Network Rendering?

[sigerson]

I was wondering if anyone can point towards a network capable MPEG renderer? Mainconcept, CCE, Procoder, Sorenson Squeeze, Cleaner, etc.? Can any of these do it? I currently use Tsumani which can't.

[twodogs]

Vegas 5 does it.

[jimmalenko]

I don't exactly understand what you mean by "network capable" ? Do you simply mean accept input from other machines on your network and/or output to other machines on your network ?

Please explain

[edDV]

Vegas 5 does it.

Vegas 5 has built in network rendering features but these don't work for MPeg2 rendering due to the licensing requirements. Only the primary machine can render MPeg2.

You can use cluster rendering for effects processing on two other machines with the standard Vegas 5 license and more rendering machines can be added at extra charge. The process isn't automated. You still need to dispatch rendering tasks manually. At least its a start. See this white paper.

http://mediasoftware.sonypictures.com/download/step2.asp?DID=515

I use Premiere on one machine and Vegas on the other so I can render MPeg2 in parallel on both.

[sigerson]

I don't exactly understand what you mean by "network capable" ? Do you simply mean accept input from other machines on your network and/or output to other machines on your network ?

Please explain

Multiple computers rendering the same video. Similar to the way 3D modeling/animation software can render to multiple computers. So instead of one computer trying to encode a video, you'd have multiple computers over a network working on it...speeding up the process. At least, thats the idea.

[jimmalenko]

I don't exactly understand what you mean by "network capable" ? Do you simply mean accept input from other machines on your network and/or output to other machines on your network ?

Please explain

Multiple computers rendering the same video. Similar to the way 3D modeling/animation software can render to multiple computers. So instead of one computer trying to encode a video, you'd have multiple computers over a network working on it...speeding up the process. At least, thats the idea.

Thanks

Boy, am I glad I didn't recommend something stupid

[Soopafresh]

It can be done rather easily, if you're comfortable with Batch files and some scripting. Conceptually, I think it would be much more efficient to give each render node a chunk of source file rather than frame by frame feeds. Four nodes each getting a .VOB file from a DVD comes to mind.

What kind of source material are you looking at rendering ?

[sigerson]

Vegas 5 does it.

Vegas 5 has built in network rendering features but these don't work for MPeg2 rendering due to the licensing requirements. Only the primary machine can render MPeg2.

Doh! Oh well... My hope was to find a way to have multiple computers rendering out a 2-pass VBR mpeg2 for DVD. Ah well. I did find one other possible solution. It's called Vidomi, but it seems to be more specialized towards Divx/Xvid encoding. Nothing wrong with that, but like I said earlier, I was hoping for something in MPEG2.

I'm still intrigued with Vegas, though... Speeding up a Quicktime/Windows Media/Realplayer render is always a good thing.

[sigerson]

It can be done rather easily, if you're comfortable with Batch files and some scripting. Conceptually, I think it would be much more efficient to give each render node a chunk of source file rather than frame by frame feeds. Four nodes each getting a .VOB file from a DVD comes to mind.

What kind of source material are you looking at rendering ?

Sorry, I should've been more clear...it's for creating the mpeg2 files necessary for the DVD authoring. Source being DV.

[edDV]

Vegas 5 does it.

Vegas 5 has built in network rendering features but these don't work for MPeg2 rendering due to the licensing requirements. Only the primary machine can render MPeg2.

... My hope was to find a way to have multiple computers rendering out a 2-pass VBR mpeg2 for DVD. Ah well. ....

That was my hope too until I got to the fine print.
Its good for other tasks though. You can assign a filter or two and keep working on the primary machine.

[lordsmurf]

The word is "ENCODE", not "render".
Render is what you do to CG.
Encode is what you do to video.

[edDV]

Well, I somewhat agree. Lets try this.

Encode is when you do the "CO" side of the CODEC
Decode is what you do on the "DEC" side of the CODEC

Render is what you do when you fill out CGI frames.
Process is what you do when you apply a filter or create effects frames.


It's Sony that calls it "Network Rendering"
http://mediasoftware.sonypictures.com/download/step2.asp?DID=515

Everyone else calls it "cluster computing" or "distributed processing" or "distributed computing"

I call it cool. Power to the little guy. Bring it on.

[jimmalenko]

Encode is when you do the "CO" side of the CODEC
Decode is what you do on the "DEC" side of the CODEC

No.

Compression is when you do the "CO" side of the CODEC
Decompression is what you do on the "DEC" side of the CODEC

[lordsmurf]

A "render farm" is what is used to speed up the render of CG (computer graphics) ... fun things like Shrek, where they assign wire frames and gradients, etc ... all to various CPUs and make them share the effort using a special kind of network.

This is not video.

[bugster]

Encode is when you do the "CO" side of the CODEC
Decode is what you do on the "DEC" side of the CODEC

No.

Compression is when you do the "CO" side of the CODEC
Decompression is what you do on the "DEC" side of the CODEC

Actually, I think both are valid definitions.

Encoder/decoder
Compressor/decompressor

[BJ_M]

renderman, shake and fusion are 3 very popular network rendering apps used for compositing or cgi rendering ..
also AE of course ..

nvidia now has a render engine also to work with 3dstudio and maya .. not network yet .

[NightWing]

Soopafresh is correct.

I use the same for when I get backed up or need thing a bit quicker but dont want to turn it over to a real time encoder.

1) Splite the video into "chunck's". What would be your chapter points, skip points whatever. Or just cut it into time chuncks.

2) Turn each over to a seperate machine. With the bigest chunk going to the fastest while the smallest going to the slowest.

3) When assembled in your Authoring package. You get clean chapter points, skip points. etc.

I do this on complex projects. Simple and it works! Either give each machine a seperate "copy of the DV file" or pull off of a central host. You are only looking at 13gb per hour. Large HD are cheep these days.

The problem on handing different layers of the encoding by different computers is the nature of MPEG2. Esp 2 pass. By the time a "rendering" encoder farm could work. It would be in the ballpark of the multi segmented concept anyway. Else spend huge amount of money for one of the pro compresson boxes.

Example:

I like to tape Enterprise even if they are going to release it starting next year. Not sure but will pick up a set even with that uggly box!

I record it in either DV or MJpeg format. I use ad's, opening, closing whatever as my break points. I set each up in TMPGE batch file. The copy that file to the others along with the video. At each. I keep the segment it will do and drop the others. Turn them loose and walk away. Simple huh. basically a form of Sneeker net so to speak.

Or just turn it over to the fastest computer and do somthing else on the others.

[sigerson]

Soopafresh is correct.

I use the same for when I get backed up or need thing a bit quicker but dont want to turn it over to a real time encoder.

1) Splite the video into "chunck's". What would be your chapter points, skip points whatever. Or just cut it into time chuncks.

2) Turn each over to a seperate machine. With the bigest chunk going to the fastest while the smallest going to the slowest.

3) When assembled in your Authoring package. You get clean chapter points, skip points. etc.

Heh, why didn't I think of that!

I can use Tsunami to join the clips after again, too!