Welcome to our community

Be a part of something great, join today!

Log in
  • Hey all, just changed over the backend after 15 years I figured time to give it a bit of an update, its probably gonna be a bit weird for most of you and i am sure there is a few bugs to work out but it should kinda work the same as before... hopefully :)

Red transcode workflows for people in a hurry

C

chlowden

Banned
Joined
Jan 2, 2010
Messages
6
Reaction score
0
Points
0
I have 7 hours of RED RUSHES to transcode to DNXHD 36 MXF (it has to be MXF and not QT). After 10 hours of transcoding on a dedicated Mac 2.66 ghz 10.5.8, I have not done 3% of the lot. My processors are working hard. The rushes drive is on FW800 and the reception drive is internal. I am using REDCINE-X (Metafuze 2 does not work for my needs at the moment due to SRT issues). I have had better results using REDRUSHES and QT but I can't do this, this time.

So I am wondering, how do others do it? How do those transcode labs do it? What techniques do others using? Is the RED ROCKET card much faster for transcoding? Has anyone put a render farm in place using REDCINE-X? Are there tools available to optimise my mac's processors for the render?

So many questions, so little time. Any suggestions / ideas are very welcome.
 
We use Clipfinder on a few Mac Pros with 8 cores and 16 GB of RAM. They are networked and use the PMDR function in Clipfinder. All cores fire at their limit and we get pretty good times. But then, it's Quicktime…
 
  • Thread starter
  • Banned
  • #3
What render times are you getting?
 
Around 1:10 on a single machine with half-high. These are older Xeons, not Nehalems.

Throughput of the whole system depends on clever load balancing, in Clipfinder you can adjust how many cores are allocated to each task. Our three machines scale very well to around 1:3 , since Clipfinder's distributed rendering relies on dummies for exchange, not shuffling the whole data around all the time. We keep the results local while rendering and later collect with a fast RAID, since our network is slow.

Full de-bayering takes much longer, but is only needed for critical low-light scenes.
 
Redrocket & Scratch is how we do it - but that's to QT. All Avid customers I deliver to accept QT - why does it have to be MXF?

Anyhow, with that setups we get faster than realtime for 1/2 debayer, realtime for full debayer. Quicktime doubles the rendertime, but that's a quicktime problem, not really Red/Assimilate's bailywick. All formats supported - 4.5k, anamorphic, etc...

cheers -

JT
 
Why not use REDrushes, do the import yourself into MC (30 day trial version) and deliver MXF media to the client? Your times are very slow - sounds like you are doing a full debayer which is wasted on DnxHD36 for offline work.

Michael
 
yeah, i think so, it looks like you're on full debayer. For offline task, this is way overkill.
 
Q: Is your target internal drive separate from your internal OS drive? Will help speed things up a bit if they are separate.

But there is nothing like the speed of the RR for transcodes. Best investment ever.
MXF support is not far off I believe... but have to agree with other posters.

If its an offline, why no QT's? Why are they insisting on MXF? DO they want to deliver the offline as online ala Prores QUAD?
 
ROCKETCINE-X already does MXF wrapped DNxHD and other than a memory lead in the Avid libraries (for batched over ~30 clips or so) it works great. It is nice to have the option of the hardware to accelerate the R3D debayer process.

Michael
 
Thank you for your replies. Why MXF? I have to re-transcode rushes on an existing project. If the media is in MXF, I can relink automatically. With Batch Import with changed clip names, i have to relink by hand.
My render times are very slow because I was using REDCINE-X MVF export and I thought that the medium button on the debayer menu on the detail window had a consequence on the export. I read in the READ ME that all exports in REDCINE-X are only in FULL DEF. I wish that this detail was in the REDCINE-X manual. It would have saved me alot of time.
I think I will buy a RED ROCKET ...
 
Why can't you relink based on tape name and timecode? See last section in Step by Step Guide on www.avid.com/red

Michael
 
Just fyi... the beta of SCRATCH v5.1, with a RED Rocket card...

Renders to any flavor of DNxHD (ANY flavor) at very close to realtime. This past weekend, I watched one of our beta customers in Los Angeles render 1,500 R3D clips - an entire feature of dailies - to MXF with audio. Smooth process and not a single crash or abort over 23 hours of rendering.

Best,

Lucas

Lucas Wilson
--------------
Director, Business Development
ASSIMILATE, inc.
LA, CA, USA
 
Lucas - that's great! How much metadata you passing through? I am trying to ensure that all metadata is consistent from different vendors as they come in and out of an Avid environment. Trying to match the ALE from ROCKECTCINE-X as much as possible.

Michael
 
MichaelP said:
Lucas - that's great! How much metadata you passing through? I am trying to ensure that all metadata is consistent from different vendors as they come in and out of an Avid environment. Trying to match the ALE from ROCKECTCINE-X as much as possible.

Michael
Click to expand...

Michael,

I'll give you a call - will probably end up w/Tony, but I want to make sure we are as much in line with preferred Avid workflow as possible!

Lucas
 
Hey,
you could use my red dailies service.
pointblankBANNER.gif

http://thecoloringroom.com/reddailies.html
ok.. shameless plug....

I transcode for a local avid house and they have me do dnx175 then they just import that for offline/online. I'm sure it takes a bit to import but it's still way faster than what they can do. They were saying they did not want mxf because of some issues with relinking audio. But I can do mxf as well.

Anyway, get a rocket = all work flow problems solved.

j
 
You must log in or register to reply here.
Back
Top