Wow, Jan, you sure are a devoted Panasonic employee. Sony needs someone like you too. :)

They *all* need someone like her!

;-)

<<<-- Originally posted by Chris Hurd : They *all* need someone like her!

;-) -->>>

If the product wasn't so cool, it wouldn't be nearly so much fun!

Thanks guys!

Jan

sergio,

this is exellent!!!!

the more I await and long for a new piece of equipment, the more I realize the time I'm wasting by not filming with what I have!

filip

<<<-- Originally posted by Toke Lahti :Acquisition and distribution codecs are whole diffrent ballgame. Compression should be done only to final product after all effecting and color correcting. Otherwise you just loose quality with any lossy compression.-->>>

compression is always done during acquisition... dv is 5:1 compression with a severely reduced color space, for instance... how many video cameras have you seen that record uncompressed video??

so the issue here isn't the bitrate, it's the efficiency of the codec that counts.

there are things like wavelet compression, that offer far greater advantages than the standard dct-based compression that we have been forced to live with all these years.

the reason that these video camera companies keep using vastly inferior codecs is so that they can get away with charging obscene amounts of money for the hardware that it takes to handle the huge bitrates that these lousy codecs generate.

i've seen jan around the 'net for many years, and i'm not slamming her personally for the decisions that her company makes... all i'm trying to do is to get people to ask companies like panasonic why they refuse to embrace new codec technologies.

<<<-- Originally posted by Dan Euritt : compression is always done during acquisition... dv is 5:1 compression with a severely reduced color space, for instance... how many video cameras have you seen that record uncompressed video?? >>

There are very few cameras that record. In fact all cameras are uncompress. It is the recorders that do the compression. Life is full of little trade offs and frankly with each trade off comes a price. So uncompressed, will cost you vastly more than compressed.

<< so the issue here isn't the bitrate, it's the efficiency of the codec that counts. >>

I don't think you can show me a codec that can be a non-proprietary, meaning anyone can access it and make it happen that is more effiecient that the DV Codec.

<< there are things like wavelet compression, that offer far greater advantages than the standard dct-based compression that we have been forced to live with all these years. >>

Wavelet is in fact older than DV. It was used in the IMX edit system, which was purchased and remade into the Stratosphere. It is/was no more efficient and its payload is heavy for the benefit. It still had compression artifacts.

<< the reason that these video camera companies keep using vastly inferior codecs is so that they can get away with charging obscene amounts of money for the hardware that it takes to handle the huge bitrates that these lousy codecs generate. >>

What you may not understand is that there is a difference between a delivery codec and an acquistion codec, which is what Toke was alluding to. In an acqusition codec it is more desireable to have each frame stand on its own, than to have them all mushed together in a large GOP like what happens in MPEG. For delivery, sure there are some really awesome codecs, for production, don't see it. Not only do you have to make machines that record but there has to be enough of a promise that the NLE guys will produce an edit system that will handle it.

<< all i'm trying to do is to get people to ask companies like panasonic why they refuse to embrace new codec technologies. >>

Like what, HDV? MPEG2? MPEG4? take a look at our consumer line up of cameras. We may have something in the very high end that will look at the huge payload that a 2K or 4K signal would need, and that is only an R & D, we're taking a look maybe; but why would should we want to walk away from DVCPROHD. It can do 24P, variable frame rate, with edit systems that support that notion today, in fact yesterday and even last year. Why would we walk away from that? It is still one of the most efficient algorithms on the planet for production. If you want to take it back to an uncompressed domain you can, but the data rate is in the native codec very viable. Granted is isn't 25Mbs, but it isn't cramped by a 15 frame GOP, compressed audio, and a color sample of 4:2:0. That is more appropriate for distribution than it is for production.

How about our DVD players, we use incredible codecs there, but they are inappropriate for production. Frankly there is little you could say that would convince me to produce an important documentary on a long GOP format. Just wouldn't do it. Distribute on it. You bet.

Better to look at the compression algorithms in the light of what they for which they are designed to be used before you make a judgement about what is and isn't the right thing to do.

My .02,

Jan

Maybe I've missed it in my skim over this thread but is this new camera everyone is talking about one to compete with the likes of the "prosumer" HDV cams. (i.e in the same ballpark range) with p2 storage?

I was disappointed with Canon's offering ala the Xl2, and I just shudder at the idea of HDV and have yet to be convinced it's a worth while format (For me, for me). For the last several months I've been thinking to myself "Panasonic must come out with something. I mean they blew us away with the DVX, they're sure to do it again"

Aaron

<<<-- Originally posted by Filip Kovcin : sergio,

this is exellent!!!!



filip -->>>

Thanks, Filip! But, back to topic, I'm really awaiting with antecipation for this new Panasonic camera!...wOoops... :)

Panasonic is going to sink the rest of them when this comes out. Think about it. They played it smart getting in bed with Macintosh and FCP HD. They've got an easy affordable workflow setup from shooting to a finished product.

Face the music. Panasonic did it right. We are in debt to them as fimmakers, guys. June will show it.

<<<-- Originally posted by Jan Crittenden : <<<-- Originally posted by Dan Euritt : compression is always done during acquisition... dv is 5:1 compression with a severely reduced color space, for instance... how many video cameras have you seen that record uncompressed video?? >>

They are making these uncompressed cameras right here!
http://www.dvinfo.net/conf/showthread.php?s=&threadid=25808
(Only 166 pages for now!)
<<< so the issue here isn't the bitrate, it's the efficiency of the codec that counts. >>
I don't think you can show me a codec that can be a non-proprietary, meaning anyone can access it and make it happen that is more effiecient that the DV Codec.
<< all i'm trying to do is to get people to ask companies like panasonic why they refuse to embrace new codec technologies. >>
Like what, HDV? MPEG2? MPEG4?>

Dan, of course it's a matter of both bitrate and codec.
Speaking of reduced color space, I don't understand why we are not getting more depth than 8bits...

Mjpeg2000 might be a good acquistion codec.
Jan, if you are interested, there is Florin Popescu discussing in the thread above, who works at Fraunhofer, which has been developing mjpeg2000.
If I remember correctly Arri's D20 uses it.
There might also be something more advanced than dv codec in mpeg4's studio profiles?
You can't claim that there has been no progress at all in acquistion codecs for last decade, can you?

