaced video. Of course shooting 60i does give you the option of converting to 1280x720x60p instead of just 30p or 24p for great smooth HD video. If you use a really good deinterlacer that can rebuild the missing lines then you may be able to get a lot more detail.

Laurence Kingston · June 22nd, 2006, 02:58 PM

My current mode of operation is to capture with HDVSplit and batch convert the clips with HDLink to Cineform codec. This gives you the option of doing things like converting a given clip to 720 60p for adding full slow motion range or converting to 30p and slowing down to 24p to slow down just a little.

Christopher Glaeser · June 22nd, 2006, 06:32 PM

Quote:

Originally Posted by Laurence Kingston

My current mode of operation is to capture with HDVSplit and batch convert the clips with HDLink to Cineform codec.

Why not use HDLink to capture, scene split, and convert?

Best,
Christopher

Laurence Kingston · June 22nd, 2006, 08:00 PM

Because then you end up with one long m2t file that is not scene split. If you archive this and convert it later with HDLink, you end up with one long unsplit Cineform file. If something happens later like Cineform goes and updates their codec to a much more CPU friendly version 3.0, if you have the individual m2t files you get with HDVSplit, you can reconvert them to the new codec without recapturing them from tape.

Laurence Kingston · June 22nd, 2006, 08:16 PM

Here's another reason I capture with HDVSplit than batch convert using HDLink: Time! If I capture and convert at the same time with HDLink, I end up spending a lot of time after the first tape has captured waiting for the conversion to complete. My aging 3.06 P4 seems to take forever to do this. Anyway, when it's finally done, I load another tape to capture and convert and it's another two hours or so before I can load the third tape. I find there's a lot less waiting around if I just capture the tapes in real time with HDVSplit, then batch convert the all the files overnight. Yeah it probably takes longer, but I don't have to worry about being there and timing when to change tapes or turn the camera off.

This could all be fixed in either of two different ways:

1/ split the m2t files into scene separated clips on capture.
2/ add the capability to timecode separate Cineform clips from a single large m2t clip.

Until one of these approaches is put into use, I'll continue to capture with HDVSplit and batch convert the scene split m2t files later.

Christopher Glaeser · June 22nd, 2006, 09:25 PM

Quote:

Originally Posted by Laurence Kingston

Because then you end up with one long m2t file that is not scene split.

HDLink can scene split both the m2t files and avi files during the capture.

Best,
Christopher

Laurence Kingston · June 22nd, 2006, 10:09 PM

Somebody else told me that recently. I haven't tried this for quite a few revisions so I must have missed this. It wouldn't surprise me. The folks at Cineform seem bound and determined to make the product we users want. I'll try again tomorrow.

Christopher Glaeser · June 22nd, 2006, 10:30 PM

Quote:

Originally Posted by Laurence Kingston

I haven't tried this for quite a few revisions so I must have missed this.

It works fine for me. I capture, scene detect, create m2t files, and convert to avi files, all in one pass. I have proposed a couple changes to the file number format which I think they plan to develop for a future release, but otherwise it works quite well, at least on a fast desktop.

Best,
Christopher

June 21st, 2006, 11:03 PM	#1
Laurence Kingston Major Player Join Date: Aug 2004 Location: Clermont, FL. Posts: 941	Capturing 1080 60i as 720 24p or 30p It seems to me that since you are losing vertical resolution anyway, and since the Sony HDV cameras interpolate some of their pixels, that converting to 1280x720 at 24p or 30p really isn't giving up much actual resolution that you haven't already lost in going to progressive. In any case, the resulting footage looks great and makes my 3.06 P4 with external firewire hard drives seem to absolutely fly. How much real world resolution am I actually giving up if I go this route?

June 21st, 2006, 11:23 PM	#2
David Newman CTO, CineForm Inc. Join Date: Jul 2003 Location: Cardiff-by-the-Sea, California Posts: 8,095	For spatial resolution you are losing very little. All you are losing is temporal resolution by going from 60 fields to 30 or 24 frames. Some don't consider that a lose at all, unless you shooting reality TV or sports which use the higher temporal resolution. __________________ David Newman -- web: www.gopro.com blog: cineform.blogspot.com -- twitter: twitter.com/David_Newman

June 22nd, 2006, 10:26 AM	#4
David Newman CTO, CineForm Inc. Join Date: Jul 2003 Location: Cardiff-by-the-Sea, California Posts: 8,095	yes, I'm in agreement. __________________ David Newman -- web: www.gopro.com blog: cineform.blogspot.com -- twitter: twitter.com/David_Newman

June 22nd, 2006, 09:40 AM	#3
Laurence Kingston Major Player Join Date: Aug 2004 Location: Clermont, FL. Posts: 941	Just to make sure we're on the same page, what I was asking is how much actual practical resolution you are losing when you downrez to 1280x720p as you go progressive, given that you are already losing about half your vertical resolution and at least some of the horizontal resolution was interpolated anyway.

