how much better is HDV when its downrezed to SD??? at DVinfo.net

Nicholas Natteau · March 12th, 2005, 05:23 PM

I've read on many web sites, that even though you have to downrez from HDV to SD on the Sony Z1U or FX1 to get your footage into Final Cut Pro, you will still get a better picture quality from material that was originally HDV and downrezed to DV, than if you shot your footage in DV mode on the Sony Z1U.

I would very much appreciate feedback.

How much better is downrezed HDV footage to DV footage??? I've heard things said like: more colorspace, and downconverted source originally had more pixels to draw from.

Please respond!

- Nicholas

Xander Christ · March 12th, 2005, 06:21 PM

I just shot a short film with the FX-1. I downconverted the m2t files to DV for editing purposes. The downconverted footage looks like you can see more detail than a Canon XL1s, GL2, or Sony VX2100. Colorwise, the FX-1 looked just like any other 3-chip camera.

I used a DV codec for compression, so it was converted from HD's 4:2:0 colorspace, to DV's 601 4:1:1 colorspace. Did I notice that much of a difference? No. The final film is going to DVD, so I don't expect much of a color difference anyway.

However, for the visual effects sequences, I converted the m2t files to an uncompressed AVI file and downsampled that to 480i. Oh my gosh!!! The footage keyed out perfectly with very little adjustment to the key settings in After Effects (Keylight plug-in). So yes, I think the colorspace expanded to 4:2:2 or even 4:4:4 in the down conversion, but again, this is also dealing with HD's colorspace, not DV's because it's uncompressed RGB data.

The film is going to SDTV DVD anyway, but the FX-1 really doesn't look any different from other 3-chippers in SD color-space.

Jasenn Robertson · March 13th, 2005, 12:45 AM

Xander,

That is encouraging news about the keying. Can you explain what you used to do the conversion or bring the m2t files onto your computer? I hadn't thought about down-conversion after import. Thanks for sharing!

Peter Rixner · March 13th, 2005, 01:32 AM

Thats really interresting. I did a key test yesterday and it was horrible. Just what I expected from a mpeg file :(

BUT: I have to admit, that the setup was extremely bad. So maybe it wasnt really a good test.

Has someone also had good results with hdv keying ?

And why did you downsample in advance to the keying ? Your loosing so much resolution.

And where sould the new colorinformation come from ?

Maybe I'm totally missing something :)

Thanks

Peter

Xander Christ · March 13th, 2005, 03:14 AM

You can check out a keying test I did a few weeks ago on my website: www.pseudofilms.com/compare. This was just a real simple, fast test on a professionally lit greenscreen. I didn't spend too much time on it.

I'm a MainConcept junkie (www.mainconcept.com). I use their MPEG Pro 1.0.5 for Adobe Premiere Pro to import the m2t files. I then render the movie as uncompressed RGB output to my 1.6TB RAID. The movie file is now 1440x1080 uncompressed 8-bit RGB translated from the HD colorspace (I don't know the spec name for it - ITU-702R???).

I then use After Effects to resize and deinterlace the uncompressed AVI HDV footage from 1440x1080@60i to 720x480@30p where I compress it to DV using MainConcept's DV Codec. The original file has a 4:2:0 sampling, so that means the Y channel is 1440x1080 and the U and V channels are 720x540 (alternating), so resampling the entire image to 720x480 should render a higher color resolution than 4:2:0 or 4:1:1 because all the samples are the same resolution (something like that). But remember, the DV Codec renders the footage to 4:1:1 in a 601r colorspace (NTSC).

I use the "DV" footage to offline the HDV footage; then, I create an EDL and then import the m2t files with the DV's EDL. For the compositing sequences, I export those frames as uncompressed RGB and import the corresponding AVI file into After Effects for keying or what not. The higher resolution of HDV enables better mattes as the resolution is some 4x larger than SD.

The keying is done prior to the downsampling. When you downsample, the anti-aliasing helps smooth out the key. Since my distribution is DVD, my final output project is 720x480, so HD resolution really doesn't matter.

Wow, after rereading all that, it seems complicated, but it's really not. I just try to keep compression down to a minimum which is why I uncompress the footage and then edit off-line.

Peter Rixner · March 13th, 2005, 04:33 AM

Meanwhile I did a test with a very good greenscreen setup. And Your absolutely right. I used nearly the same way like you described, with the new Premiere 1.5.1 HDV codec and cineframe 25 (as I am from PAL-Country).

I also played around with Pictureprofile settings and I think that even improved it. But thats hard to say, as one couldnt really to a A/B comparison with one camera.

The HD keys is as good as I know from SD and even better than 16mm Film HD, which I recently had. But downsampling is really unbelievable.

