JVC HD100 FAQ -- includes addressing split-screen at DVinfo.net

Barry Green · October 27th, 2005, 04:30 PM

JVC has posted a FAQ section about ProHD, HDV, and the HD100 on their website, and it includes a mention of split-screen. For those who've been asking for a statement from JVC, this would appear to be it.

http://pro.jvc.com/prof/Attributes/f...ture_id=13#sse

Stephen L. Noe · October 27th, 2005, 05:19 PM

Great find and it supports alot of our findings.

Paticularly the part about editing in HDV resolution is far better than converting to SD for the intent of DVD delivery. 420 to 420 is far better than 420 to 411 to 420. Another thing I liked about what they've written is that ProHD is HDTV. The notion that ProHD and HDV is not a pro codec is hogwash. There is not much (if any) difference between m2t and HDTV ts.

really good find Barry....

Diogo Athouguia · October 27th, 2005, 05:33 PM

I love the title... "RARELY, a faint vertical..."

Graeme Nattress · October 27th, 2005, 05:41 PM

However, there's a fair bit of propoganda in there, mostly against the Z1, which is rather unfair, and also against DVCproHD, which is rather unfair:

"Finally, a DCT based compression is applied to reduce the amount of data so that it conforms to the requirements of the storage media. Some DCT compression systems are inherently more efficient than others. MPEG2, pioneered by JVC and utilized in the GY-HD100U, is generally considered to be about 5 times as efficient as conventional frame bound systems because the compression is applied to frame groups rather than single frames."

5 times efficiency is rather an exaggeration, IMHO. I tried to find some white papers on MPEG2 efficiency a while back, and didn't find much, but in speaking to other experts, the effeciency over a traditional DCT codecs be in the region of 2.5 at best, and certainly no-where near 5 times. Perhaps if you put a totally still frame in there, you might approach that level of efficency, but no way would you get it in real life.

Similarly to criticize the Z1 for being 960x1080 chips is a bit rich when the Z1 delivers equally as good pictures as the JVC.

"Compressing progressive signals is more efficient than compressing interlace signals. At any given compression ratio, progressive images can be reconstructed more faithfully and with fewer artifacts. That's one reason all high end digital cinematography systems utilize progressive scan."

No, they use progressive scan because interlace is for TV, not the big screen - 24p is "progressive" - I mean!

"If a DV dub is made from an HDV recording, the extra resolution is lost, and the chroma subsampling effectively becomes 4:1:0. This is less than optimal for DV which records at 4:1:1 (in the U.S.), and less than optimal for DVD which utilizes MPEG2 at 4:2:0. "

Well, no, not if the downconversion is handled correctly. This would only happen if you did:

HDV 4:2:0 -> SD 4:2:0 -> DV, leading to 4:1:0

whereas I'm sure Sony do

HDV 4:2:0 -> DV 4:1:1, giving a great 4:1:1

Come on JVC - credit us with a little intelligence....

Graeme

Michael Maier · October 27th, 2005, 06:49 PM

Are you going to tell me Sony, Canon and Panasonic don't do the same? It's just marketing.

Graeme Nattress · October 27th, 2005, 07:17 PM

And if Sony said something like that, I'd challenge it to! Same goes to Panasonic or Canon. I'm sick of video mis-information masquerading as engineering fact. Marketing it may be, but that doesn't mean I have to like it.

I've just spent the last weekend with the JVC camera and I shot some really nice stuff with it, and I'm just in the process of editing it into a little short. The 30p mode worked very well as long as you took care with it, and when you lighted the scene, the image came alive, just like it should. There were things it didn't handle well, and others it did a great job with. There are limitations and compromises with the camera, but for goodness sake, it's blinking cheap for what it offers, and if you don't accept such compromises, you'd better be rather rich when looking for your next camera. That said, I don't accept JVC trying to sell their camera on bogus engineering faqs - they should sell it on it's picture quality in a well lighted controlled environment, where it produces great results, and indeed, excellent when you consider the price.

Graeme

Steve Mullen · October 27th, 2005, 07:21 PM

Quote:

Originally Posted by Graeme Nattress

They should sell it on it's picture quality in a well lighted controlled environment, where it produces great results, and indeed, excellent when you consider the price.

Graeme

I couldn't agree more which is why they are writing a Guide and I'm at work on a book.

But folks were demanding a public "response" from JVC so now they have one.

