Welcome to our community

Be a part of something great, join today!

Register 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 :)

Dream Red Rushes Feature

M

Mark L. Pederson

REDuser Sponsor
Joined
Dec 28, 2006
Messages
7,145
Reaction score
1
Points
0
Age
59
Website
www.offhollywoodny.com
Here's one for you Deanan -

What do you think of this -

For the post houses and boutiques -

Assumptions:

a) you have your Redcode media on a SAN
b) you have more than one worksation accessing that media over the SAN, and all of those workstations have RedRushes installed on them
c) You have fast storage

First -
Let's have a folder (location set in preferences) that holds RSX look files - so that you can drag ones you have made there, or make new ones in RedAlert and put them there.

In RedRushes, if you don't choose "use metadata" - you get a dropdown of all the RSX files in the folder.

You open RedRushes, import all the Redcode you want to process.
You HIGHLIGHT various clips (based on the THUMBNAIL that RedRushes generates on import - did I mention the thumbnails??? and SURE, it would be cool to have a checkbox for "create thumbnail" - so you can opt. out of that if you want.

Let's assume you make the thumbnails. You highlight multiple shots - access the Look Source tab - pulldown and select a RSX file - and that gets applies to all your selected shots.

Now, there is a checkbox for "process with multiple workstations".

You check it, and it drops a menu down of every workstation on the network.

You select all the workstations that you want to use for the render. When you start the render, it splits the file list between the workstations proportionally - and each of them runs RedRushes - processing their portion of the media - to a defined, shared output folder.

Unless I am missing something, I am pretty sure you could kick Qmasters ass - Compressor isn't exactly the most "optimized" software on the planet.

Anyway, it would be pretty awesome to automate the process of splitting up one big batch render job over a SAN - and have the most optimized processing possible - which I am pretty sure would be running the app independently on separate workstations.
 
Completely agree with you Mark, that would be a dream come true. At the moment I am running around to all the edit suite at night and manually managing the files from the San. I spend at least an hour setting all the suites going!!

Cheers, Chris
 
Even just being able to create a master Redrushes file, and selectively render parts of it manually on different machines wold be awesome
 
If we all pitch in a buck or two, we can send Mark to programming school so he can make the changes to RedRushes... fast. I didn't know before I read his request, but apparently I need all of his requests - brilliant suggestions.

Nick.
 
Suede said:
If we all pitch in a buck or two, we can send Mark to programming school so he can make the changes to RedRushes... fast. I didn't know before I read his request, but apparently I need all of his requests - brilliant suggestions.

Nick.
Click to expand...

Agreed Nick, Deanan did the RedRushes over weekends and at the expense of his kids' time - a huge huge sacrifice already. I prefer he concentrates on getti ng the last 1% on B16 to release. I will pitch in the buck for Mark - Lim
 
Qmaster is actually just distributing Redline commands... so the pieces to do what you're asking for are already there. All that's required is the code in RedRushes to select an RSX file, and to search/select which Qmaster clusters you are sending to.
 
SOME type of network rendering is desperately needed and Redrush is the perfect app for it. Add me to the list for this request.
 
the OTHER dream feature -

the OTHER dream feature -

also Deanan -

here's an important one -

under "Phase 1: Render" - please add a checkbox for Dynamic Debayer

what this would do is the following -

let's say your client wants all Redcode to HD - and you determine,based n the material, that extracting from the 2K debayer is fine. (there IS a difference between extracting from 4K debayer and sampling to 2K/HD vs. debayering from the 2K layer - but you need to have some very serious detail - and a very serious monitor to see it)

you would select Half Res

HOWEVER, in the project, there are off-speed shots in BOTH 2K and 3K.

so when it hits a 2K clip, INSTEAD of using the 1K layer - the Dynamic Debayer function tells the app to use Full Res for 2K clips AND 3K clips.

If you selected Quarter Res - it would use Half Res on 2K and Half Res on 3K

Make sense?

I think for this function - we can just have RR assume the project is 4K - as that is by far the most common. And it is VERY common our 4K jobs that there are 3K and 2K shots here and there. Maybe just make it 4K Dynamic Debayer

We have been bugging the Assimilate team add this into Scratch - because you need to do some "workarounds" for mixed RES with a half debayer output node, etc.
 
twlim said:
Agreed Nick, Deanan did the RedRushes over weekends and at the expense of his kids' time - a huge huge sacrifice already. I prefer he concentrates on getti ng the last 1% on B16 to release. I will pitch in the buck for Mark - Lim
Click to expand...