<How about our DVD players, we use incredible codecs there...>

I wouldn't call dvd's mpeg2 incredible. Maybe it was a big improvement a decade ago, but it always hurts my eyes after seeing a digibeta or dv master of a program and then looking the same on dvd.

ps. Jan, about YLE (FinnishBC) replacing their vcr's, I don't know who's in charge over there. It's a huge place, I've been working there only as a freelancer and heard about it as a rumour. If you want to contact them, maybe you should do it through your official channels. Salespeople at PannyEurope knows who they are selling to.

<<<-- Originally posted by Jan Crittenden : I don't think you can show me a codec that can be a non-proprietary, meaning anyone can access it and make it happen that is more effiecient that the DV Codec. -->>>

didn't panasonic develop dvcpro hd, and put it out there for the public to use? nothing too revolutionary there, tho, when compared to the phenomenal advances in internet codecs.

i was a compressionist at intervu back in the late 90's, before they were bought out by akamai, so my viewpoint is different than most out here... i do agree with your point wrt the problems with long gop for professional acquisition, but i also have a ton of first-hand experience encoding with the very latest codecs... which is why i can see how horrible the hdv mpeg2 concept is.

hdv is basically the same codec that panasonic is using on their dvd players, yuk!... i would suggest that you go out to www.wmvhd.com and play some of those clips back on your hdtv to see where codec technology is today... take a hard look at the bitrate vs. quality, and how it could be applied to a prosumer format like hdv.

wavelet codec development is alive and well, http://www.wavelet.org/index.php... and i think that the people at cineform have things to say about the viability of their codec :-) but wavelets aren't the only compression technology out there.

thanks for discussing these issues with us! i shoot with an xl1s, but i also own a panasonic av100 that records mpeg2 to sd cards, so i am sold on solid state recording... i want to see it become the standard, but i think that it's going to take codec development to make it really happen... to put it another way: in the computer business, we say that software sells hardware.

<<<-- Originally posted by Jan Crittenden : I don't think you can show me a codec that can be a non-proprietary, meaning anyone can access it and make it happen that is more effiecient that the DV Codec. -->>>

Then posted by Dan Euritt :
>didn't panasonic develop dvcpro hd, and put it out there for the >public to use? nothing too revolutionary there, tho, when >compared to the phenomenal advances in internet codecs.

First when I say DV Codec, the DVCPROHD is a DV based codec. Secondly an internet codec is a delivery codec. Different objectives than production.

>hdv is basically the same codec that panasonic is using on their dvd players, yuk!...

Actually this is not true, as there is no HD on a DVD at this time, soon, but not yet. The commonality is MPEG2 with a longer GOP, but it is not at 25Mbs, the current DVDs are more like 9Mbs. HDV is more similar to the transport stream that the broadcaster uses to to get HDTV to your home HDTV set over the airwaves.

>i would suggest that you go out to www.wmvhd.com and play some of those clips back on your hdtv to see where codec technology is today... take a hard look at the bitrate vs. quality, and how it could be applied to a prosumer format like hdv.

Again, the discussion here is irrelevant to the production domain. Delivery is absolutely different that production.

>wavelet codec development is alive and well, http://www.wavelet.org/index.php... and i think that the people at cineform have things to say about the viability of their codec :-) but wavelets aren't the only compression technology out there.

Find a camera that can handle the load. Find an NLE that can deal with the vairable frame rate, Find the NLE that can deal with on the fly ramping of speed, find a codec with any of those that can do it on the fly with the switch of a menu item and and NLE or two that can accomodate the rest of the handshake. The cineform is a platform codec, and the wavelet is not in the consideration either. Different applications, not acquisition. Might work in the Desktop as long as the concatenation doesn't get in the way but I wouldn't put my money on it till I saw it.

>thanks for discussing these issues with us! i shoot with an xl1s, but i also own a panasonic av100 that records mpeg2 to sd cards, so i am sold on solid state recording... i want to see it become the standard, but i think that it's going to take codec development to make it really happen...

Frankly the Solid State is happening at the professional level on a I frame only on the codecs that are designed for production. The little Panssonic camera you have there is not a production codec. Nice for home movies but not for serious production.

You cannot introduce a new codec today and then expect all sorts of NLE guys to fall all over the idea, fawning to just hold your hand. Getting the NLE guys to deal with the stringent principles that we set into DVCPRO and its deriviatives took several years. DV was being pushed over firewire for at least three years before the NLE guys realized that if they want the News market they needed the DVCPro codec, and then a whole bunch of other things started to happen. DVCPRO50 over Firewire, DVCPROHD over firewire. These sorts of advances take time, energy and money. Codec du jour does not work.

The HDV camera from Sony has had the benefit of the JVC camera being out there for 2 years before its delivery. It takes money and the promise of return on investment to get the NLE guys to dive in and spend the money to make the system work. And this solution still challenged.

There is a whole business of getting folks to buy in to the codec and put together the entire production plan and then continue the work on it so those cool ideas can come to reality; like 24P, variable frame rate, scalable codecs that go from SD at 4:1:1 to 4:2:2 to 4:2:2 HD and all over firewire in native codecs. This is not all something that happens in one year, this happens over years.

To sum up, production codecs have different objectives than delivery codecs. There are some codecs that work well as a platform codec, that allows for special things to happen, like HDV to edited with better efficiency, like the cineform codec. But these do not have acquisition. And then there are delivery codecs, which the HDV codec could be, or the DVD codecs or the internet codecs. The consumer codecs for consumer cameras generally are slanted at the delivery end of the curve as many people are not nearly as fussy as production people.

I hope this helps you understand that 1. the DVCPRO codecs are here to stay, and therefore are a safe bet for building future ideas on. 2. that there are delery codes and the efficiency and low data rate are the goal. and 3, it has take a lot of time out of my day to write this, and for you to say that we should just drop DVCPROHD and go with CODEC du Jour is something that I feel needs to have attention and education aimed at it, because CODEC du jour does not always serve the production community well.