Graeme Nattress · October 27th, 2005, 07:29 PM

I'm just finishing up editing a 720p30 short I shot with the JVC HD100 over last weekend. I had a fun time, if a frustrating one at times, with the camera. But the biggest frustration was the lack of NLE support (particularly FCP for the 24p - 30p worked great hence my short is in 30p), not the camera. The camera is not perfect (what is?), but it makes up for it by being rather cheap considering. What is needed is good advise to operate the camera within it's best quality working range, so that you're unlikely to ever see the nasties. If you do that, you'll get great images.

Graeme

Jiri Bakala · October 27th, 2005, 07:29 PM

Quote:

Originally Posted by Diogo Athouguia

I love the title... "RARELY, a faint vertical..."

My favourite is: "...and the chips would self-destruct."
:-)

Nate Weaver · October 27th, 2005, 07:38 PM

"Doc, everytime I do this, my knee hurts"

...and the doctor replies: "Well, don't do that!"

Michael Maier · October 28th, 2005, 12:37 AM

Quote:

Originally Posted by Graeme Nattress

There were things it didn't handle well

Like what?

Steve Mullen · October 28th, 2005, 03:34 AM

Quote:

Originally Posted by Graeme Nattress

What is needed is good advise to operate the camera within it's best quality working range, so that you're unlikely to ever see the nasties.Graeme

Have you considered writing a filter that either automatically or via manual control reduces/eliminates SSE?

This would seem ideal for those times when you get SSE.

Werner Wesp · October 28th, 2005, 05:05 AM

Quote:

Originally Posted by Graeme Nattress

However, there's a fair bit of propoganda in there, mostly against the Z1, which is rather unfair, and also against DVCproHD, which is rather unfair:

"Finally, a DCT based compression is applied to reduce the amount of data so that it conforms to the requirements of the storage media. Some DCT compression systems are inherently more efficient than others. MPEG2, pioneered by JVC and utilized in the GY-HD100U, is generally considered to be about 5 times as efficient as conventional frame bound systems because the compression is applied to frame groups rather than single frames."

5 times efficiency is rather an exaggeration, IMHO. I tried to find some white papers on MPEG2 efficiency a while back, and didn't find much, but in speaking to other experts, the effeciency over a traditional DCT codecs be in the region of 2.5 at best, and certainly no-where near 5 times. Perhaps if you put a totally still frame in there, you might approach that level of efficency, but no way would you get it in real life.

Well, I'm one too (a compression expert. I have to say 5 times is about right. You have to look at what they're doing with wath kind of bandwith. 6frame gops, with 4 P-frames each that take second to none of datarate: one I frame, one B-frame....

that's roughly 6/1.5 (that is 4) times the efficiency of intraframcompression...

Quote:

Originally Posted by Graeme Nattress

Similarly to criticize the Z1 for being 960x1080 chips is a bit rich when the Z1 delivers equally as good pictures as the JVC.

"Compressing progressive signals is more efficient than compressing interlace signals. At any given compression ratio, progressive images can be reconstructed more faithfully and with fewer artifacts. That's one reason all high end digital cinematography systems utilize progressive scan."

No, they use progressive scan because interlace is for TV, not the big screen - 24p is "progressive" - I mean!.

Well, I think it really IS a selling point to emphasize the fact that it is a true-full-resolution-camcorder. It really has its benefits in terms of quality....

Quote:

Originally Posted by Graeme Nattress

"If a DV dub is made from an HDV recording, the extra resolution is lost, and the chroma subsampling effectively becomes 4:1:0. This is less than optimal for DV which records at 4:1:1 (in the U.S.), and less than optimal for DVD which utilizes MPEG2 at 4:2:0. "

Well, no, not if the downconversion is handled correctly. This would only happen if you did:

HDV 4:2:0 -> SD 4:2:0 -> DV, leading to 4:1:0

whereas I'm sure Sony do

HDV 4:2:0 -> DV 4:1:1, giving a great 4:1:1

Come on JVC - credit us with a little intelligence....

Graeme

No no, there is something to say about this. You actually have a lot more data in ProHD-mode, it's just compressed better, so it "seems" about the same. I think aquisition in ProHD would be better then DV25...

Graeme Nattress · October 28th, 2005, 06:58 AM

Quote:

Originally Posted by Werner Wesp

Well, I'm one too (a compression expert. I have to say 5 times is about right. You have to look at what they're doing with wath kind of bandwith. 6frame gops, with 4 P-frames each that take second to none of datarate: one I frame, one B-frame....

that's roughly 6/1.5 (that is 4) times the efficiency of intraframcompression...