June 22nd, 2006, 11:34 AM	#5
Laurence Kingston Major Player Join Date: Aug 2004 Location: Clermont, FL. Posts: 941	Boy is that cool! My current project has a lot of spontaneous zooms and pans, so it will look better in 1080 60i, but starting with my next project, I'll try to be a little more careful with the zooms and pans and edit either 24 or 30p. Another observation: a lot of new HDTV TVs upconvert SD DVDs and movies from blockbuster seem to look pretty darned good even on the larger screens. Movies shot with SD DV cameras don't get treated quite so well. The uprezzing algorythms ignore the interlaced video and the 60i gets deinterlaced and actually loses resolution. This has got me going on a progressive kick much more than before. None of my stuff is going to film and I've never worried so much about progressive vs interlaced because I actually preferred the extra temporal resolution of 60i. Viewing my stuff on a relatively inexpensive LCD HDTV has changed my mind however. 60i gets deinterlaced on the fly and not only does the vertical resolution drop, but it looses the temporal resolution anyway. It seems to me that being a little more careful about pans and zooms and converting to 30 or 24p 720 on capture will give me an image that will uprez rather than deinterlace and downrez on one of these 1280x720 HDTV sets that seem to be so common these days. By the way, does anyone know if an average 1280x720 HDTV will detect and uprez 30p properly. I know that mine at least does a pretty good job with 24p. I have a feeling though that when it sees 30p video it bypasses the uprez routines and deinterlaces just like it would with 60i. Aside from this issue I would rather work with 720 30p, but this has me worried.

June 22nd, 2006, 02:49 PM	#6
Thomas Smet Trustee Join Date: Mar 2004 Location: Milwaukee, WI Posts: 1,719	This does work great. The only thing to think about is that interlaced video is filtered so the resolution is actually a little bit lower than if the camera was to shoot progressive. For example on the SONY Z1 in the CF modes you do get around 540 lines which is about all one filed can give you. In interlaced mode however each field may only have around 405 lines of detail. So if you shot CF and down sized to 1280x720 you would actually get a little bit more detail than if you threw out a field from normal interlaced video. Of course shooting 60i does give you the option of converting to 1280x720x60p instead of just 30p or 24p for great smooth HD video. If you use a really good deinterlacer that can rebuild the missing lines then you may be able to get a lot more detail.

June 22nd, 2006, 02:58 PM	#7
Laurence Kingston Major Player Join Date: Aug 2004 Location: Clermont, FL. Posts: 941	My current mode of operation is to capture with HDVSplit and batch convert the clips with HDLink to Cineform codec. This gives you the option of doing things like converting a given clip to 720 60p for adding full slow motion range or converting to 30p and slowing down to 24p to slow down just a little.

June 22nd, 2006, 08:00 PM	#9
Laurence Kingston Major Player Join Date: Aug 2004 Location: Clermont, FL. Posts: 941	Because then you end up with one long m2t file that is not scene split. If you archive this and convert it later with HDLink, you end up with one long unsplit Cineform file. If something happens later like Cineform goes and updates their codec to a much more CPU friendly version 3.0, if you have the individual m2t files you get with HDVSplit, you can reconvert them to the new codec without recapturing them from tape.

June 22nd, 2006, 08:16 PM	#10
Laurence Kingston Major Player Join Date: Aug 2004 Location: Clermont, FL. Posts: 941	Here's another reason I capture with HDVSplit than batch convert using HDLink: Time! If I capture and convert at the same time with HDLink, I end up spending a lot of time after the first tape has captured waiting for the conversion to complete. My aging 3.06 P4 seems to take forever to do this. Anyway, when it's finally done, I load another tape to capture and convert and it's another two hours or so before I can load the third tape. I find there's a lot less waiting around if I just capture the tapes in real time with HDVSplit, then batch convert the all the files overnight. Yeah it probably takes longer, but I don't have to worry about being there and timing when to change tapes or turn the camera off. This could all be fixed in either of two different ways: 1/ split the m2t files into scene separated clips on capture. 2/ add the capability to timecode separate Cineform clips from a single large m2t clip. Until one of these approaches is put into use, I'll continue to capture with HDVSplit and batch convert the scene split m2t files later.

June 22nd, 2006, 10:09 PM	#12
Laurence Kingston Major Player Join Date: Aug 2004 Location: Clermont, FL. Posts: 941	Somebody else told me that recently. I haven't tried this for quite a few revisions so I must have missed this. It wouldn't surprise me. The folks at Cineform seem bound and determined to make the product we users want. I'll try again tomorrow.