Ideal Pixel Box
Published : 5th July 2004
I have been doing some research into this subject and many have fallen short of my practical expectations of durability, Mobility, and simplicity, along with a simple yet cost effective plan for transport and storage to and from post.
Basically what you are looking at is a Tapeless Digital Deck, A box that collects Pixels actually Data. I realize that this device will have application in all places that Decks currently see employment. Production, Post, VFX, Downconvert, Duplication, Transmission, Projection, etc. Can or should one box for one project be used for all uses in that project, is this at all realistic??
Most work on Linux with HD cards to input data into big fast hard drives making DPX files. which are packets of data. At least as I understand it.
So I am still on a Hunt for the Ideal Pixel Box.
Be nice to see Dual Stream 10 bit LOG 4:4:4
One stream 4:2:2
One stream 0:2:2
One Stream 4:2:2
One Stream 4:2:2
Able to be used by Viper, Sony 950 and for 3D configurations synched into one Box with Sound.
Removable Data with enough Storage to realistically shoot 2 cameras all day then get the Data to Post or where ever storage will be going on. Its nice to be able to Edit and color correct using the same box on the set, but That's not what I am paid to do. And as long as it doesn't slow me down and always works, I don't care what other features it has.
What Boxes exist that can pull off such a feat.??
Now is the Time Gents there are smart Guys out there listening to our suggestions, What will it be.
I realize the best non linear solution for all around durability, storage, and cost may be...dare I say...Tape
B. Sean Fairburn
HD Director of Photography
Doesn't the S-two DMag already do these things you just described?
BTW, it has a network connection on the box, or an alternate box with network connections so you don't need that HD-ingest card and can have access to the DPX files on any platform that supports network protocols.
Post Production Artist
Virginia Beach, VA
>Doesn't the S-two DMag already do these things you just described? >BTW, it has a network connection on the box, or an alternate box with >network connections so you don't need that HD-ingest card and can >have access to the DPX files on any platform that supports network >protocols.
All of that is correct.
GEORGE C. PALMER
Take a look at www.stwo-corp.com ... the web site contains the basic info on the capabilities of their "pixel box".
I've used it on several shoots, and it has all of the capabilities described and exceeds some of those capabilities such as flexibility for ingest to post and access to DPX files.
GEORGE C. PALMER
At MotonFX in the UK they use an s-two and Quantel iQ combined with their Viper camera, with the iQ in a flight case so it can be used on-set to initiate editorial & grading. It looks to me to be the ideal combination and is presently being used in northern Italy on a promotional project.
Digital Praxis Ltd.
>Viper Sony 950 and for 3D configurations synched into one Box with >Sound.
Strange that you didn't bring up the S-Two DMag. I thought of you while it will not record single deck 3D (two cameras in 4:2:2 mode) yet, S-Two told me that they are working on that. Otherwise, to echo the other responses, it has removable "magazines", will record both single and dual link (and therefore will work with the two cameras you mention in either configuration). You can look at one at Plus 8.
Cobalt / 3ality
Steve Schklair wrote:
>While it will not record single deck 3D (two cameras in 4:2:2 mode) yet, >S-Two told me that they are working on that.
Maybe they should get it to do basic things like read external time code first.
BTW, I just got finished working with your brother on the "Point Pleasant" pilot. A great guy. Say hi for me.
IATSE Local 600
Sean Fairburn writes:
>I realize the best non linear solution for all around durability, storage, >and cost may be...dare I say...Tape
Could be (though tape is hardly non-linear!). But transfer speed remains a serious bottleneck.
Can anyone offer a reason why it's impractical to multiply current tape speeds by, say, 4x? Would the heads overheat and/or burn the tape? Surely there must be a way around this.
Of course, if you're dumping lots of data you can always run several systems simultaneously.
And if you're only "printing" selected takes, that further reduces your transfer time.
