View Full Version : another screamernet problem...

01-18-2004, 01:03 AM
i think this has been posted already but i'm in kind of a hurry so here goes-

i've successfully setup my screamernet now my problem is that my animation could not be saved. i've also read a screamernet manual i downloaded somewhere and t said that avi, mov or animation files cannot be saved by screamernet. is this true? and is there any workaround on this?

if u any of u gauys could kindly point me the thread for this if this has been posted already. thanks in advance:)

01-18-2004, 08:09 AM
If you wanted to save an AVI file from a screamernet rendering, then its true, you cannot do this. Instead, save your files as an image sequence, then take all the files into AfterFX or Premiere, etc. and make them into an AVI from there.

The benefit to this is that if your screamernet rendering crashes or you need to stop it, you won't lose your renders up to that point.

Tom Wood
01-18-2004, 08:10 AM
Originally posted by peji
i've successfully setup my screamernet now my problem is that my animation could not be saved. i've also read a screamernet manual i downloaded somewhere and t said that avi, mov or animation files cannot be saved by screamernet. is this true? and is there any workaround on this?

Probably this manual:


Screamernet will not save animations, only individual image file sequences. That's because each node has to be able to create and store an individual image, and can't presently do that -and- place it within a larger animation file in the correct location relative to the other incoming images from other nodes. Use Aura (or Mirage) or similar to convert the image sequences to an animation.



01-18-2004, 08:23 AM
You guys know that NewTek has an RTV-screamernet capable saver already?
Insert edits; recompile old anims; all kinds of good stuff.
I hope they release it one day soon.

Harborer of info tidbits,

Tom Wood
01-18-2004, 09:03 AM
I think it's actually already in the list of render options, but I got instant LW crash and burn when I tried to use it. :p

Seems like it would necessarily have to be slower to give it time to check for proper sequencing each time it placed a frame. Would it cache frames that were waiting on another? Hold them in RAM?


01-18-2004, 09:21 AM
yup Tom Wood! that's the tutorial which i got. finally, i can get on ith my life:) i have 1 more thing that's really bugging me-do nodes render the same frame OR is it possible to render this way? or do nodes wait for their turn to render their own frame? and lastly, which image file (best) should use when i render usinig screamernet and then compiling it into one movie in premiere or after effects?

01-18-2004, 09:24 AM
The way it was explained, none of that would be necessary: it would build the file asychronously.
You could play it while frames were being inserted, no less.
The problem I have with it is that I don't know where to set it up to look for the rtv info. During a render you can see the error message that goes by relating to this issue.

Truly, this type of rendering capability would surely be revolutionary for desktop systems.

01-18-2004, 09:28 AM
Getting back to your post and your thread... ;)

Depends on what you are using to control screamernet. Go look up Spider or Lightnet for free controllers.

Pick the file that serves your purpose.
Need Alpha embedded? Make it a 32 bit format like png, tga, tiffs.
Don't care about alphas? Any 24 bit format will do. Those above would suffice here, as well.

Want buffers galore? Use ImageFilters for saving as rpf, rla, etc. Export some buffers the same way.

Lots of choices. The more you research the more you'll know. ;)

01-18-2004, 09:46 AM
thanks for the info dude:) i'll look into that:)

Tom Wood
01-18-2004, 09:47 AM
The way I understand it, each node renders one complete frame at a time. I use uncompressed .TGA images, but I might try .PNG images to see if there is a discernable difference. I take them into Mirage, render out a RTV sequence that goes to VT3, and then render out a DV that is firewired to DVCAM tape. Watched the first narrowcast on cable access TV last night, and the image looked pretty good. (Better be, for all the disk space this uses.)

I don't use a -controller- for Screamernet, what do they bring to the process? I set up the render options in the render options panel where the file type, save location, and frame range is set. Then just add the scene to the render list in Screamernet. If there is a faster way to do that, I'd like to hear about it. I am doing a lot of small animations that are compiled later, and setting all the render options each time is tedious and error-prone.


01-18-2004, 09:58 AM
Oh gosh, Tom,
That sent a shiver down my spine.