LOL!

Save your buck. Life is too short for me to code.

Besides, I'd rather just spew ideas and have the rockstars with skills do what they do best -

I do intend to hire a "Code/Flash Ninja" full-time at Offhollywood - but that is slated for a Jan. hire - then - I'll drive that poor person crazy with endless ideas and Deanan can get a bit more sleep!!!
 
Trevor Meier said:
Qmaster is actually just distributing Redline commands... so the pieces to do what you're asking for are already there. All that's required is the code in RedRushes to select an RSX file, and to search/select which Qmaster clusters you are sending to.
Click to expand...
Well .. I could be wrong here - but distributed rendering via Qmaster by selecting clusters - is NOT the same as having individual versions of the application independently processing -

I just want the application to automate and execute the process of splitting up the batch - I want to- for example - have 300 Gigs of Redcode in RR - and select 3 workstations on the network and set the destination folders(s) for the workstations - and press ROCK - and (although PROCESS would be fine too) - RR starts the RR applications on each of the three workstations - and gives them each 100G's of the batch to process -

My issues with Qmaster are - when you do get it humming - it's crazy awesome - but you can spend almost as much time setting it up to get it humming that you save - and I know for a fact that I am not alone in this -

Deanan has significant network render knowledge - he will know the best way to crack this nut -

MAYBE it's Qmaster - but I suspect there's a easier, faster way automate splitting a batch to multiple workstations on a SAN -
 
For those that need an extremely robust, crossplatform (Linux, Windows, OSX), software agnostic render queue management system, may I strongly suggest RUSH (Deanan can probably tell you a bit about RUSH as well).

It's been used in high-end vfx pipelines for many years, and is a script based system. You pretty much have to set up your own scripts for any customization, but there are some sample wrapper-scripts included that can get you started for some of the more popular rendering tasks (not Redline... yet).

It is not necessarily the easiest to use, but it is one the most powerful render distribution software I've had the pleasure to work with. With this software you can easily customize a facility-wide render pipeline for any task - from Redline renders to Maya renders to simple distributed file copies.

I've run it on as few as 3 systems, and as many a 600+ cpus at a time with multiple stage render dependencies from different software packages and custom TCL, Python, and shell scripts all rendering on all three platforms simultaneously with error correction, automated frame sequence/header parseing, automated dropped-frame pickups, etc... sky's the limit!

RUSH could really do wonders re. render management for a facility like OFFHOLLYWOOD...

For those wanting to check it out, go to the website, contact Greg (the author), and get a rental quote - it's very inexpensive to try that way.

Perhaps Deanan would be kind enough to write a RUSH wrapper script for Redline that implements the requested functions?

FYI, I do not sell or work for Rush or have any financial relationship with Greg or Rush, I'm just a satisfied user who would love to see digital cinema post benefit from the kind of extensibility and power that visual effects pipelines have been exploiting for many years - the similarities between the two disciplines are greater than they might seem at first glance.


cheers,

john t.
 
Offhollywood said:
I just think there's an easier approach -
Click to expand...

I agree that 'easier' can be better depending on intended use.

If I'm an individual banging out smaller quantities of footage/shots then simplicity is probably where the efficiency balance lies.

But if I'm turning around any sort of volume in a general digital imaging pipeline, then flexibility and scalability are my top priorities.

My perspective on this comes from having worked in moderate and large scale digital imaging pipelines for feature visual f/x, where cumulative render time can add up to astronomical turnaround times, and not having to manually flip through every f#@*ing frame for 1st pass QC and render re-submissions can save not only weeks/months of manhours, but also tons of $$$ in salary spent elsewhere.

cheers,

jt
 
Nice thinking Mark, thats a great idea, just have a REDrushes server on one machine find REDrushes "client" software running in background on all systems, it could be any system on the network. even wireless for that mater. Have a few Mac Book Pro's in the studio let them join in via wireless while you keep working on something else.

Now if we could get this to be cross platform. :)
 
You must log in or register to reply here.
Back
Top