I'd love to see you point to some kind of white paper on this, because it's a question I've asked again and again, and there's very little out there.

The P frames will only be small on limited movement though, and given the artifacts I can see on the JVC, I can assure you the compression does not look as good as 5 times as efficient as plain DCT. Also, if that's JVC's arguments, that a 6 frame GOP can be that efficient, why don't they say that Sony's 15 frame GOP is more efficient still, or are they saying that because it's interlaced they NEED a 15 frame GOP to get the efficiency of their 6 frame progressive GOP.

Argh. Now you've got me started. Discussing relative compression efficiencies just leads to arguments and war. The fact is that JVC have pulled a number out of thin air that doesn't fit right with me. There's no justification for that number, and really, they're trying to "prove with numbers" - a fruitless task at best, that their HDV is better than anyone elses and better than DVCproHD at the same time.

In JVC's defense - they've edited the document and now say "is generally considered to be UP TO 5 times as efficient as conventional frame bound systems", whereas last night they were saying "is generally considered to be ABOUT 5 times as efficient as conventional frame bound systems". Sure, MPEG2 can be, at maximum, on perfectly still images, about 5 times as efficient. If you're shooting still images though, buy a Canon 1dsMKII and be done with it though..... Then, of course, they apply their MAX possible figure as if it were some kind of average. That's cheating.

That's a nice bit of revision there JVC.....

Werner Wesp · October 28th, 2005, 08:33 AM

Well, the longer the GOPs, the more 'efficient', that much is true. But efficiency of a codec is a mathematical concept (on which I've written my master thesis - or how do you translate that in English) and something around 5 times more efficiency isn't hard to believe in the aspect of interframes vs. intraframe. It's all about redundant information. Information that's already there, so there's no need to write them to tape twice...

The longer the GOP, the more efficient, but with quite large implications for visual quality. Very simple: The best picture has the worst compression: all I-frames. The worst picture has just one I-frame at the beginning of a complete movie. But somewhere in between is the best trade-off for optimum quality (and I think 6-frame GOPs seems to be the good choice). 15-frame-GOPs are even better if there is virtually no motion, and non-fixed-length GOPs would be the best for all situations...

October 27th, 2005, 04:30 PM	#1
Barry Green Barry Wan Kenobi Join Date: Jul 2003 Location: North Carolina Posts: 3,863	JVC HD100 FAQ -- includes addressing split-screen JVC has posted a FAQ section about ProHD, HDV, and the HD100 on their website, and it includes a mention of split-screen. For those who've been asking for a statement from JVC, this would appear to be it. http://pro.jvc.com/prof/Attributes/f...ture_id=13#sse

October 27th, 2005, 05:41 PM	#4
Graeme Nattress RED Problem Solver Join Date: Sep 2003 Location: Ottawa, Canada Posts: 1,365	However, there's a fair bit of propoganda in there, mostly against the Z1, which is rather unfair, and also against DVCproHD, which is rather unfair: "Finally, a DCT based compression is applied to reduce the amount of data so that it conforms to the requirements of the storage media. Some DCT compression systems are inherently more efficient than others. MPEG2, pioneered by JVC and utilized in the GY-HD100U, is generally considered to be about 5 times as efficient as conventional frame bound systems because the compression is applied to frame groups rather than single frames." 5 times efficiency is rather an exaggeration, IMHO. I tried to find some white papers on MPEG2 efficiency a while back, and didn't find much, but in speaking to other experts, the effeciency over a traditional DCT codecs be in the region of 2.5 at best, and certainly no-where near 5 times. Perhaps if you put a totally still frame in there, you might approach that level of efficency, but no way would you get it in real life. Similarly to criticize the Z1 for being 960x1080 chips is a bit rich when the Z1 delivers equally as good pictures as the JVC. "Compressing progressive signals is more efficient than compressing interlace signals. At any given compression ratio, progressive images can be reconstructed more faithfully and with fewer artifacts. That's one reason all high end digital cinematography systems utilize progressive scan." No, they use progressive scan because interlace is for TV, not the big screen - 24p is "progressive" - I mean! "If a DV dub is made from an HDV recording, the extra resolution is lost, and the chroma subsampling effectively becomes 4:1:0. This is less than optimal for DV which records at 4:1:1 (in the U.S.), and less than optimal for DVD which utilizes MPEG2 at 4:2:0. " Well, no, not if the downconversion is handled correctly. This would only happen if you did: HDV 4:2:0 -> SD 4:2:0 -> DV, leading to 4:1:0 whereas I'm sure Sony do HDV 4:2:0 -> DV 4:1:1, giving a great 4:1:1 Come on JVC - credit us with a little intelligence.... Graeme __________________ www.nattress.com - filters for FCP