With many of the controllers, you have a way to save directly to the que. For example, I use Tequila Scream and he has written (Chepe Nicoli, author) an lscript that pumps it right into TS for rendering, with all the settings from LW. So, I just setup a scene and press 9 to send it to the render garden. As I have other machines in this garden, they pick up the rendering so I cna continue to use my machine at full. IF it's not in use, I start the two lwsn's on the main machine to assist with the rendering and walk away.

I remember reading about email connectivity, too. The first time it paged me as I drove down the road was awesome. It told me it had just finished rendering oodles of frames. Very cool stuff.

Tom, there are a plethora of reasons for a solid, professional render controller. Spider is the one I would recommend right now as it's what the pros were using and is now free.

Tom Wood
01-18-2004, 10:32 AM
Thanks Robert,

If I can impose just a little more...

I am doing animations using characters like my avatar that are edited together to create a mock news show. (Talking Heads Newz...) In order to get the lipsynch to work (I use MagpiePro) I have to be able to place the WAV file and the RTV file for each clip at the same start frame in VT3. Which means a lot of little clips, which works good too to simulate a talk show that cuts back and forth between cameras.

So I save each Lightwave image sequence to its own folder, the name of which has to be entered -somewhere- in the process. I do it in the LW render options panel. Then I have to tell it what to name the files inside the folder, which is a pain because I just use the same name as the folder, so that's a step I could lose. (Images are then automatically labelled sequentially with the 001, 002, 003, etc. suffix.)

Then I also have to tell it to render the entire scene, which seems to me like it should just be a default. I guess adding a controller that would always do that could save a step, but if it has to be told that too, then I haven't gained anything.

Will a controller automate any of this?



And thanks peji, for letting us leverage some discussion off your thread.

01-18-2004, 10:49 AM
hehe no prob tom...as if i did this intentionally hehehe:)

i have another que on this screamernet thing-

i have a P3 800 and Athlon XP2500 comps right now and set both as nodes but my XP2500 mainly as master also. if the theory goes - "nodes render their own frames" then would it be the same if i just open the same scene on both of my computers but dividing the frames to be render (500 frames on each comp to have a total of 1000)?

can u guyss orient me more on these "spider,lightnet,controllers". i'm not really sure hat they are..


01-18-2004, 10:53 AM
and also "RTV"..

Tom Wood
01-18-2004, 11:03 AM
Yep, that will work as long as you have a way to composite them together later. You can even open two instances (or more) of LightWave on a single machine and divide the frames between them. I did that before I bit the bullet and set up Screamernet. If you go either of those routes, you could also just let LightWave render an animation, again just as long as you have a way to connect them, or edit them, together later. I animate all day and render all night, so Screamernet is a necessity for my workflow.

I'll defer to Robert about the controllers, I don't know.


01-18-2004, 11:13 AM
I can't (won't) tell you all about controllers in this thread but instead suggest you do a search and have a read about them.
Suggested terms:
ButterflyNet (I think) - http://www.liquiddreamsolutions.com/
Tequila Scream - http://www.digitaltoons.com/shop/
Spider - http://www.stationx.com/
Muster - http://www.vvertex.com/

All of these should alot you copious amounts of reading, in just this forum.
And to your question, no, it's not the same. With a controller it will distribute the files to the next empty node. So in your situation, you could literally sit and watch as the Athlon render 2 to 1 frames. A controller would do the averaging. With the half and half as you describe, the master box will sit for a bit of time as the 800 caught up.

It would be like setting up a lw scene and saving it. Yes, you have to start all the nodes but if it's 2 or more, having a centralized on/off switch would be a savings.

Check out Spider. I think you will see the benefits rather quickly. Where are you doing all that naming? In LW, proper? Or in the screamernet panel?
IF in lw proper, then that is still the procedure. IF you mean on the screamernet panel, that would make me scream. [Can I admit: it's been a long time since I looked at that panel. ;) ]

In your situation I can see a great benefit being gained from the use of a controller. And it wouldn't take long to get spoiled. ;)

Let us know how it goes.

01-18-2004, 11:14 AM
thanks dude:)