DVCPRO HD - Storage & throughput requirements

[fiercecurry]

sure faster drives help but a faster bus (sata) will allow for quicker processing and transfer of hd data streams - especially uncompressed. i know dvcprohd is compressed but that doesn't mean it has to stay that way in post.

What is the point in turning your compressed DVCPRO HD footage, into uncompressed HD footage. Your only making yourself a bigger data file, and creating more problems for yourself in terms of processing speed and hard drive requirements. You're certainly are not going to get a better image out of it. Why make life more difficult?

 

[ChuckS]

The issue is not so much compressed versus uncompressed, it is 8bits versus 10bits.
8bits was marginally OK when the viewer watched on a crappy 525 line TV, but there is a very different look and feel to HD that shows all the limitations of 8bits [banding etc]. DVCPro is probably the best acquisition format to tape, but you want to get into a 10bit log workflow in post as soon as possible.

 

[wigby]

mac are not pcs

mac are not pcs

Wigby,
that's the way it is. Some effects are realtime, some are not. Which ones qualify for realtime depends on your installed RAM memory and your graphics board. There are better ones and cheaper ones. And it depends on available video memory. The more the better for realtime editing.

You can't say "I have a G5 and that will do." You have to tweak the whole system to boost and optimize its performance. Wouldn't try DVCProHD under 2 gigs of RAM and with a videocard with less then 128 GB of video memory. So a Powerbook 1.67 GHz with 2 gigs of RAM and an ATI 9700/128 GB is a save bet. Not to forget the hard drive - 5400 at least, 7200 better.

of course you can say i have a g5 and it will do...that's the way apple designs them. we're not talking about the latest nvidia release this month and overclocking cpus. apple designs complete systems that are not readily upgradeable. you may have noticed how few options there are out there for upgrading mac.

besides, this goes right to my original question - what third party hardware accelerators might be the best suited for working with dvcprohd in fcp?

 

[Pais]

There's also a relationship between read speed and the number of simultaneous files being read. Just because a drive array can read data at 75/MB per second does not mean it can also read from three files at a sustained 25/MB per second.

Also, if you guys are really curious, couldn't you just scale up some video into the target format (i.e. DVCPRO HD at the desired resolutions) and try editing, playing back, etc.? It doesn't really matter whether the content was 'uprezzed', since DVCPRO HD has a constant bitrate.

Alas, I do not have a disk array, so I can't do it myself.

 

[Pais]

you may have noticed how few options there are out there for upgrading mac.

Well, in this case, we're talking about the RAM, disk, and video card -- all of these are readily upgradeable (and identical to their PC counterparts, except the video card's firmware). What you can't replace yourself is the CPU, the motherboard, the power supply, etc.

 

[Jan_Crittenden]

8bits was marginally OK when the viewer watched on a crappy 525 line TV, but there is a very different look and feel to HD that shows all the limitations of 8bits [banding etc]. DVCPro is probably the best acquisition format to tape, but you want to get into a 10bit log workflow in post as soon as possible.

Chuck, it is not always necessary to go to 10 bits. There have been a good number of programs made on the VariCam, edited in FCP over firewire and out to broadcast or film. It really is more dependent on what else you are trying to do with the image. Editing in the native codec can be outstandingly effective in maintaining the original quality.

Best,

Jan

 

[KnockOutFilms]

As far as IO cards, the Kona 2 from AJA is listed as a DVCPROHD accelerator. This card is supposed to be the shiznet.

Cheers,
Chris

 

[fiercecurry]

I think hardware accelaration is the way to go. But I am curious what FCP 5 and a G5 (with 2GB of RAM)can do on its own. FCP 5 ships today. I dont know why people would really need more than 2 HD streams while editing anyway? 2 is enough for transitions and effects. With celluloid you werent running more than one stream. For hardcore effects youre better off working in an animation/compositing program anyway. I guess some people like to impress others with how many streams they have running at the same time. Some sign of manhood?

 

[ChuckS]

Chuck, it is not always necessary to go to 10 bits. There have been a good number of programs made on the VariCam, edited in FCP over firewire and out to broadcast or film. It really is more dependent on what else you are trying to do with the image. Editing in the native codec can be outstandingly effective in maintaining the original quality.