October 27th, 2005, 07:17 PM	#6
Graeme Nattress RED Problem Solver Join Date: Sep 2003 Location: Ottawa, Canada Posts: 1,365	And if Sony said something like that, I'd challenge it to! Same goes to Panasonic or Canon. I'm sick of video mis-information masquerading as engineering fact. Marketing it may be, but that doesn't mean I have to like it. I've just spent the last weekend with the JVC camera and I shot some really nice stuff with it, and I'm just in the process of editing it into a little short. The 30p mode worked very well as long as you took care with it, and when you lighted the scene, the image came alive, just like it should. There were things it didn't handle well, and others it did a great job with. There are limitations and compromises with the camera, but for goodness sake, it's blinking cheap for what it offers, and if you don't accept such compromises, you'd better be rather rich when looking for your next camera. That said, I don't accept JVC trying to sell their camera on bogus engineering faqs - they should sell it on it's picture quality in a well lighted controlled environment, where it produces great results, and indeed, excellent when you consider the price. Graeme __________________ www.nattress.com - filters for FCP

October 27th, 2005, 07:29 PM	#8
Graeme Nattress RED Problem Solver Join Date: Sep 2003 Location: Ottawa, Canada Posts: 1,365	I'm just finishing up editing a 720p30 short I shot with the JVC HD100 over last weekend. I had a fun time, if a frustrating one at times, with the camera. But the biggest frustration was the lack of NLE support (particularly FCP for the 24p - 30p worked great hence my short is in 30p), not the camera. The camera is not perfect (what is?), but it makes up for it by being rather cheap considering. What is needed is good advise to operate the camera within it's best quality working range, so that you're unlikely to ever see the nasties. If you do that, you'll get great images. Graeme __________________ www.nattress.com - filters for FCP

October 27th, 2005, 07:38 PM	#10
Nate Weaver Wrangler Join Date: Oct 2002 Location: Los Angeles, CA Posts: 2,100	"Doc, everytime I do this, my knee hurts" ...and the doctor replies: "Well, don't do that!" __________________ My Work: nateweaver.net

October 27th, 2005, 05:19 PM	#2
Stephen L. Noe Trustee Join Date: Apr 2005 Location: Chicago Posts: 1,214	Great find and it supports alot of our findings. Paticularly the part about editing in HDV resolution is far better than converting to SD for the intent of DVD delivery. 420 to 420 is far better than 420 to 411 to 420. Another thing I liked about what they've written is that ProHD is HDTV. The notion that ProHD and HDV is not a pro codec is hogwash. There is not much (if any) difference between m2t and HDTV ts. really good find Barry....

October 27th, 2005, 05:33 PM	#3
Diogo Athouguia Major Player Join Date: May 2005 Location: Portugal Posts: 282	I love the title... "RARELY, a faint vertical..."

October 27th, 2005, 06:49 PM	#5
Michael Maier Trustee Join Date: Apr 2005 Posts: 1,269	Are you going to tell me Sony, Canon and Panasonic don't do the same? It's just marketing.

October 28th, 2005, 08:33 AM	#15
Werner Wesp Major Player Join Date: Aug 2005 Location: Belgium Posts: 497	Well, the longer the GOPs, the more 'efficient', that much is true. But efficiency of a codec is a mathematical concept (on which I've written my master thesis - or how do you translate that in English) and something around 5 times more efficiency isn't hard to believe in the aspect of interframes vs. intraframe. It's all about redundant information. Information that's already there, so there's no need to write them to tape twice... The longer the GOP, the more efficient, but with quite large implications for visual quality. Very simple: The best picture has the worst compression: all I-frames. The worst picture has just one I-frame at the beginning of a complete movie. But somewhere in between is the best trade-off for optimum quality (and I think 6-frame GOPs seems to be the good choice). 15-frame-GOPs are even better if there is virtually no motion, and non-fixed-length GOPs would be the best for all situations...