Marin County, CA
Hereâ€™s my input for you :
As I know there are four systems available, and most of them do MOST of the features you asked for - maybe we'll see more at NAB:
- Directors Friend (stopped production, only rentals available) did single 4:2:2 and dual 4:4:4 and would have done dual 4:2:2 with newer software / development
- S-Two http://www.stwo-corp.com (please correct me if Iâ€™m wrong!) does single 4:2:2 and dual 4:4:4 but no dual 4:2:2
- Keisoku Giken http://www.keisoku.jp/htmlE/vw/udr2e.htm (please correct me if I'm wrong!) does single 4:2:2 and dual 4:4:4 but no dual 4:2:2
- Screen disk http://www.digitalvideo.de (please correct me if I'm wrong!) does single 4:2:2 and dual 4:4:4 but no dual 4:2:2
All of them capture lin or log files - for example with the viper (log) or F950 (lin) and all of them capture sound, mostly via AES Inputs.
Almost all of them have removable drives or can attach removable drives and can/do record/export dpx frames.
Colour Correction is only available on the DF and the Screendisk (please correct me if I'm wrong!); Editing only on Screendisk.
The only dual 4:2:2 device will be HDCam SR, haven't seen it work but believe it will... Didn't Steve Schklair already tested it for 3D?
But you should check these systems not only on tech specs, see also the way to capture on set (for example start/stop recording from camera, user interface), the integration in your production and postproduction workflow (Timecode Input, Metadata integration, Timecode in DPX Header, Backup...), mechanical specifications and dimensions, AC/DC Power supply, Raid levels (security) and so on...
I captured some 4:4:4 shots with the df and now IÂ´d love to play with the "final" sTwo and the "final" Keisoku soon...
What I like :
On the df I like the capture interface (software called MUNGO) because its simple but powerful and the timecode to dpy header works now good. On the Dmag I like the compact unit and the heavy duty housing with small mags. On the Keisoku I like the raid 5 level and on the screendisk I like the ability to use all kind of software from combustion to speed grade on a open system.
Hope that helps...
+++ Florian Rettich +++
+++ Europe based DIT / vision control +++
With regards to color correction on the S2 products we are working to qualify our software on them to provide realtime grading straight off the Viper/950 captures.
This is combined with our color management plug-in cubes so you can achieve an accurate representation of the sequence as it would look on the target medium, film or HD. We've tested realtime 2k with cubes from ARRI, Cinespace, Filmlight, Imagica and Kodak.
You can also send the grade bin back to the post house along with full resolution reference plates for each of the shots to preserve the 'look' you were trying to create.
Dan Drasin writes
>Can anyone offer a reason why it's impractical to multiply current tape >speeds by, say, 4x? Would the heads overheat and/or burn the tape? >Surely there must be a way around this.
I keep trying to forget about tape but it won't go away.
Unfortunately just speeding up tape by 4x won't work because at the data rates we are interested in most of the head to tape speed comes from the rec/play head whizzing round and round (helical scan) not from the linear tape speed.
The limit on speeding up the head speed is to do with the shortest recorded wavelength and to push that on requires new tape formulations.
The way you up the data rate is by running more recording channels in parallel inside the tape deck, I'd give examples but here my memory is failing. To make space for the data you increase the linear speed. Now Mr commerce enters the debate - more channels equals higher capital costs and higher running costs and more tape means less on a cassette and higher tape costs. It can be done - Voodoo D6 manages non compressed 1 Gb/s or so recording but it's hardly mainstream.
The alternative linear data recording solutions are way cheaper than helical scan and their data rates are increasing but they are at about 1.5x SD video rates now so have a way to go. Do a search for LTO2 Ultrium or SAIT for more.
Post person at Quantel, ex large Japanese VTR maker - no the other one!
>The only dual 4:2:2 device will be HDCam SR, haven't seen it work but >believe it will...Didn't Steve Schklair already test it for 3D?
Sony tells us the SRW field recorder will record dual 4:2:2 streams for 3D work, but I have not yet had the opportunity to play with one. Would very much like to though, as it removes a layer of complexity from a 3D shoot. Not to mention the need for half as much raw stock.
