MAC/FCP Uncompressed capture system

[Jacques Mersereau]

I am in the early process of designing an uncompressed HD capture system
and I would like your informed opinions.

Computer: MacPro dual quad core/16 gig ram
Hard Drive: Apple Xserve RAID
Interface: AJA Kona 3 or BlackMagic Eclipse
Software: Final Cut Pro 6
-----
Cameras (4): Several Choices with the goal of high quality in low light
and 1920x1080 output.
Front Runners:

Budget if High: RED or Panny HPX 500.

Budget limited: Canon XL-H1 (although I don't like interlaced and do not
know if you can hot switch 24F)

Low Budget: Canon XH-G1

_____

Switcher choices:

Big Money: Grass Valley Kayak

Less Money: Data Video SE-1000

Low Money: Panasonic AVHS300G

________
Monitors:
Camera Preview monitors (5): Panny Panasonic BT-LH1700W
Output Monitor: JVC DT-V24L1DU
(2) Edit Monitor: Dell 24"

____

Do I need a trisync generator to make this play together?

[Glenn Chan]

1- In my opinion, the Red camera can give you better quality VS "uncompressed" even though it is compressed. The sensor itself is very low noise to begin with and their compression looks very good from old stills (on the CML site; the codec has changed since then).

1b- 4:2:2 is usually called uncompressed, but the subsampling behaves much like compression (lowers bandwidth, causes some mathematical loss). IMO it is inferior to Redcode RAW and is not visually lossless.

2- That might be missing the point however. Please state in your posts what you're trying to do.

Are you trying to put together a studio for live productions with live switching? Or is it for EFP use? (Reading your post, I can't figure it out.)

Is this for a university?
If it's for a university, then the educational discounts/wheeling-and-dealing that you can get can affect the price equation dramatically.

If it for teaching, or a combination of teaching and in-house production purposes?

[Jacques Mersereau]

[QUOTE=Glenn Chan;725517]1- In my opinion, the Red camera can give you better quality VS "uncompressed" even though it is compressed. The sensor itself is very low noise to begin with and their compression looks very good from old stills (on the CML site; the codec has changed since then).>>

Agreed. Since this system is for the University of Mich.
and since RED is of yet vapourware, I am unsure as to
if our bosses will foot a $120K for four Reds (alone).
That is my first choice, but even if REDs start flowing tomorrow, according
to their production schedule, it would be next summer at the soonest
before we could get our hands on them.
There are other issues such as delivery (we are not going to do film outs),
presentation (4K projectors are starting to appear) and what our clients
can deal with when we hand them a 4K RED CODE file (most are use to DV or HDV).

<<1b- 4:2:2 is usually called uncompressed, but the subsampling behaves much like compression (lowers bandwidth, causes some mathematical loss). IMO it is inferior to Redcode RAW and is not visually lossless.>>

I agree, with the exception of visually lossless. Uncompressed 10 bit
HD is probably lossless to my old eyes and way better than NTSC
or HDV. :)

<<Are you trying to put together a studio for live productions with live switching? >>>

Yes. Currently the video studio serves a wide variety of functions,
but multi cam shoots of live performances, interviews and
greenscreen stuff is what we do most of.

<<Is this for a university?
If it's for a university, then the educational discounts/wheeling-and-dealing that you can get can affect the price equation dramatically.>>

Yes, and I hope so. First I have to get a plan together and then
approach vendors and manufactures . . . right, Joe, Jim and Jan? :)

[Glenn Chan]

Some info on why 4:2:2 is not always visually lossless:
http://glennchan.info/articles/techn...ma/chroma1.htm

It's situations like red text on a black background where you run into some problems. While my article's examples are 4:1:1, the same deficiencies exist in 4:2:2.

Yes. Currently the video studio serves a wide variety of functions,
but multi cam shoots of live performances, interviews and
greenscreen stuff is what we do most of.

If you want to do live switching, then you will need either:
A- Genlock all your cameras. You will need a sync generator and possibly a distribution amplifier for additional outputs. If the runs are long, you should adjust the cameras to compensate for the delay in the signal.
B- Use a mixer with frame synchronizers. They buffer the whole frame and this helps get around the genlock problem if your cameras don't have genlock in. (Genlock is also called sync.)

Personally I'm not a big fan of live switching... you need a crew of people who are good at it. With students they are generally inexperienced and they graduate before they have enough time to get good at it. Plus, if I were a student I wouldn't be that interested in multicamera work.

1b- It's still very nice to have intercom system, monitors for a director to preview all the cameras at once. Though you don't necessarily need the latter if your camerapeople are good and you work out a system where you always know you have coverage (e.g. leave one camera on a tripod on a stage shot).

2- For editing...
It might make sense to capture Prores for some work so that you can deal with multiple projects on the system at once. Depending on how your curriculum is setup, there may be crunch periods where there are multiple people trying to online their project at once.