I hope that helps,

Jan

<<First when I say DV Codec, the DVCPROHD is a DV based codec. Secondly an internet codec is a delivery codec. Different objectives than production....

....1. the DVCPRO codecs are here to stay, and therefore are a safe bet for building future ideas on. >>

Jan, thank you for taking the time to post in this forum, it's really nice to have some info from a person with the knowladge!

I do a lot of color corection and finishing from a range of formats: Digibeta, DVCPRO 50 /HD and HDCAM.
Besides the the limited color resolution and aliasing of saturated colors of the 4:2:2/3:1:1 formats I am finding it odd that only the old Digibeta format supports 10 bit sampling.

As you talk a lot about the difference between acquisition and delivery codecs, I am wondering why all current acquisition codecs are 8 bit only?
Especially for film out, banding occurs VERY easily with both DVCPRO 50 and DVCPRO HD and it's VERY obvious in all smooth graduations on noiseless material.

Also using the cinegamma and other DSP functions on the DVX100, the SDX900 and the Varicam makes this worse because to retain the highlights, even fever than 256 levels will be used.

Can you comment on this? Is there a future possibility of finer quantizing in your codecs? or is it locked at 8 bit by being based on the old (but very funtional ) DV technology

I like the very functional workflow we have today, especially using Apple / Panasonic equiptment, but it's too bad we are limited in this regard. (even though it's much better than 4:1:1/4:2:0 DV/HDV)

> You cannot introduce a new codec today and then
> expect all sorts of NLE guys to fall all over the idea,
> fawning to just hold your hand. Getting the NLE guys
> to deal with the stringent principles that we set into
> DVCPRO and its deriviatives took several years.

I tend to disagree. Although there is something to be said about DV and the way it works transparently in an NLE, it is perfectly feasible to use a different codec for aquisition. This has been demonstrated with Sony's IMX. It's MPEG2 based, high quality and is treated by the NLE as uncompressed.

If you give me a camera with an efficient new codec that delivers better than DV, which could be something based on the the MPEG4 AVC, I will gladly spend time transcoding that to DC30 or some other low-loss codec. Granted the workflow won't be the best for ENG, but this camera can be aimed at the indie and documentary markets, as it can be --because of the cost savings in storage-- much less expensive.

I like the logic behind HDV's MPEG2 trick, except that they have ruined it by commiting to tape and thus limiting the data rate. So there is potential to do much better. Sure, less compression is an option, a good one. But *better* compression, coupled with disk-based storage, can also do the job for many of us. The ideal: both! Options, that's what we want. A camera with as many options as a computer.

Well, I'm really curious about the low light performance of the DVCPRO-HD camcorder. If it's just as good as that of the FX1/Z1, then I'll definitely buy it as a replacement for my little cam. (MX5000)

>>>You cannot introduce a new codec today and then expect all sorts of NLE guys to fall all over the idea, fawning to just hold your hand.<<<

actually, that is exactly what has happened with hdv, it's a delivery codec being used for acquisition... all of the major software editors on the market immediately tweaked their systems to accept it... if panasonic suddenly started selling a prosumer hd camera based on wm9 or h.264, the software editors would integrate it asap.

in fact, the software editing systems that use cineform as an intermediary codec should be able to accept it immediately... all panasonic has to do is to create the in-camera silicon for it, no minor detail.

are you aware of the role that the cineform codec played in the making of that new bruce brown baja 1000 movie? i know people who saw the sneak preview onscreen, no complaints about the quality... but based on my encoding experience, hdv/cineform is not a work flow that i want to use for the type of content that i shoot.

so i'll pass on hdv, i know that panasonic will come up with a superior quality solution... and if it'll integrate the solid state work flow i've seen with the av100, it's going to rock! btw, i use the av100 for recording an external camera video feed inside of race cars, because tape-based recorders won't stand up to the vibration.

thanks again for your time out here, i can't wait to see the new panasonic camera!

Anders asked:
Besides the the limited color resolution and aliasing of saturated colors of the 4:2:2/3:1:1 formats I am finding it odd that only the old Digibeta format supports 10 bit sampling. As you talk a lot about the difference between acquisition and delivery codecs, I am wondering why all current acquisition codecs are 8 bit only?

The reason is that those last two bits can be added to any of the 8 bit formats buy going out SDI. Additionaly there is always the added cost for those last two bits. The DigiBeta entry camera is $45,000, the DVCPRO50 which makes a comparable picture is $27,000, Add the deck cost and you are looking a a bigger chunk of change. If you ingest over SDI, the stuff on time line is very similar. So in some ways it is economics, some ways it is the facility to achieve the same net result for less money. These two questions basically get the same result.

All of what this is built upon is generated out of a much more exhaustive piece of work done by people vastly more learned than I will ever be. See http://www.smpte.org/engineering_committees/pdf/tfrpt2w6.pdf

Staring in the Annex C, you can see all of the stnadard definition digital formats compared, even DVCPRO50/DigitalS, Digital Betacam, IMX-referred to as MPEG50/I Frame, DVCPRO. It is pretty heavy but interesting.

>Especially for film out, banding occurs VERY easily with both DVCPRO 50 and DVCPRO HD and it's VERY obvious in all smooth graduations on noiseless material.

I have not seen this as much as you have I suppose and I guess neither did the SMPTE/EBU. Perhaps it may be the rest of the signal path? I don't know. The footage that I see up on the big screens at NAB, 70 feet, don't seem to show what you are saying and those programs are done with the DVCPROHD material. So can't explain your experience.

I can't say that I have never seen it, but then I have backed up and captured the material differently.

>Also using the cinegamma and other DSP functions on the DVX100, the SDX900 and the Varicam makes this worse because to retain the highlights, even fever than 256 levels will be used.

Here I could not disagree more. Noah Kadner has a shot in his film Formosa that is at least 8-9 stops of latitude, inside the car to the bright blue skies of sunny New Mexico, shot with the SDX, and there is detail in the shadow and blue sky. I can point to and endless number of pictures that have huge amounts of detail in the wite areas, where on other cameras they have been blown out. Perhaps you experience could be based on video you have seen shot by people that may not understand how to use the controls provided to them with these cameras. But there is no way I could begin to agree with you assessment of the cine-gamma.