Best,

Jan

Jan, I’m sure your aware that this is by no means a new debate. Quantel tried for years to convince people that 8bits was good enough because all of their software was 8bits. When Greg Kennel and Brad Hunt developed the Cineon spec and workflow, the spec that film I/O facilities adhere too, just about all software and hardware solutions began working in 10bit log workspace. One of Apples big announcements at NAB was that Motion, FCP and Shake all had a 16bit workflow.

DVCProHD is a great codec that works with the limitations of the recording media (tape) for acquisition, but I would post and output in the 10bit world. Also, saying that a good number of broadcast and film programs were shot with the Varicam and edited in FCP in 8bits in context of this discussion is a weak point, very reminiscent of Quantel, because it could also be argued that many more broadcast shows and films were acquired with a PD150 and edited on an Avid.

 

[Jan_Crittenden]

Chuck,

While I am sure there are times when 10 bit has a pay off, it does not always have to be played. Since there are so few HD codecs that are 10 bit in acquistion, namely none, the time and space to change it on the time line should be done for reasons that are warranted. It is not necessary in every session. It can be prudent in some situations, but the fact remains that there are many projects that are indeed edited in the 8 bit domain, vastly more than those in 10 bit.

That was my point,

Jan

 

[ChuckS]

Sure, if you’re going to acquire in DV and master to SD, use the DVX and stay in 8bits. But if you’re going to the extra trouble and expense of shooting HD and mastering to HD or film, 8bits doesn’t cut it. It didn’t ten years ago and it doesn’t today.

The problem from Panasonics perspective is that if your position is that 8bits is good enough – because there are no 10bit acquisition codec’s, well you know that’s going to change. I’ll bet you a shiny new HVX200 that cameras will appear before the end of this year with 10bit codecs…

My point is that all of the projects that we have done that were 720P 8bits edited with FCPHD looked pretty bad compared to the original footage. By the time the project was output – FCP image processing had considerably stepped on the image. I have noticed that a lot of Panasonic reps defend the integration of the DVCPro HD codec in FCP, I guess that’s cool, but I think it would be much cooler if Panasonic would open it up, give us enough source code segments so that we can develop tools that make it easy for us to ingest it into our facility and integrate it into our own workflows.

I want one format for acquisition that I can standardize on and another for post so that I can control the workflow, efficiency and the quality of the look so I can differentiate my productions from my competitors. Apples concept of “production in a box” is a little archaic, if you’re a one person shop Apple’s pretty cool, but I often have two editors, two graphic designers and a compositor working on the same project that need to seamlessly share files and resources concurrently.

 

[D_and_G]

Sure, if you’re going to acquire in DV and master to SD, use the DVX and stay in 8bits. But if you’re going to the extra trouble and expense of shooting HD and mastering to HD or film, 8bits doesn’t cut it. It didn’t ten years ago and it doesn’t today.

The problem from Panasonics perspective is that if your position is that 8bits is good enough – because there are no 10bit acquisition codec’s, well you know that’s going to change. I’ll bet you a shiny new HVX200 that cameras will appear before the end of this year with 10bit codecs…

My point is that all of the projects that we have done that were 720P 8bits edited with FCPHD looked pretty bad compared to the original footage. By the time the project was output – FCP image processing had considerably stepped on the image. I have noticed that a lot of Panasonic reps defend the integration of the DVCPro HD codec in FCP, I guess that’s cool, but I think it would be much cooler if Panasonic would open it up, give us enough source code segments so that we can develop tools that make it easy for us to ingest it into our facility and integrate it into our own workflows.

I want one format for acquisition that I can standardize on and another for post so that I can control the workflow, efficiency and the quality of the look so I can differentiate my productions from my competitors. Apples concept of “production in a box” is a little archaic, if you’re a one person shop Apple’s pretty cool, but I often have two editors, two graphic designers and a compositor working on the same project that need to seamlessly share files and resources concurrently.

So chuck, what would be your solution given the current paradigms of today ( without resorting to high end equipment - Adrenaline, DS etc...) ?

