Can you shoot uncompressed?

[Brian Duke]

I searched the forums, but I couldn't find an answer whether or not you can shoot uncompressed, and if so, how to do it? Is it available or will it be available?

Thanks D

[Adam Grunseth]

Yes you can shoot uncompressed with 4:2:2 color space.

How you go about it depends on which camera your using, but basically you need an off camera recorder (like a powerful computer connected to a sufficiently fast RAID). You need a capture card in the computer capable of recording uncompressed HD, and finally you need a way to connect your camera to the capture card (with the HD100 you would use the analog compenent outs).

[Jim Press]

Brian,

I am very, very new to video production so please take what I say with a grain of salt! We're looking at getting up to 3 HD251e's because of the HD-SDI output--it's the only way to capture uncompressed. You'll need a grunty machine to capture it to-- see Mike Curtis' excellent article on suggested uncompressed workstations http://www.dv.com/features/features_...leId=196602702.

My understanding is that "uncompressed" with ProHD gives you a 4:2:2 10bit signal. If the data rate is just too massive for your system to cope with (and it's unrealistic to drop $15-$20k on a new workstation) you could run it through the AJA Io to downsize the file to ProRes422 while still maintaining relatively uncompressed footage quality (720p24f in ProResHQ is approx 38GB/hr). Obviously if you're on location (like a lot of your shooting Brian) being tethered to a RAID is going to make it tough.

Hopefully someone much more knowledgable than me will help out! BTW really enjoy seeing your flicks--we laughed a lot at The Date!

[Jack Walker]

Originally Posted by Brian Duke

I searched the forums, but I couldn't find an answer whether or not you can shoot uncompressed, and if so, how to do it? Is it available or will it be available?

Thanks D

With which camera?

They all "shoot" uncompressed, but only the 250/251 will let you get out of of the camera digitally, through the SDI.

The 200/201/110/111/100/101 have uncompressed at the analog component outputs only.

In all cases, how you then take the signal and deal with is the issue.

[Jim Andrada]

Do they really shoot uncompressed, or do they compress to an M2t intermediate in the camera and decompress at the output stage?

In other words, is there any documented evidence from the manufacturer or otherwise equally technically astute entity that the video stream at the output port is exactly what was captured with no intermediate compression and decompression in camera?

I've never been able to get a clear answer to this. I know that I WISH it were true, but...

[Brian Duke]

From reading the posts, I guess its possible. Now my question is could you then edit in FCP6 with it? It seems like the files may be too large. Maybe not. What size is a second of uncompressed HDV footage on a raid?

D

[Jim Press]

Yes FCS2 will edit uncompressed HDV--as I understand it, but I haven't tried (yet).

Re "What size is a second of uncompressed HDV footage on a raid?"

From Mike Curtis' article (cited above): "Even asking, "What's the data rate for uncompressed 24p HD?" can give an answer anywhere from 42 MBps (8-bit 4:2:2 720p) to 190 MBps (10-bit 4:4:4 1080p). Here's the short answer: for 10-bit broadcast work, you need a system capable of sustaining 200 MBps across the entirety of the array."

Again, you could consider compressing the uncompressed data to a "lossless" codec like ProRes (see the blurb at http://www.aja.com/html/products_Io_IoHD.html).

We're trying tp find a middle path where one can maximize the output from the 251, but in a realistic budget for an indie. Does anyone have any other recommendations?

[Brian Drysdale]

You could check these:

http://www.dvinfo.net/conf/showthread.php?t=102312

http://www.dvinfo.net/conf/showthread.php?t=87645

[Daniel Weber]

First to clarify a few things.

There is no such thing as uncompressed HDV. HDV is either of two data rates, 25mb/s as in the Sony and Canon HDV cameras and 19mb/s as the JVC cameras use.

These data rates are recorded to tape or DTE.

When you use either the Component or SDI outputs on these cameras when shooting live, the video signal is sent out as uncompressed. I believe that it is 8 bit.

You can use a variety of capturing setup to take advantage of these signals. The problem is that you will be tethered to a computer to do the capturing.

A new option is the new Compact Flash recorder that Convergent Design is coming out with. This unit does not capture uncompressed though. I believe that you can do 50mb/s XDCAM though which would be a 4:2:2 signal.

Dan Weber

[Jim Andrada]

I think the question I have is whether the camera sends out uncompressed data by decompressing previously compressed data or whether there is truly a completely 100% uncompressed stream that passes through the camera and out the SDI path.

In the case of component I'm sure the digital has to be converted to analog, but again I'm not sure whether the digital that gets converted is what originally came in the front of the camera, or a decompressed version of the M2t stream.

[Adam Grunseth]

Originally Posted by Jim Andrada

I think the question I have is whether the camera sends out uncompressed data by decompressing previously compressed data or whether there is truly a completely 100% uncompressed stream that passes through the camera and out the SDI path.