When we move to the first generation of the DVX100, the cine-look there was much easier to blow out the highlights and unfortunately folks didn't know how to shoot with that for the proper handling. As the DVX100 moved to the DVX100A, the Cinelooks expanding to include some that have a more rounded shoulder. Again, what shows up in the edit suite is often predicated by the strength of the person using the tool.

>Can you comment on this? Is there a future possibility of finer quantizing in your codecs? or is it locked at 8 bit by being based on the old (but very funtional ) DV technology

I feel that life is full of little trade offs. When you look at the potential of what is possible versus what are people going to pay, I believe that there is a balance point. DV is younger than DigiBeta, why didn't Sony take DigiBeta and make DigiBeta Lite? the 4:1:1 version with 10 bit? Because that would be silly, no advatange and the machine would still be stunningly expensive. So what a manufacturer has to do is find the price point, and the performance that is pleasing to enough peole that it will sell and then figure out how to make it so that it holds up and is reliable. I mean really, if the 8 bit formats were so awful as you imply, why would any body buy them. Shouldn't they all be lined up at Sony's door with fists full of dollars.

The question is not so simple, and covers more ground than that which can be covered here. I suggest you read Annex C of the SMPTE study above. It might enlighten your perpective. Life is full of compromises.

>I like the very functional workflow we have today, especially using Apple / Panasonic equiptment, but it's too bad we are limited in this regard. (even though it's much better than 4:1:1/4:2:0 DV/HDV)

But arguably, a 4:1:1 2/3" camera can look stunning in the hands of the right operator as can the DVX100A. I really believe that much of the dreck I have seen is based on the fact that the operator is not taking the time to optimize the tool.

Video in and of itself is so immediate that it almost works against the learning curve. They shoot, they have a picture, they break, and move on. Not until they get back in the edt suite do they see the problems in what they shot, because they only had a little $500 dollar LCD panel that you could juge anything on, if your life depended on it. But then the format takes the hit.

When I was learning to expose film, 35mm still, I cannot begin to count the number of rolls of film I went through before I was able to come out with more properly exposed shots out of a roll of 36 that badly exposed. But I took the time to analyze v=because I knew the variable was not me, not this 100 year old medium. I mean jeez if all of these other photogs can get good shots, with the same film and same equipment what am I doing that makes the difference.

Anyhow I hope you can see where I am going. Lif is full of little trade-offs. Many things influence which way you go, and then it is up to you to make the best of your choices. For me I see the trade of of 4:20, a long GOP, and compressed audio, huge in comparison to simple little 4:1:1 DV with progressive imaging, uncompressed audio. I know what the algorithm is going to do when I look into the viewfinder.

Anyhow, good questions.

Best regards,

Jan

Ignacio said:
>This has been demonstrated with Sony's IMX. It's MPEG2 based, high quality and is treated by the NLE as uncompressed.

However, we are talking about a system, that is 4:2:2 and has a starting price point at $34,000. Show me something in the $4-6000 range here. In fact, you cannot.

>If you give me a camera with an efficient new codec that delivers better than DV, which could be something based on the the MPEG4 AVC, I will gladly spend time transcoding that to DC30 or some other low-loss codec. Granted the workflow won't be the best for ENG, but this camera can be aimed at the indie and documentary markets, as it can be --because of the cost savings in storage-- much less expensive.

But in the delivery of MPEG4 or whatever, you cannot invent what you gave away.

>I like the logic behind HDV's MPEG2 trick, except that they have ruined it by commiting to tape and thus limiting the data rate. So there is potential to do much better. Sure, less compression is an option, a good one. But *better* compression, coupled with disk-based storage, can also do the job for many of us.

I would agree that HDV is bit starved, but can't say that any form of GOP is a great idea for production work. I am afraid that what is going to have to happen here is that I will have to stop responding as frankly we will never agree as to what the perfect sets of trade-offs are. What everybody here is saying that we should be able to get full uncompressed quality at low bit rates and have no penalty for making those choices. Sorry it doesn't work that way.

Best,

Jan

<<<-- Originally posted by Dan Euritt :

actually, that is exactly what has happened with hdv, it's a delivery codec being used for acquisition... all of the major software editors on the market immediately tweaked their systems to accept it... if panasonic suddenly started selling a prosumer hd camera based on wm9 or h.264, the software editors would integrate it asap.

Obviously you missed the point in my previous post that said, the reason that the NLEs are where they are is because of the leadership role that JVC took 2-3 years ago, and frankly it still doesn't make the HDV codec less problematic. As far as developing a small consumer camera that does what you suggest, that may happen but I think there needs to be higher adoption in the consumer industry for HDTV.

>in fact, the software editing systems that use cineform as an intermediary codec should be able to accept it immediately... all panasonic has to do is to create the in-camera silicon for it, no minor detail.

IMHO, the cineform codec is a platform codec. I dont think it would lend itself to a camera/price performance payload.

>thanks again for your time out here, i can't wait to see the new panasonic camera!

Thanks, I really think you guys will see it and think it rocks!

Best regards,

Jan

<<<-- Originally posted by Ignacio Rodriguez: ...it is perfectly feasible to use a different codec for aquisition. This has been demonstrated with Sony's IMX. It's MPEG2 based, high quality and is treated by the NLE as uncompressed. -->>>

IMX as is a I-frame only, so it's as far from delivery codec as you can be with mpeg. Practically same as mjpeg.

<<<-- Originally posted by Jan Crittenden: The reason is that those last two bits can be added to any of the 8 bit formats buy going out SDI. -->>>

Can you give us any reason why 10bit colors can't be implemented to firewire transferred stream?

I think the reason for not going to greater colordepths is just because industry has been thinking that 8bits is "good enough for television".
After digital video and especially HD has became widerly used, people are starting to noticing the quality more.

Having more than 8bits is not a price point in still cameras, so same way it shouldn't be with video cameras.

<<<-- Noah Kadner has a shot in his film Formosa that is at least 8-9 stops of latitude... -->>>

Question with greater color depths is not just about how many stops. It's about how many tones you get in one stop.
And having more colorspace gives you more room in production, so you can work faster and still get higher quality in color correction in post production than with 8bit.
Very similiar thing like with acqusition and delivery codecs. They shouldn't be the same.

