[thxraph]

a friend of mine, coder one told me that it could export a MXS as usual, then a matrix file with frame-by-frame coordinates changes for each object...

ok, i guess another coder can understand that


also, what could save my HDD and my network speed could be simply compress data when MXS is written...

for ex. a simple 100*100*100 subdivs cube is about 7Mb in MXS, and only 20Ko zipped...

ok, time needed to compress / decompress datas, but ok, HDD place savec (plenty) and time saved sending over the network...

Pretty good idea an Xml file could handle that with the bonus of it could be generated easily

Naxos... je vous conseille d'investir en raid avec bcp de disk dur pour vos rendu anim... ca pese vite tres tres lourd, de plus tu as omit le mxi, qui peu etre lourd egalement...qui n'est pas indispensable cela dit!

[naxos]

le MXI est lourd, mais il n'y en a qu'un, fort heureusement...

[naxos]

back in that topic for a real big problem about what i thought a minor limitation...

cooperative does not work when animation is on : each serverf renders a single frame from anim...

we all think final rendertime is the same : that is true...

but : imagine you have a renderfarm where all PCs are not identicals...
the way maxwell render is working gives 2 hours to each PC to render a frame... but ok, slower machines will give toll different results than faster ones, more grainy pix...
so, finally, you'll get some pictures grainy and others less grainy...

ok, we can ask to stop after a SL reached instead of time... but then we can no longer plan the full time to go for final... and so we loose the user-friendly behavior of MX...

so, turn this into a request : please let us render in real cooperative way an animation, please... so all frames are identical, quality speaking...

thanx

[naxos]

ok, it seams that none from NL seams to find my problem serious...

i think we will not get Maxwell rendered animations soon...

Hello ? is anybody (from NL dev team) there ?

[Peter_K]

Hi Naxos,

Im really curious about to find out what kind of rendertimes your having when doing animation?! I´ve only played very very little with maxwell sofar since the network rendering has been nothing except pain. So how long time does it take to render out a NTSC/PAL frame in good quality?

How many days can one student render his animation for? You said 2,5 minutes of completed animation on 90 CPUs.. Thats 3750 frames. I guess the computers are used during day time since its a school or? So lets say one computer renders one frame / day maybe?.. Thats like 3750/90=42 days to complete one animation... So well. I hope the students are collaborating on each project or only 9 students will have time to render out their project/year

Well.. Im interested to hear what you typical rendertimes are and what specs you got for your machines... I haven't got as many CPUs that you have but I got speedy ones, and also plan to use maxwell for animation..

Btw.. It would be good if nextlimit did come up with some point cache baking for maxwell, so we dont have to spit out an .mxs file for each frame ..

/Regards Peter K

[Maxer]

My render times seem to be around 1 or 2 hours per frame at 720x480 but they look really good. I'm working on one now that I hope to share sometime next week.

[Peter_K]

What are the specs on your computers Maxer? Im currently replacing my P4 for AMD 5000+ (or Intel E6600, if I can find a matx mobo with gblan)... 1-2 hours/frame is totally acceptable for me.. That would atleast let me render over 10 seconds/day

/Regards Peter K

[Maxer]

I've got a dual core dual processor AMD Opteron 275 with 4 Gigs of ram running Windows XP64. It's a good machine but I'd look at the new Intel chips to see which gives you the better performance.

[lllab]

i just recieved a quad xeon woodcrest (apple) will test soon against my 275 quad opteron:-)

i will post test comparisions as soon i have it installed.

cheers
stefan

[beppeg]

Hi lllab, please can you install m~r 1.0 first and make the speedtest v1.0 ?

[Maxer]

Why are you still using V1.0?

[dyarza]

i just recieved a quad xeon woodcrest (apple) will test soon against my 275 quad opteron:-)

i will post test comparisions as soon i have it installed.

cheers
stefan

Yes, but you'll be running under Rosetta while you wait for a Universal Binary version of M~R. Render times will not show the true potential of the new CPUs. Or are you going to load XP on it?

D

[beppeg]

Why are you still using V1.0?

It's only for making the speedtest for a comparison with the others cpu
http://www.maxwellrender.com/forum/view ... hp?t=15924

I would lunch the speedtest v1.1 after the exit of m~r mac-osx UB version for interesting comparison

[Kabe]

I don't think that creating a scene for each frame is a big problem.
Changing that would mean major loops and hoops for NL, and there's a lot
of bigger fish to fry currently. Also this would make the SDK harder to use
and I don't think that this would be an improvement over all.

What could be done and would be well worth doing, not only for animation
but also to save HDD and network traffic, is to zip the MXS files internally.
This reduces file size to ~5%, the affected applications could support both
the compressed and uncompressed MXS type without major redesign.

Kabe

[lllab]

i am using win xp on the macpro;-)

so no rosetta stuff...

test will come later..with v1.1

cheers
stefan