In the case of component I'm sure the digital has to be converted to analog, but again I'm not sure whether the digital that gets converted is what originally came in the front of the camera, or a decompressed version of the M2t stream.

If you are taking a live feed out of the camera's SDI or the component outputs you get an uncompressed signal that never was compressed. However, if you record the signal and play it back via SDI or component outputs it just decompresses the compressed recording.... but as long as you are taking the video stream live it is not compressed.

[Jim Andrada]

Adam

Thanks for the response. This is what I'm hoping is true. But I still have a couple of doubts. Probably because I'm an old computer guy and can't shake my suspicion of how computer logic tends to be implemented. And after all, these so called cameras are actually just computers with a lens and sensor system stuck on the front.

For example, as I understand it, the chroma information is not really sampled for every pixel in the first place, so I wonder where the color space "compression" to 4,2,0 happens. It would seem most logical to do this at the front end of the process.

We do know that anything recorded to tape is already compressed, so it would have to be decompressed before being output from the camera.

This means that there would have to be two separate paths through the camera and it would be different for live capture and for playback

Somehow I wonder if the in camera process might not be to take the images off the sensors and compress them directly to a memory buffer. Then a Tape Write process could look at the contents of the buffer and re-arrange it as needed (inserting error correcting codes etc) and write it to tape. An Analog Out process could be looking at the same buffer and just doing whatever was necessary to put it on the analog out lines, and other output processes (including SDI) could work the same way. It would be very neat because there wouldn't be any inter process communication or alternate data flow or pathing logic, just a bunch of independent converters sucking off the same buffer and outputting the converted data.

It would also be simple for the playback process because the tape input process could just drop the data from tape into the same buffers, and the back-end processes would just continue to look at the buffer and do their things. In other words, in this model, none of the output processes would care how the data got to the buffer, or whether it was live or playback, they'd just act like the display adapters in a PC and shove whatever showed up in the buffer our their respective ports in appropriate formats. It would also mean that the whole set of front end processes like sampling and compression and tape read could be completely common for each camera in the line-up, whether it had SDI or not. And in fact, it would be easy to offer a variety of ouput processes because they could also be common from camera to camera.

I know I worry about this too much! It just bothers me to think that there are two different ways for an image to get to the ouput bus depending on whether it's live or from tape. Possible of course!

[Jack Walker]

From the JVC webpage on the 250:

Uncompressed HDTV live signal output

In addition to providing superior quality HD recording in the HDV format, the GY-HD250 can capture and output both digital and analog HD signals at 60 frames per second. Ideal for live broadcasting, the uncompressed full-resolution signal (either 720p or converted 1080/60i) can be output via the built-in HD-SDI and component analog terminals, providing an ideal feed to a video server, HD switcher or microwave link.

Here is the link to this page:
http://pro.jvc.com/prof/attributes/t...&feature_id=02

If you think about it, the JVC 250 cameras change the video to a number of different formats: HDV, 1080i HD, DV etc.

The signal comes off the camera block to the processors and then goes a whole number of different ways... some at the same time (e.g. uncomressed out the component and HDV to tape) or some only separately DV or HDV to tape.

If you want to talk about what is "stuck on" the unit, it is the VTR. That's why is is called a camcorder. Most studio cameras, for example, only send out a signal--they don't have a recorder. This is how the SDI works, like a standard video camera that sends out a signal to be recorded elsewhere, or to go through a processor for broadcast.

The VTR part of the camera is incidental to the camera's operation. However, to record to tape, a format must be used that is recordable to the kind of VTR used. Just one of many formats outputable. Since the VTR is a DV and HDV recorder, the video must be put into one of these formats before recording. But the main signal is uncompressed (technically not, just 4:2:3) and send to the SDI. The signal (I suppose at the same time) is sent to the component outputs, and to do this it goes through a digital to analog converter, but remains uncompressed.

If JVC were sending a 4.2:0 up-compressed video out the SDI, I don't think it would still be a secret. I think their literature is accurate.

[Jim Andrada]

No question that what comes out the back is as described - only suspicious that somewhere between the lens and the output it gets compressed and then decompressed.

Oh well, I guess the only way to tell for sure is to get one and make a test:<)

Unfortunately the 250 is out of my buf\dget range and I'm thinking of geetting the 110 - so I guess I'll never know.

Sort of like knowing whether the light in the refrigerator really goes out when you close the door!

[Bill Ravens]

FWIW...

The JVC HD110 will output 8-bit, 4:2:0 HDV over firewire or 4:2:2 over the analog component outs. Internal D-A/A-D processing is 12 bit, so the analog output is generated with a 12-bit engine.

The problem remains how to record the component outs. That can be done, live only, to a RAID array. You'll need something like a Blackmagic Intensity Pro PCI bus card to capture the component live, and you'll need approximately 250 MB/sec thruput to the RAID Array. Current state of the art arrays, capable of this kind of data rate are available from manuf. like Dulce Systems. Take a look at theri website for more info on how to configure hardware for this kind of capture.