Digital displays are also late with quality aspects. In DVI specs there is a possibility with dual link to carry MSB with first link and rest in second link. However there is no display or graphics adapters in the market that supports this.

<<<-- Originally posted by Toke Lahti : <<<-- Originally posted by Jan Crittenden: The reason is that those last two bits can be added to any of the 8 bit formats buy going out SDI. -->>>

Can you give us any reason why 10bit colors can't be implemented to firewire transferred stream?

I think this is a question best answered with"life is full of little tradeoffs.

>I think the reason for not going to greater colordepths is just because industry has been thinking that 8bits is "good enough for television".

I think it is based on what people are willing to pay for technology.

>After digital video and especially HD has became widerly used, people are starting to noticing the quality more.

Only the people that produce it are noticing. Average consumer does not.

>Having more than 8bits is not a price point in still cameras, so same way it shouldn't be with video cameras.

It is a question of Pay Load. I know at the highest resolution I get fewer pictures and they are slower to open and slow down my system. I opt for the more lighter load based on my application.

<<<-- Noah Kadner has a shot in his film Formosa that is at least 8-9 stops of latitude... -->>>

>Question with greater color depths is not just about how many stops. It's about how many tones you get in one stop.
And having more colorspace gives you more room in production, so you can work faster and still get higher quality in color correction in post production than with 8bit.

There is no argument here, but if you read the question to which I am responding it was about the Gamma curve, which is about the latitude and the CineGamma and not about the color depth only the tonal range. When you look at the camera front end it is a 12 bit camera, it is much easier to deliver this picture in 8 bit without the banding that the Anders was referencing. It can be delivered with the economy of 8 bit but it dependent on where it starts. The detail is preserved in the highlingt and I have detail in the shadow area, and this is what Anders was say ing that you could not get with the cinegamma curves that we have implemented.

>Very similiar thing like with acqusition and delivery codecs. They shouldn't be the same.

They aren't!

Hope that clarifies,

Jan

<<<-- Originally posted by Jan Crittenden: I think this is a question best answered with"life is full of little tradeoffs. -->>>

I think you can answer with that to every question on the earth, but it isn't very productive.

<<<-- I think it is based on what people are willing to pay for technology. -->>>

People are willing to pay for 8bits when they don't know any better.
Manufacurers primary goal is to make money for share holders, so they will give cheapest solution with quality level that is just slightly above the general acceptance. Nothing new here other than new technology would allow better quality with reasonable costs and when one competitor raises its quality, others have to follow to answer the competition and then we are heading to the right direction.
Now people are building their own 4:4:4 10bit hd-cameras out of industrial cameras for less than $10k, when same commercial "movie"camera costs over ten times more. Maybe next year we will have same kind of commercial cameras on shops shelves...

So we definetly are in a big turning point for moving pictures.
Hacking dvx100 for getting uncompressed stream out is one example. This is what some people wants and maybe manufacturers will have to answer to it.

<<<-- Only the people that produce it are noticing. Average consumer does not. -->>>

I belive that average consumer can notice a diffrence of a cheapest gonzo-tv shot with smallest miniDVs compared to biggest hollywood movies, even if they both are seen from dvd with same datarate.
Noticing migt be in the subconscious level and surely they can't analyze why there is difference.

So consumers dont' notice the need for more than 8 bits, because they don't need it. The need is in the production, not the delivery. Color correction is done before consumers sees the program, so consumers don't have a slightest idea how many bits were used during film scanning or other post production stages.

Every cinematographer who has worked with film knows the need for more color depth than 8bits. Go ahead and ask them!

And I think it's very sad if the industry is changing to new technology only to save money, but quality is decreasing, instead of improving like it should be with technology advancements.

<<<-- They aren't! -->>>

They do have same colorspace!

Thank you for the very good and through reply. I hope you didn't read my post as another format bashing comment, I was just wondering about if it was possible to implement 10 bit in the DVCPRO format.

By the way I really like my DVX100A, and compared to sonys offerings at the time of purchase, it was almost the perfect DV camera. Alright real 16x9 sensors Is still on my wishlist but as you also rightfully stated life is full of tradeoffs and all in all this is a great package at a great price.

Also from the reports on these boards, hdv editing is really not working efficiently yet, at least not as good as DVCPRO/HD is currently

Regarding codecs do you think that constant quality VBR codecs will be a viable option compared to CBR, in the future as we are moving to randam access storage like solid state technology?

<<- Originally posted by Jan Crittenden :
The reason is that those last two bits can be added to any of the 8 bit formats buy going out SDI. These two questions basically get the same result. >>

Yes, I have captured from both varicam and 750p and 900 thru HDSDI into uncompressed 10 bit 4:2:2 HD, bypassing the data truncation/compression of the tape decks. Ofcause this does not only awards me with 10 bit quantizing but also bypasses the 6.7:1 compression so actually what a 10 bit 4:2:2 6.7:1 format would look like I really don't know....


> See http://www.smpte.org/engineering_committees/pdf/tfrpt2w6.pdf
Staring in the Annex C, you can see all of the stnadard definition digital formats compared....The footage that I see up on the big screens at NAB, 70 feet, don't seem to show what you are saying and those programs are done with the DVCPROHD material....I can't say that I have never seen it, but then I have backed up and captured the material differently.

Yes the report is a very interesting read (I did read it back in 2000) but it is also written from the perspective of broadcast TV and not filmmaking which is what I do.
Some of the conclutions are still interesting, and of cause Panasonic is still a broadcast company not a film company like Arri or Panavision and still the tools that you supply will be aimed at both groups of people.


> Here I could not disagree more.... there is no way I could begin to agree with you assessment of the cine-gamma.

I think you misunderstood my statement. What I ment was that when you expose for the highlights and is using a workflow where you will shift the levels later in the gradesession, you will see a huge difference in the finer grauations going from 8 bit to 10 bit in quantizing.
When you aim for that 8-9 stops with cinegamma very few levels sre the for subtle thigs like skintones etc. I firmly believe going 10 bit would make quite a bit of difference.
I am impressed at the imagequality of the varicam/sdx900 and the dynamic range that it will capture, especially using the excellent gamma settings. We are used to 10 bit log in digital film for a reason, and with the gamma settings we are closer to that.