16GB of RAM might be a little much?

3- So, as I understand it, you're trying to put together a studio for live switching work?

Some considerations:
You should probably look at getting a tape deck, otherwise how would you archive your projects?

Monitors... you can probably get away with something cheaper and/or smaller. Or, some mixers can preview multiple sources at once on one large monitor. Ross and For-a might be some manufacturers also worth looking at.

It might be worth putting some cabling infrastructure in place...
Raised flooring to run cabling underneath. (Though you can run it in the ceiling too.)
You need to run video and audio cables that carry the picture, sound, and sync. You can also share decks this way by running them out of an equipment room.
It's also ideal to run some computer networking infrastructure if you ever move towards shared storage.
Patchbays to help manage the cabling.
Most studios are setup this way because it makes things more efficient for them. For your situation it may or may not make sense. Ideally, it's good for the students to know how to operate the patchbay since it's good to know for entry-level positions (e.g. assistant editing).

And then there are some other considerations like...
Audio...?

Titler?
Still Store? (Maybe these are out of fashion now.)

Playback VTRs... for pre-packaged elements.

Teleprompter?

4- There are some systems integrators that deal with putting together studios and such. They would know much more about putting a studio together.

For fancier studios, it gets a lot tricker. If you have upconverted sources for example, suddenly you need to start delaying everything to stay in sync.

[Jacques Mersereau]

Hey Glenn,

I just wanted to start by thanking you for taking a lot of very
precious time to help me out. I very much appreciate it.

>If you want to do live switching, then you will need either:
>A- Genlock all your cameras.

The Canon's spec'd have genlock and timecode in/out. I am wondering
whether I would need a tri sync generator.
Some of the new switchers have genlock out on them, so yes, a DA to
distribute that around to everything would probably be a good addition.

<Personally I'm not a big fan of live switching... you need a crew of people who are good at it. With students they are generally inexperienced and they graduate before they have enough time to get good at it. Plus, if I were a student I wouldn't be that interested in multicamera work.>

We give class credit for some students and other's find friends to run camera.
I have found that for many performances a live switch can
get a high percentage of the edit accomplished . . . by the end of the performance.
That greatly enhances productivity. Since we have a comprehensive intercom system,
the video director can choose to keep close control of camera ops' shots and direct those
who are inexperienced.
The big preview monitors provide the ability to acquire and direct sharp focus to the
operators and smaller monitors are only a hundred or two less
and usually provide far fewer pixels in addition to the other things these monitors
provide like scopes.

<if your camera people are good>

Sometimes you get folks with natural talent and other times you have to keep right on
top of them. Some of the student directors don't seem to
care. ;-)

<2- For editing... It might make sense to capture (snip)>
Yes, I definitely want each camera to capture to some on board
recording system. That way if a live switch is blown or the client hates the video director's
style, there is another way to reconstruct an edit. For RED it would be a hard drive, for
the Canons it would be HDV to tape.

<16GB of RAM might be a little much?>

No. Talking to some of our graduates who are now working for Apple,
because there are 8 processors working, that is only 2gig per processor.



<Some considerations:
You should probably look at getting a tape deck, otherwise how would you archive your projects?>

The plan (as of now) is to record uncompressed to hard drive.
That can be changed to another codec if the client wants, but
uncompressed means that the switch can later be rendered to whatever
format the client wants to do the final edit in.

<Monitors... you can probably get away with something cheaper and/or smaller. Or, some
mixers can preview multiple sources at once on one large monitor. Ross and For-a might be
some manufacturers also worth looking at.>

Sure, but I really like to see exactly what each camera is outputting and all the monitors spec'd
have certain "value added" features like scopes or can display HD-SDI imbedded audio.

<It might be worth putting some cabling infrastructure in place...
Raised flooring to run cabling underneath. (Though you can run it in the ceiling too.)>

Unfortunately that is not a possibility in our studio.

<You need to run video and audio cables that carry the picture, sound, and sync. You can also
share decks this way by running them out of an equipment room.>

Yes, that is what we have in place now.

<It's also ideal to run some computer networking infrastructure if you ever move towards shared
storage. Patchbays to help manage the cabling.>

A shared fiber channel SAN is going to have to wait for the next step, but we do have fiber cable in place.
Patchbays between studios and master control I installed last year.

<Titler?
a) Still Store? (Maybe these are out of fashion now.)

b) Playback VTRs... for pre-packaged elements.

c) Teleprompter?>

a) Some switchers have still stores and we have lots of computers.

b) We have several HDV decks and more computers. I would like a
pro HD tape deck, but those suckers are BIG money. We do have a digital betacam SD deck however.

c) We have a teleprompter.

<4- There are some systems integrators that deal with putting together studios and such.
They would know much more about putting a studio together.>

Yes, although many vendors seem to have problems with these very
questions.