BTW isn't the Kona 2 10 bit ?

 

[fiercecurry]

Chuck,
if you like the 10 bit workflow so much, couldn't you just re-render your DVCPRO HD captures at 10 Bit, using some other codec? Or go even more extreme and use Blackmagics 12 bit 4:4:4? You would be the man!

Or are you against the whole DVCPRO HD format? If you are than the Panasonic's gear isn't for you, just acquire footage with the Sony F950 or the Viper, and edit native 10 bit 4:4:4 or 4:2:2 using a Kona 2 or Blackmagic HD cards.
I am not sure what is the point you are trying to make?
That DVCPRO HD isn't as good as 10 bit 4:2:2 HD?

 

[Jan_Crittenden]

The problem from Panasonics perspective is that if your position is that 8bits is good enough – because there are no 10bit acquisition codec’s, well you know that’s going to change. I’ll bet you a shiny new HVX200 that cameras will appear before the end of this year with 10bit codecs…

Actually Chuck DVCPRO HD is 8 Bit and will continue to be 8 Bit. HDCAM is 8 Bit and will continue to be 8 Bit. D5 HD is 10 Bit and above that is 10 bit. But life is full of compromises and frankly 8 bit can be good enough. I don't think you will see too many 10 bit acquisition codecs because it makes the data load pretty huge and we are already having to compress it to get it to a manageable size in 8 bit.

My point is that all of the projects that we have done that were 720P 8bits edited with FCPHD looked pretty bad compared to the original footage.

Can't imagine what you were doing but perhaps you may have had a lot of effects in it and this would be one of those situations where I would have recommended to work in an uncompressed domain, i.e. 10 bit.

I have noticed that a lot of Panasonic reps defend the integration of the DVCPro HD codec in FCP, I guess that’s cool, but I think it would be much cooler if Panasonic would open it up, give us enough source code segments so that we can develop tools that make it easy for us to ingest it into our facility and integrate it into our own workflows.

Open what up? the data stream so you are going to write your own codec? Why not just transcode to the cineform codec, or go in uncompressed. I don't think that you have to stay in 8 bit and that 8 bit is good enough, it is sometimes and that is what I said.

I want one format for acquisition that I can standardize on and another for post so that I can control the workflow, efficiency and the quality of the look so I can differentiate my productions from my competitors.

So you can do that, I am not saying that you can't or shouldn't. But not every one has to nor should they. If you want to go through the process, have at it. I have stared at the Apple output on a 60" plasma and found it to be stunning, If it is going to look bad, it would look bad there, so I don't know what you are doing but that really doesn't matter. If you want a different codec for post, they are out there. Enjoy yourself.

Best regards,

Jan

 

[thisiswells]

The problem from Panasonics perspective is that if your position is that 8bits is good enough – because there are no 10bit acquisition codec’s, well you know that’s going to change. I’ll bet you a shiny new HVX200 that cameras will appear before the end of this year with 10bit codecs…

I know you're talking Panasonic specifically, but getting outside that box, you do realize DigiBeta is 10bit aquisition, right? It looks fantastic, but it's also 90Mbps for SD. That's an enormous amount of data. So, there is such a thing as 10 Bit cameras, but they're super data hungry and very expensive. And, the visual difference is (I can say this since the organization I worked for last year had four cameras and two decks) isn't that much different from 8 Bits in terms of an aquisition format. Editing is another story. I don't think anyone ever said the 8 Bit domain is better than 10 Bit for image quality with lots of effects added... But, it is a lot less data intensive than 10 Bit and for some (or most) folks, that is plenty good enough. I have edited plenty of stuff in the 10 Bit Uncompressed 422 preset in FCP4 and it definitely renders better than the DV timeline, but it's also consuming, what, like eight times the bandwidth? The difference isn't eight times better, so I can understand recommending to use the native codec for that reason alone.. Surely Cineon or Uncompressed will render better, but for me that difference isn't night and day, but it is better, you're right.

 

[someday]

10 bit vs 8 bit, the matter is simple: unprocessed editing (no plugins) can be done at 8bit.
But so far you apply a plugin (i do always) you need extra bits for cleaner color and image.