> Again, what shows up in the edit suite is often predicated by the strength of the person using the tool.

> I mean really, if the 8 bit formats were so awful as you imply, why would any body buy them. Shouldn't they all be lined up at Sony's door with fists full of dollars.....I suggest you read Annex C of the SMPTE study above. It might enlighten your perpective. Life is full of compromises.

I not implying that theese formats are awfull, these formats a now proven technology that is used every day professionally. I just find it's a limiting factor that it's 8 bit only. Will I


> But arguably, a 4:1:1 2/3" camera can look stunning in the hands of the right operator as can the DVX100A.

It can indeed!


> Anyhow I hope you can see where I am going. Lif is full of little trade-offs. Many things influence which way you go, and then it is up to you to make the best of your choices. For me I see the trade of of 4:20, a long GOP, and compressed audio, huge in comparison to simple little 4:1:1 DV with progressive imaging, uncompressed audio.

Yes, it sounds scary to me to. And therefore I am not buying the Sony offerings but waiting for the new panny...

> Anyhow, good questions.

Thank you, good anwsers as well...

Jan, have you personally seen the picture (output) from the upcoming 1/3 lil DVCProCam?

Interested in your take on it.

Toke, there are several issues with 10bit color with todays systems.
1. Most video cards only support 8bit per channel. Matrox has one that supports 10bit on your screen, but you give up most of your alpha channel to get it.
2. 10bit codecs put a lot of strain on most desktop computers, both cpu and IO. If you want a real world test, go to Blackmagic designs' web site and download their free codecs for Quicktime (both Apple and Windows). They will show up for any nle that supports Quicktime.
Both Premier and Vegas on Windows can use them. Try editing and rendering and everything you usually do. The do not require BlackMagic hardware to run.

3. Maybe this will all change once we get 64bit windows, but who knows.

I too would like 10bit or better yet 12 or 16bit, but those systems are out of my price range. 10bit uncompressed HD(1080p) single stream requires minimum 1.2gigabit throughput. (over 150Mbytes per stream). thats something that can even put lots of strain on Ultra320 Scsi drives. Most HD is edited offline in some form of SD.

If you want 10bit, scream at MS and Apple and the video card manufacturers (Apple has repeatedly refused to offer 10 or 16bit). Until they support and deploy it, it's unlikely the camera makers will make affordable 10bit cameras. No market outside of very high end systems.
(btw, don't mistake the 10bit ramdac outputs the video card makers talk about. Thats strictly for video/dvd playback, not for cg or editing).

Your frustration is felt by many of us, but it's not the camera makers at fault on this one. really.

<<<-- Originally posted by Toke Lahti : <<<-- Originally posted by Jan Crittenden: I think this is a question best answered with"life is full of little tradeoffs. -->>>

>I think you can answer with that to every question on the earth, but it isn't very productive.

But it is reality.

<<<-- I think it is based on what people are willing to pay for technology. -->>>

People are willing to pay for 8bits when they don't know any better.

Unfortunately this is not held up with evidence. If everyone had refused to use DV, the course of compression would have been much different. If everyone was willing to pay for the 10 bits, then they would have made DigiBeta the most popular format and not DV based products. So it is fairly obvious from here, that no, people are not willing to pay. DigiBeta was in the market for almost 2 years prior to any DV based format!

And all new products, even HDCAM, IMX, HDV and XDCAM are 8 bit. So it is a matter of what people are willing to pay balanced against a percieved quality performace. Can you pay for better, can you renegade something like Juan's product. Sure, but there is a much smaller market. And if there is something huge there, then Juan will make a lot of money.

Best,

Jan

Excellent question, Michael!

I was thinking the same thing myself. Hope you don't mind me adding to your question...
Jan,
have you seen footage from the camera? If so, what size screen(s) and what was your initial reaction?
I remember mine when I first downloaded a M2T file from the FX1 and holy @#!@$. Anything like that?

<<<-- Originally posted by Jan Crittenden: If everyone had refused to use DV, the course of compression would have been much different. If everyone was willing to pay for the 10 bits, then they would have made DigiBeta the most popular format and not DV based products. -->>>

Quality was not the reason why everybody didn't buy digibeta. It was price.

My frustration of this slow technical development in digital video might become in working in different fields.
When I bought vx1000 in -95 it seemed that development will be rapid.
Before that people were using u-matic with inferior quality and analog beta with over ten times higher prices.
After that almost nothing has happened.
Maybe prices in higher quality digital video has halved in last decade (not digibeta of course), but that's really nothing when you compare advancement in computers or digital still photography. There one zero has dropped from prices and quality is ten times better than decade ago.

I also bought 16:9 televison in -95. Still before now you couldn't buy 16:9 camera with at least 1/3" sensor less than $20k.

I think that reason for this slow development is mainly lack of competition. There is sony with everything very expensive and panny with something less expensive. If you compare situation in still cameras there is about a dozen of top grade manufacturers. And there is lot of choises in the "middle class".

Another thing might be that better than consumer class quality is instantly thought to be needed only in big money television productions in top economical areas where one hour of program can easily cost eg. $500k. How about areas where hour of program can cost only one tenth of it or less? Well, just use the same cameras than home hobbyist...

Technically it would have been very easy to offer 10bit upgrade to dv-formats for years. That would make 25Mbps to 32Mbps that could be easily handled with existing tape and transfer technics. Camera heads already have higher than 10bit dsp's, so there would be no expenses over there either.
Every $1k computer today could handle 10bit even if converted to 16bits.

I hope that this new Panny's camera will strike to this non-existing $2k-20k range with good quality for smaller economic areas and non-commercial productions!

First you say that people are willing to pay for 10 bit, and when I show you that they aren't; you say that it is because it it is too expensive. Hey, you said that people were willing to pay for it. Apparently not.

You make asumptions, you think are correct. In reality, you don't know because you are not in the manufacturing end of things, you are not in the R & D Department, you don't know how long it takes to make things happen and or even make a new product. I have tried to explain. You don't care for what I have said and do not want to beleive it. You think it is based on the lack of competition and that isnt it at all.