For the 3D shoot at the Super bowl, we chose tape (SRW) over disc recording for reasons of the sheer amount of data we would be capturing with multiple camera systems. The capture was just part of the decision, backing up the data was the other part. Running two discs per camera for redundant back-up was out of the question, and we were not going to let the truck leave the stadium until a duplicate copy of the material went off-site. We ran dubs all night immediately following the game, so that we could pull out the next morning.
BTW...I also worked with the Directors Friend and Florian during one of the early sports shoots, and owe them an apology for not including them in the last post. They were great to work with. The biggest issue we had was backing up the data after the game, and then storage methodology. The only solutions to this were expensive, as we could not make tape or standard raid disk back-ups in real time.
Storage could be an entire new discussion thread. Our workflow is designed such that a tape in hand is still a tangible item. It contains "X" number of shots. When we are looking for a shot or a scene, it is easy to pull the tape. When we are done, we archive the entire box of tapes with a label on the outside of the box indicating what scenes are contained within.
Disc recording is a great way to record uncompressed full bandwidth data, but there is no tangible physical object to put in your hand with the shots you need. Also, digital copies of shots seem to proliferate faster than rabbits, and suddenly you have a dozen copies of the same shot on multiple drives. This should not be the case, but the discipline to track every copy made of the data, and then to follow a very complex file naming procedure is not yet SOP.
However, always pick the right tool for the job, and while we are very happy with the results from the SRW recording, on another project with other needs we may choose to record to disc.
Cobalt / 3ality
>Unfortunately just speeding up tape by 4x won't work because at the >data rates we are interested in most of the head to tape speed comes >from the rec/play head whizzing round and round (helical scan) not from >the linear tape speed.
I'm confused also. Sony sold a 1" digital uncompressed VTR for several years also, And I seem to remember demonstrations of others. It's more a cassette convenience and overall cost issue, not a technical one, to record 1.2 Gbs. Also, their main goal is to make a shoulder mounted news camera, not something useful for everyone else.
>The limit on speeding up the head speed is to do with the shortest >recorded wavelength and to push that on requires new tape >formulations.
Or wider tape...
They made 1" field recorders in the old days before Betacam...
Steven Bradford wrote :
>I'm confused also. Sony sold a 1" digital uncompressed VTR for >several years also,
I remember Sony's compressed tape format at the time (1995/6), but the only 1" decks they used to have at the Culver City and Basingstoke 'edit suites' and they were big boys; fridge sized machines. Perhaps there was something more recent that I'm not aware of for the field.
Steve Schklair wrote :
>"Sony tells us the SRW field recorder will record dual 4:2:2 streams for >3D work, but I have not yet had the opportunity to play with one.
I assume that the VTR runs the tape twice as fast. So just as much tape as running two other machines. Also there is no playback VTR other than the model these tapes were recorded on. So Sony's stated plan is that you playback on the SRW-1 and record to a disk array for editing. The SRW-1 is not an editing playback VTR. It is just a capture device.
>They made 1" field recorders in the old days before Betacam...
And 2" ones as well.
I lugged an Ampex 3000 quad machine around for about four years when I first started in the biz (ca. 1971).
20 minute loads, mono audio, no timecode, B&W playback only, weighed about 80 lbs. and the size of a small suitcase. Ahhh, those were the days...
I must Give Credit to Rob Legato in a meeting with Mike Kanfer (Award winning VFX, Apollo 13, Titanic, Harry Potter, many others)
Robs Idea was to take the 3D camera model (Mirror version) and set it at "0" interocular perfectly aligning the cameras to see the exact same frame and FOV.
Then shoot one exposed for Highlights one for Shadows to create a much greater dynamic range. You could also specifically tune each camera to maximize and support the area of the frame that it is intended to capture not worrying or compromising the other end of the scale.
Yes Film already does this. especially Vision 2.
The cameras could be phased together (Genlocked) so that motion blur on each frame would match both cameras.
You would then put the elements together later in Post
Kind of a hyper Dynamic range capture mode.
Going into a Pixel box would be a wonderful way to test this on site.
Especially for Explosions at night where it is important to see subjects before the explosion goes off. or anywhere where 3-5 stops of difference kicks your butt.
Just another wild thing to experiment with.
B. Sean Fairburn SOC
Director of Photography