<For fancier studios, it gets a lot tricker. If you have upconverted sources for example, suddenly
you need to start delaying everything to stay in sync.>

Good point, but since we have pretty good control and good relations with our clients we will
do that work in pre production.

Thanks GLEN!!!!!!!!!!!!

[Glenn Chan]

For monitoring each camera individually, you might be able to setup a station with a single monitor (and possibly scopes if desired) with a switch (or video router??) to switch between the cameras and other video sources. Probably loop the signals through your mixer to this station.

Then you could potentially use HDLinks with 1920x1200 computer monitors for preview monitors.

2- Sync/delay issues:

A LCD monitor will likely impose a signal processing delay when de-interlacing the image... the amount of delay varies from deinterlacer to deinterlacer. It should be worth asking what that delay is, and figuring out whether or not you want to delay the control room audio to match.

3- There may be a vendor which makes an add-on pack for certain cameras for studio use so they get tally lights and such. I can't remember the name.

4- Something like this...
http://www.evertz.com/products/7750SRG-HD
will generate stable HD sync. I don't know enough to know whether you need that over something else (which would drift).

[Jacques Mersereau]

<[QUOTE=Glenn Chan;726228]For monitoring each camera individually, you might be able to setup a station with a single monitor (and possibly scopes if desired) with a switch (or video router??) to switch between the cameras and other video sources. Probably loop the signals through your mixer to this station.>

You could, but experience says that it is better to have all cameras
displayed at all times in front of the director. I think your call about LCD
delay makes a lot of sense, so I need to check with the manufacturers to
see if the HD-SDI In/Out is hard-wired pass-through or "reclocked".
If there is a delay in the monitor output, then I might need to send
both HD-SDI camera outs to master control and one hits the
monitor and the other the switcher, though that means that the
cable to the monitor might be sending good signal and the one
inputting the switcher might have issues. I really prefer that the
ONE signal is good to the monitor and then it only has to jump several
more feet to the switcher.

Thanks again Glenn. I will visit that site and take your recommendations
into careful consideration. You ROCK!

[Glenn Chan]

You could, but experience says that it is better to have all cameras
displayed at all times in front of the director.

For that you can use cheaper monitors like the HDLink right? And depending on what computer monitor you use, you can adjust the color temp on it to get all the monitors in your control room with matching white point (just eyeball it).

You don't need waveform monitor on those monitors since you'd never switch to the waveform view.

So you would do something like:
Camera (4) --> Switcher --> monitors(4) (for director) --> (switch) --> engineering station/whatever you want to call it (1). (If you had a video shader, they would use this station.)

2- The delay in the LCD (that I was talking about) is between when the signal arrives and when the signal is ready to be displayed. I don't think the HD-SDI signal going through it gets delayed... it's when the picture appears that gets delayed.

[William Hohauser]

One of the problems with many of the new HD cameras is integration into a live studio situation. Most of the low end models have limited gen-lock capability and even more limited monitoring abilities. The traditional studio camera is connected by a multi-core (or triax) to a CCU unit that has multiple outputs for monitoring, switchers and single camera back-up recording. The lowest priced HD camera that can do this is the JVC HD-250. However the CCU is limited in that it can only output analog signals, fine for monitoring but if you want to connect that HD-SDI signal you'll have to run a separate line. The new Panasonic HPX500 has a remote unit that can output one SD video signal but all the other signals (genlock, SDI, component) have to be run separately. Feh, excuse my Yiddish. You can set up a decent SD studio at under $20,000 per camera, HD I don't think so.

The next issue is your switcher. Many interesting low priced HD capable A/V switchers, the Grass Valley Indigo for example, have limited HD inputs. If you are expecting to have 4 HD cameras switched at a time, you'll have to go to a real studio switcher. Ross is a company to look at.

[Jacques Mersereau]

<You don't need waveform monitor on those monitors since you'd never switch to the waveform view.>

But if one had a funky looking image, you could switch to the
waveform and check the signal's integrity and I am a big fan
of 1:1 LCD pixel rendering.
Good stuff if you have the dough to blow. :) We can always
reduce things if our budget is downsized.

<So you would do something like:
Camera (4) --> Switcher --> monitors(4) (for director) --> (switch) --> engineering station/whatever you want to call it (1). (If you had a video shader, they would use this station.)>

Yes, and if we go with Canon H1s, we could extend
firewire and have complete control of all the camera's functions
using CONSOLE from master control. That would be very cool.

<2- The delay in the LCD (that I was talking about) is between when the signal arrives and when the signal is ready to be displayed. I don't think the HD-SDI signal going through it gets delayed... it's when the picture appears that gets delayed.[/QUOTE]>

Delay is something to test and then determine how to proceed.
I would hope that the delay on a good monitor when
measured would be less than 10 milliseconds.

The switchers in consideration at this time are the Grass Valley Kayak,
the Datavideo SE1000 (five HDSDI inputs) and the Panasonic
AVHS300G (also 5 HDSDI inputs).