All in all, I must COMPLETELY say that I was wrong with my test yesterday. It was only the bad setup of mine there.

I'm back from 99% to 110% Satisfied with Z1 :)

Greetings

Peter :)

Nicholas Natteau · March 13th, 2005, 07:31 AM

Please answer the question.

Keying should be posted as an entirely new thread.

That is not the question I asked. I wanted to know about downconverting from HDV to SD,

I would appreciate it if those of you posting replies in this thread answered my question or else start another thread about keying.

This was not helpful feedback.

- Nicholas

Douglas Spotted Eagle · March 13th, 2005, 09:02 AM

Nicholas, while the thread has taken a turn, I believe you've had two answers that offer what you've asked. It's unfortunate the thread turned into a keying thread; at the same time there is no need to become offended nor offensive about it.

You've asked a question that can be answered with a simple yes or no. It also requires some qualification, which is why (I believe this thread took a slight turn. You brought up colorspace and more pixels, which is what the other posts are alluding to.

Yes, footage shot as HDV and downsampled is superior (generally) than footage acquired as SD. There are LOTS of comparitive images on this website and others. HDV has more pixels from which to draw, and has a different colorspace offering the full 8 bit spectrum.

Nicholas Natteau · March 13th, 2005, 09:20 AM

Thanks for your feedback.

I'm really hoping to hear more in this thread about downconverting from HDV to DV.

I was not offended and I'm sorry if I offended. That was never my intention.

Cheers,

- Nicholas

Douglas Spotted Eagle · March 13th, 2005, 09:22 AM

Specifically, what did you want to know, other than wondering if HDV downsampled to SD is better than acquiring in SD?

Nicholas Natteau · March 13th, 2005, 09:54 AM

OK, has anyone posted HDV downconverted images side by side with DV images? Is there a noticeable difference? Or is the difference just in the numbers and too marginal to be noticeable?

You mentioned that HDV employs the full 8-bit color...is DV less than that. Sorry, I'm so new to all this.

Thank you in advance,

- Nicholas

Douglas Spotted Eagle · March 13th, 2005, 10:00 AM

If you visit the link below my name, and click on the HDV portal, you'll see several comparison images in several compression formats. It's DEFINITELY noticable. Once FCP gets their act together on HDV, it should be a good downsample in the app, too. Currently, only Sony Vegas downsamples better than the camera, but all indications point to FCP being equal to the task. Otherwise, you'll want to downsample from the camera, which is also quite good.

HD (all HD) is in the ITU709 colorspace. This is compatible with sRGB as well, so it's web-friendly among other things.

Luis Caffesse · March 13th, 2005, 10:55 AM

Douglas,

You mentioned that HDV "has a different colorspace offering the full 8 bit spectrum."

Does DV not offer the 'full 8 bit spectrum'?

I've heard this mentioned before, but never with a clear explanation. I was under the impression that DV was an 8bit medium...is it not using the full 8 bits?

Thanks for any clarification.

-Luis

Douglas Spotted Eagle · March 13th, 2005, 04:34 PM

I'm on a short schedule, but there is a paper to be found at http://visarts.ucsd.edu/grad/lab/dv-craigb.html that should explain it, albeit a long read. It's not as big a deal as one would think, but it is something to be aware of.

Soeren Mueller · March 13th, 2005, 04:44 PM

I thought that ITU709 was the colorspace for analog transmission of HD signals!?

MiniDV is 8bit (of course not full 8bit for every pixel ;o) as is HDV - though the compression of color information is slightly different but it's there with HDV too. None of the both formats offers much superior color information over the other!
And why HD should be web-friendly in contrast to MiniDV is something only Douglas seems to know. ;-)

Cheers,
-Sören

Additional info: I'm from PAL country... PAL MiniDV color compression is 4:2:0 here too - the same as with HDV

March 12th, 2005, 05:23 PM	#1
Nicholas Natteau Regular Crew Join Date: Feb 2005 Location: Boston, MA Posts: 70	how much better is HDV when its downrezed to SD??? I've read on many web sites, that even though you have to downrez from HDV to SD on the Sony Z1U or FX1 to get your footage into Final Cut Pro, you will still get a better picture quality from material that was originally HDV and downrezed to DV, than if you shot your footage in DV mode on the Sony Z1U. I would very much appreciate feedback. How much better is downrezed HDV footage to DV footage??? I've heard things said like: more colorspace, and downconverted source originally had more pixels to draw from. Please respond! - Nicholas

March 13th, 2005, 12:45 AM	#3
Jasenn Robertson New Boot Join Date: Jan 2005 Location: Alexandria, VA Posts: 19	Xander, That is encouraging news about the keying. Can you explain what you used to do the conversion or bring the m2t files onto your computer? I hadn't thought about down-conversion after import. Thanks for sharing! __________________ Jasenn Robertson, Editor Alexandria, VA