Please, this part of the thread is going nowhere. You know it, I know it, all that are reading it know it.

I will not respond to this part of the thread again.

Best regards,

Jan

I feel your pain Jan. Thank for all the enlightening insight you have painstakenly written. Having been in product design over the years I perfectly understand you points.

jh

Jan, I am very gratefull for your presence here, it's a great opportunity to voice our concerns, ask intelligent questions and, even though you can and will not "leak" information that we might be eager to know, you do add enormously to the quality of these conversations and go to great care and detail in answering when you can. Please don't go away just because someone is obsessing about an issue or two. Stay with us.

Hey I am not thinking about going away, I just refuse to waste my time on that part of the thread as t benefits none of us. No I want to be here so that when I can say something you guys will be among the first to know.

Best regards,

Jan

Hi Jan,

Thanks again for letting us know what you Can let us know. Theoretical discussions Are a waste of time when you've got a new camera just around the corner!

As I assume the new HD/P2 is still 8 or more months away, here are a few details of its design that I really hope are looked at very closely before its release. No pie-in-the-sky wishlist, just realistic tweaks that could make all the difference. I'd appreciate your comments on what has been or may be implemented re:

1. Viewfinder. --- Needs higher magnification, higher res. If it won't get a physically larger & finer viewing system, than the existing one could be at least made better by exchanging a few optics in it w/higher magn ones. Bigger is better here, and absolutely needed.

2. Viewfinder again. - The only way to exploit the camera's quality (and I believe you that it's gonna' Rock) is to be able to Focus it. Porting over the DVX100's vf will simply not be up to the task. But if the existing vf is what ends up on the HD/P2, then a focus assist toggle that would greatly magnify a central portion, a la Sony Z1/FX1, will be the minimum necessary. This also requires either a higher res vf (or LCD) to make it of much value, and if this system is used, it really should be a good bit better in res/magn than Sony's, which is just Barely adequate. And of course it should also work While recording.

3. ...Let's see... the VF! - Getting the vf much better on a budget is I'm sure a tall order, but will really show Pana's intentions for its end-use, and will make or break the camera's success in the long run. This because when down the line, someone else comes out w/a camera even better than DVCPRO HD, yours will still compete, if it has the better viewing system.

4. Image Stabilization. This camera is probably going to be Extra light in weight, so hopefully the OIS will be specially dialed-in to the nth degree. And made available in HD mode. I think a small form-factor for a camera like this is perfect, but that means it will end up being hand-held that much more often.

Along w/picture quality, and of course **low-light capability**, the ability to fine focus must be job #1 on the new cam. Thanks for listening, and if anyone in engineering/product dev needs to be reminded of any of the above before it's too late, please forward!

Agreed completely. Focus is crucial, especially considering the cost of an HD field monitor. PLEASE include an "enhanced focus" option, something like what the FX1/Z1 do, which would let us zoom in to full pixel resolution for focusing! That's probably the #1 best feature it *needs*...

Hi Vlad,
>As I assume the new HD/P2 is still 8 or more months away, here are a few details of its design that I really hope are looked at very closely before its release.

You would be amazed at how much is locked in at this time. Your suggestions are good ones and I will pass them along. From the time of the first DVX idea till the first NAB with it, was 18 months. There are many things that go into these little cameras that not everyone can appreciate.

It is sort of like the 1880 Victorian my husband and I purchased. Before we do anything really cool, we have redone the electrical, which was done in the 20's and the 40's and again i the 90's, we have redone the gas, and the plumbing. When we are all done, none of this work will show, and we have owned the house for 5 months. Of course we gratefully don't have to live in it at the same time as we are the world's slowest contractors. ;-)

>This camera is probably going to be Extra light in weight, so hopefully the OIS will be specially dialed-in to the nth degree.

People think/thought the SPX800 would weigh in significantly lighter than its tape-based brother the SDX900. They are within ounces of each other.

Thanks for the ideas, I will pass them along.

Best,

Jan

> something like what the FX1/Z1 do,
> which would let us zoom in to full pixel
> resolution for focusing!

Yes! But unlike the FX1/Z1, make sure it can be used while recording.

Since we are all speculating here (except Jan) I have to believe that on the day (or shortly after) the new Pana is officially announced, there will also be an announcement from some 3rd party about an dvcprohd hard disk recording system. There shouldn't be any reason the pana couldn't record directly through the firewire interface. If it's anything like what JVC does for their 5000 series of cams, it should be quite small and easy to haul around. Maybe even hook up to the on cam battery.
If this isn't the case... someone on this board should get venture captital and move quickly.

I really think the current crop of P2 cards are more for ENG work than short/feature production. at least for the near term.

There are too many smart people out there to ignore such an opportunity.

But with NDAs all around I'm just blowing smoke at the moment. Okay I've got that off my chest now. hehehe

Since we are wishlisting, here is the Hard Disk that fits in a PC-Card slot and can handle more than 100 Mbps. Please Jan pass on the idea that the camera should be able to use this kind of storage in it's slot:

Meet the Toshiba 1.8 inch PC-Card. The initial offering is 5 GB but it seems they can scale that up to 60 GB as they have disks in the same form factor with that capacity:
http://sdd.toshiba.com/main.aspx?Path=810000000007000000010000659800000000/81000000010D000000010000659C000003B7

Jan, I do appreciate how far in advance the many details of these things have to fall into place, which is why I limited my suggestions to those that could be Relatively easy to change, if they want to. But I cannot stress enough the importance of having a better finder in this cam. It is the vital heart of the device that has Not kept pace w/the advances in its electronics.

We're reaching the point w/this level camera, where potentially more important or more-widely viewed work could get made w/them, some of it for the big screen. Pana is obviously keeping its price point flexible (or at least under its hat) at this stage, so if adding $500 or $800 to the MSRP can allow it to have an improved viewing system, then I'm certain that would help distance the camera from the field and sell more of them, and make users who know the difference grateful for Panasonic's commitment.

Oh, and good luck w/the house!

Jan,

I love my DVX100A, it is simply an unbelievable camera for the money.

If I was going to fault anything about it, it would be the viewfinder/LCD. (with full recognition of what's possible, versus money and other"big picture" thinking.)

