Is Outboard Redcode a possibility?

[Thom Steinhoff]

Okay, so the camera can't process redcode fast enough to handle 60fps at 4k, but what about if we could pump up to 60fps out the RAW Port, directly into a computer running Redcode that is fast enough to process 60fps into Redcode?

Of course you are developing Redcode on a PC already, and choosing the speeds of your chips for heat/cost/performance, etc. But what about selling the software to run on a monster outboard or even an selling outboard box for those who want to crank 4K but don't want to sacrifice the flexability and ease of use of redcode?

Run normal, to the normal drives, at 24fps until we need that Killer 60fps shot, Rent a "Red Monster" box for the day that cranks 4K Redcode at 60fps and everyone is happy?

Idea inspired by trying to figure out a solution for Tom--Us Indies have to stick together.

[Brook Willard]

I've often wondered if an outboard box could be created to compress 2540p into REDCODE RAW and dump it onto a mini-RAID. I would love to see one come to market at some point.

[Thom Steinhoff]

Thinking about it some more, I came up with more "details"

Call it "Redcode Raid Buffer" -- It is an outboard raw to Redcode processor that takes standard Red Raid packs but also has inside of it say 5 minutes (or even 1 minute?) worth of high speed raid cache at Raw speeds. Assuming the RAW Cable also has a metadata/control track you set your default speed on camera and ramping up to 60fps. Once you go, the footage is cached in the Raid buffer until you stop. Now the box processes it, tags it with the metadata, puts it on the RedRaid pack and clears the cache--maybe the whole process takes 2x shooting time to clear the cache--so what?

This way, you can be shooting all day at normal speeds with the Red Raid, get to the ramped scene, hook up the Redcode Buffer box, transfer the Red Raid to hang off that box, shoot your highspeed scenes, then transfer your Red Raid pack back to the camera, disconnect the buffer box and keep shooting normal speeds.

What do you guys think?

[MikeCurtis]

My thoughts on this from a few months ago when they announced the specs - if 4K30p is all that can be done by the onboard compression stuff, why not put two of those in an outboard box connected to the high speed RAW port? The outboard box would need whatever flavor of I/O port that the high speed RAW port has, plus a couple of whatever the hardware is to do the compression, then enough drives to capture that datarate. 4K@60p should be around 70 MB/sec with the old numbers that it sounds like are being improved upon. 4 or 5 laptop drives could handle that - perhaps an eSATA port multiplying interface? Or does that rely on too much brains for the host? If not, a little hardware RAID config.

Hmm....call it the High Speed Red Drive?

As opposed to RED RAID?

OR....more crazy science project ideas....if splitting the load between two chips is somehow problematic, or it helps keep it more affordable, what about compressing every other frame onboard and sending that compressed data down the eSATA bus (call this the even frames), and then sending the odd frames uncompressed down the high speed port to the High Speed Red Drive, where another compression chipset resides, just like the one onboard Red One. It compresses the odd frames, and then muxes those in proper order with the even frames that all get written to the 4 or 5 drives. Is this second way worth it? My gut says the complexity could outweigh the benefits. I guess only if the cost of the high speed port stuff is markedly less than the cost of the compression stuff.

Just some fun thoughts, dunno how practicable any of that is.

-mike

[Nick Shaw]

Nice ideas Mike. One potential issue. Isn't the high speed port 2540p (up to 60 fps) or 1270p (up to 120 fps) only? REDCODE (currently) only supports 4k. I suppose the hardware in the camera is windowing 4k from 2540p before compression, so the High Speed Red Drive could do the same. Makes it a bit more complex.

Thom's buffering idea also sounds possible. It was something I had been thinking about too. That would enable 2540p REDCODE, which is something everyone would like.

Nick

[Brook Willard]

I imagine that part of the reason that we aren't able to do more in camera is that the amount of data bandwidth and processing power necessary to do this is either prohibitively expensive or nonexistent. Just a thought.

[MikeCurtis]

Nick & Thom - RAID cache=RAM. At up to nearly 1 GB/sec, one minute of 2540p60 would be 55-60 GB of RAM, five minutes 250-300 GB of it. Eeeyouch! Pricey! But in theory that could work, you just have a limitation on how long of a take you can do. But then yeah, could crunch it in non-realtime and write it to whatever. Just a matter of what's cheaper - RAM or compression chipsets...cropping 4K from 2540p doesn't seem like too big of a challenge to do prior to the data going down the RAW port - it already has to be able to window for 2K work anyway.

[Nick Shaw]

I was thinking more of a possible option for RED-RAID which as well as recording uncompressed RAW, had the option to post-process in the background to REDCODE, thus allowing 2540p60 and 1270p120 recording, but making much more efficient use of the disk space. The uncompressed RAW could be automatically deleted after conversion (or optionally not). Admittedly that would mean RED-RAID would need a CPU onboard, but it would probably need one anyway for the RAID controller.

Nick

[Thom Steinhoff]

Mike,

I think a hybrid of what we both talked about would be perfect. Dual chipsets, compressing every other frame (great idea) to an internal raid array--then, when the camera stops, streaming the footage to an attached standard Red Drive (Or Red Raid) for use and clearing the raid cache for reuse.

That way, since we only need 60 gig of high speed raid for a 1 minute buffer (do you really need more than 1 minute recording bursts at a time?) would be x number of smaller laptop drives that would be a lot cheaper than the high capacity ones (you are far better at that math).

I agree, in a few years when ram drives are reachable price wise then that would be a better solution as it would keep it somewhat solid state, but in the mean time, someone has to do something with the 40 gig laptop drives!

I love the idea of full speed, I just hate the idea of dealing with true RAW. Even if you only use it for the few shots you need it--then you have to deal with having that footage on a different workflow.

[John Cordell]

One thing I've not seen mentioned is the set of formats that can be written to REDFLASH. If 4k60p can be written onto REDFLASH, then an easy workflow would be to have the camera convert raw to REDCODE at 1/2 real time (from the REDFLASH to the RED-DRIVE.) Or use REDCINE to do the raw to REDCODE conversion, which free up the camera and quite possibly out-perform it as well.

Not sure if REDFLASH is expected to be fast enough, or if both REDFLASH and RED-DRIVE can be connected at the same time. Either way, having REDCINE be able to convert high frame rate raw to REDCODE would be very useful. If I have to capture raw to get the high frame rate, I would still want to compress to REDCODE in order to save space and normalize all my footage.