March 13th, 2005, 09:02 AM	#8
Douglas Spotted Eagle VASST Join Date: Jan 2004 Location: Stockton, UT Posts: 5,648	Nicholas, while the thread has taken a turn, I believe you've had two answers that offer what you've asked. It's unfortunate the thread turned into a keying thread; at the same time there is no need to become offended nor offensive about it. You've asked a question that can be answered with a simple yes or no. It also requires some qualification, which is why (I believe this thread took a slight turn. You brought up colorspace and more pixels, which is what the other posts are alluding to. Yes, footage shot as HDV and downsampled is superior (generally) than footage acquired as SD. There are LOTS of comparitive images on this website and others. HDV has more pixels from which to draw, and has a different colorspace offering the full 8 bit spectrum. __________________ Douglas Spotted Eagle/Spot Author, producer, composer Certified Sony Vegas Trainer http://www.vasst.com

March 13th, 2005, 09:22 AM	#10
Douglas Spotted Eagle VASST Join Date: Jan 2004 Location: Stockton, UT Posts: 5,648	Specifically, what did you want to know, other than wondering if HDV downsampled to SD is better than acquiring in SD? __________________ Douglas Spotted Eagle/Spot Author, producer, composer Certified Sony Vegas Trainer http://www.vasst.com

March 13th, 2005, 10:00 AM	#12
Douglas Spotted Eagle VASST Join Date: Jan 2004 Location: Stockton, UT Posts: 5,648	If you visit the link below my name, and click on the HDV portal, you'll see several comparison images in several compression formats. It's DEFINITELY noticable. Once FCP gets their act together on HDV, it should be a good downsample in the app, too. Currently, only Sony Vegas downsamples better than the camera, but all indications point to FCP being equal to the task. Otherwise, you'll want to downsample from the camera, which is also quite good. HD (all HD) is in the ITU709 colorspace. This is compatible with sRGB as well, so it's web-friendly among other things. __________________ Douglas Spotted Eagle/Spot Author, producer, composer Certified Sony Vegas Trainer http://www.vasst.com

March 12th, 2005, 06:21 PM	#2
Xander Christ Regular Crew Join Date: Jul 2004 Location: Minneapolis, MN Posts: 115	I just shot a short film with the FX-1. I downconverted the m2t files to DV for editing purposes. The downconverted footage looks like you can see more detail than a Canon XL1s, GL2, or Sony VX2100. Colorwise, the FX-1 looked just like any other 3-chip camera. I used a DV codec for compression, so it was converted from HD's 4:2:0 colorspace, to DV's 601 4:1:1 colorspace. Did I notice that much of a difference? No. The final film is going to DVD, so I don't expect much of a color difference anyway. However, for the visual effects sequences, I converted the m2t files to an uncompressed AVI file and downsampled that to 480i. Oh my gosh!!! The footage keyed out perfectly with very little adjustment to the key settings in After Effects (Keylight plug-in). So yes, I think the colorspace expanded to 4:2:2 or even 4:4:4 in the down conversion, but again, this is also dealing with HD's colorspace, not DV's because it's uncompressed RGB data. The film is going to SDTV DVD anyway, but the FX-1 really doesn't look any different from other 3-chippers in SD color-space.

March 13th, 2005, 01:32 AM	#4
Peter Rixner Regular Crew Join Date: Mar 2005 Location: near Munich, Bavaria Posts: 139	Thats really interresting. I did a key test yesterday and it was horrible. Just what I expected from a mpeg file :( BUT: I have to admit, that the setup was extremely bad. So maybe it wasnt really a good test. Has someone also had good results with hdv keying ? And why did you downsample in advance to the keying ? Your loosing so much resolution. And where sould the new colorinformation come from ? Maybe I'm totally missing something :) Thanks Peter