So, with respect, and from my little world, I echo the wishes of some previous posts.

In regard to data storage (tape, p2, giant p2, third party cards, camera mountable hard disks, magic codecs du jour, blah blah blah) its my hope that I would be able to take this camera on the road for a couple of weeks and return home with the aprox 600 minutes of HD footage I would shoot for two episodics, with low hassles and reliabilty. My current HD cameras are all Arriflex.

I really don't care if it writes its data on peanut butter cubes in sanscrit, so long as its reliable, affordable, and I can shoot for an hour or so without breaking down the production, and i can easily and safely store the data.

This is the message from my world. Other peoples worlds may differ.

Thanks

If you've never been directly involved in the design, development, test, and production of a product it may be hard to understand and appreciate the lead time necessary to bring a product to market.

The final design features MUST be frozen well in advance of production in order to properly test and evaluate the product. During/after testing, usually ONLY the most critical flaws/defects are corrected (and re-tested). The introduction of new design features after the start of testing can have significant, unintended, negative consequences because the product may not have been properly tested prior to production.

While we all desire lots of great features, we all require that the camera work flawlessly, and as advertised. This requires exhaustive testing of a frozen design.

Panasonic has clearly shown that they listen to their customers -- the DVX100 --> DVX100A is a perfect example.

So we shouldn't expect nor pressure Panasonic to be making design changes for the HD camera(s) to be announced at the April 2005 NAB -- it's only a month away!!

> So we shouldn't expect nor pressure Panasonic
> to be making design changes for the HD camera(s)
> to be announced at the April 2005 NAB -- it's only
> a month away!!

I see your point. But the request I have voiced --that is for the P2 slot to be able to use a PC-card hard disk-- is in essence a software feature, it does not alter the physical design of the camera. Also, some requests we are making now can be answered by Panasonic in the way of accesories, deals with other companies and so on.

Software/firmware is more difficult to test than hardware -- it does NOT have to follow the laws of physics. :-)

The last thing one wants to do is make a last minute software change -- the risk of unintended consequences are great!

Why don't we let Panasonic get the camera out the door before we tell them how to redesign it. :-) I would expect there will be future versions/updates to this camera giving us many opportunities for us to provide meaningful feedback to the manufacturer.

Jan has been very gracious and generous with her time here, and has provided incredibly detailed answers -- much more so than any other camera manufacturer I've seen. Let's not run her off. As soon as the new camera hits the streets I'm sure we'll have many questions we would like to ask her. She is an invaluable resource. Her presence here certainly makes me want to strongly consider Panasonic over other brands.

-->>> Her presence here certainly makes me want to strongly consider Panasonic over other brands. -->>>

Likewise. I'm lucky to be in the position where I don't have clients and no real need to upgrade now but I see my XM2 being a little too long in the tooth by the end of the year. Being someone who doesn't want to go the HDV route, I am very interested in what Panasonic will offer.

Aaron

Hey Jan,

You've heard of this new camera coming out by JVC, the GY-HD100. How do think your camera will compare to this one?

<<<-- Originally posted by Laurence Maher : Hey Jan,

You've heard of this new camera coming out by JVC, the GY-HD100. How do think your camera will compare to this one? -->>>

Well the JVC camera will still be a flavor of HDV, which means it is a low data rate, which means that it is still 4:2:0 and compressed audio. It should be interesting to see how they do the 24P and still make the camera record in 60Hz. Unless it is an all new HDV deviant and then that screws up the works for editing. If all of this is resolved, I would be this little camera would smoke the Sony Z1.

That said, up against the Panasonic there are features that it just can't stack up against. Higher data rate, means more info, means uncompressed audio and 4:2:2. DVCPRO, DVCPRO50 and DVCPROHD in editing systems that are supporting it for years. So this camera works in an extensible codec, that allows you to do DV one day HD the next day support the SDX900 camera the next day and the only compromise is the one that we made on compression in the first place back when we said okay, highest quality we can get at three levels of compression, all I frame, intra frame, all uncompressed audio and some have more than two channels but all are at 16 bit, 48kHz. And that is the stuff I can talk about that reassures me that the p2HD camera will be a smokin' camcorder.

Soon,

Jan

This has been one very informative thread, from codecs to bagpipes.
(I'm Scottish) Jan, since you mentioned darkroom work, and I got into Film (albeit on a pretty modest indie level) after twenty yrs. of still photography you might want to take a gander @ www.williamanderson.ca It doesn't move, but...

Thanks Jan,

Mmmmmm.

4:2:2

High data rate.

DVCProHD . . . mmmmmm . . . Jan, you're my hero.

Uh-Oh Jan,

Just heard about this very possible "uncompressed component output" on the new JVC camera. Go to:

http://www.videosystems.com/e-newsletters/HDVatWork_3_23/index.html/#1

It mentions uncompressed out at 720/60p, but the guys here (and most of them are smarter than me) pretty much insist that if it can do 60p, it will do all the standard progressive frame rates, including 24p. You might check the article and tell me how you think the camera compares now. I'd be VERY interested in knowing (given the above speculation is correct), your stance on the possible quality differences between this new HDX-100 and the JVC model.

The JVC comes with interchangeable lens system as well, mentions cinema-gamma, I think 2 XLR inputs. In order to use the uncompressed component out, it also said you have to get a special SDI converter or something, so I'm sure that can't be too inexpensive.

Personally, I like all my puzzle pieces to fit together, so I would have a hard time not going with DVCProHD codec, as my Macintosh FCP HD Editing system is already solidly compatible. If I had to go SDI, I'm sure I now have to deal with getting an extra capture card and yada yada yada. The JVC articles mention that the uncomponent signal output can be coverted to "any type quick time file" with a free shareware, which I assume will quickly become common, but I'm not so sure I trust SDI converters and software converters to not give me all sorts of problems trying to get a clean and wonderful HD signal imported, edited, then exported via FCP HD.

I'm no techy, so anything that can go wrong will. Quality meeting efficiency in filmmaking for me seems to point to the hdx-100, but I must admit the word "uncompressed" carries a lot of weight in the world of HD cinema filmmaking.

Please give me comments/comparisons on all this.

Thanks so much!

Laurence