March 13th, 2005, 03:14 AM	#5
Xander Christ Regular Crew Join Date: Jul 2004 Location: Minneapolis, MN Posts: 115	You can check out a keying test I did a few weeks ago on my website: www.pseudofilms.com/compare. This was just a real simple, fast test on a professionally lit greenscreen. I didn't spend too much time on it. I'm a MainConcept junkie (www.mainconcept.com). I use their MPEG Pro 1.0.5 for Adobe Premiere Pro to import the m2t files. I then render the movie as uncompressed RGB output to my 1.6TB RAID. The movie file is now 1440x1080 uncompressed 8-bit RGB translated from the HD colorspace (I don't know the spec name for it - ITU-702R???). I then use After Effects to resize and deinterlace the uncompressed AVI HDV footage from 1440x1080@60i to 720x480@30p where I compress it to DV using MainConcept's DV Codec. The original file has a 4:2:0 sampling, so that means the Y channel is 1440x1080 and the U and V channels are 720x540 (alternating), so resampling the entire image to 720x480 should render a higher color resolution than 4:2:0 or 4:1:1 because all the samples are the same resolution (something like that). But remember, the DV Codec renders the footage to 4:1:1 in a 601r colorspace (NTSC). I use the "DV" footage to offline the HDV footage; then, I create an EDL and then import the m2t files with the DV's EDL. For the compositing sequences, I export those frames as uncompressed RGB and import the corresponding AVI file into After Effects for keying or what not. The higher resolution of HDV enables better mattes as the resolution is some 4x larger than SD. The keying is done prior to the downsampling. When you downsample, the anti-aliasing helps smooth out the key. Since my distribution is DVD, my final output project is 720x480, so HD resolution really doesn't matter. Wow, after rereading all that, it seems complicated, but it's really not. I just try to keep compression down to a minimum which is why I uncompress the footage and then edit off-line.

March 13th, 2005, 04:33 AM	#6
Peter Rixner Regular Crew Join Date: Mar 2005 Location: near Munich, Bavaria Posts: 139	Meanwhile I did a test with a very good greenscreen setup. And Your absolutely right. I used nearly the same way like you described, with the new Premiere 1.5.1 HDV codec and cineframe 25 (as I am from PAL-Country). I also played around with Pictureprofile settings and I think that even improved it. But thats hard to say, as one couldnt really to a A/B comparison with one camera. The HD keys is as good as I know from SD and even better than 16mm Film HD, which I recently had. But downsampling is really unbelievable. All in all, I must COMPLETELY say that I was wrong with my test yesterday. It was only the bad setup of mine there. I'm back from 99% to 110% Satisfied with Z1 :) Greetings Peter :)

March 13th, 2005, 07:31 AM	#7
Nicholas Natteau Regular Crew Join Date: Feb 2005 Location: Boston, MA Posts: 70	Please answer the question. Keying should be posted as an entirely new thread. That is not the question I asked. I wanted to know about downconverting from HDV to SD, I would appreciate it if those of you posting replies in this thread answered my question or else start another thread about keying. This was not helpful feedback. - Nicholas

March 13th, 2005, 09:20 AM	#9
Nicholas Natteau Regular Crew Join Date: Feb 2005 Location: Boston, MA Posts: 70	Thanks for your feedback. I'm really hoping to hear more in this thread about downconverting from HDV to DV. I was not offended and I'm sorry if I offended. That was never my intention. Cheers, - Nicholas

March 13th, 2005, 09:54 AM	#11
Nicholas Natteau Regular Crew Join Date: Feb 2005 Location: Boston, MA Posts: 70	OK, has anyone posted HDV downconverted images side by side with DV images? Is there a noticeable difference? Or is the difference just in the numbers and too marginal to be noticeable? You mentioned that HDV employs the full 8-bit color...is DV less than that. Sorry, I'm so new to all this. Thank you in advance, - Nicholas

March 13th, 2005, 10:55 AM	#13
Luis Caffesse Major Player Join Date: Apr 2004 Location: Austin, Texas Posts: 704	Douglas, You mentioned that HDV "has a different colorspace offering the full 8 bit spectrum." Does DV not offer the 'full 8 bit spectrum'? I've heard this mentioned before, but never with a clear explanation. I was under the impression that DV was an 8bit medium...is it not using the full 8 bits? Thanks for any clarification. -Luis __________________ Luis Caffesse Pitch Productions Austin, Texas

March 13th, 2005, 04:34 PM	#14
Douglas Spotted Eagle VASST Join Date: Jan 2004 Location: Stockton, UT Posts: 5,648	I'm on a short schedule, but there is a paper to be found at http://visarts.ucsd.edu/grad/lab/dv-craigb.html that should explain it, albeit a long read. It's not as big a deal as one would think, but it is something to be aware of. __________________ Douglas Spotted Eagle/Spot Author, producer, composer Certified Sony Vegas Trainer http://www.vasst.com

March 13th, 2005, 04:44 PM	#15
Soeren Mueller Regular Crew Join Date: Mar 2004 Location: Düsseldorf, Germany / Denver, CO Posts: 137	I thought that ITU709 was the colorspace for analog transmission of HD signals!? MiniDV is 8bit (of course not full 8bit for every pixel ;o) as is HDV - though the compression of color information is slightly different but it's there with HDV too. None of the both formats offers much superior color information over the other! And why HD should be web-friendly in contrast to MiniDV is something only Douglas seems to know. ;-) Cheers, -Sören Additional info: I'm from PAL country... PAL MiniDV color compression is 4:2:0 here too - the same as with HDV