Date for R1.5?

[bojo5150]

I just purchased VRay advanced for my firm and was a little disappointed to see that distributed rendering was not availiable for max 6 yet. Basically I can't even use VRay because of it. I should have look through these fourms first before my purchase I guess. Anyway, does anyone have a date that we can expect to get this functionality? Please post any information you might have. Thanks!

[Da_elf]

just out of curiosity...why cant you use vray?

[dbuchhofer]

As of last week there was a post that DR in the new version was in 'testing' which alludes that is at least functioning again

Aka, the ever popular 'soon'.

[ShaunDon]

For the same reasons a lot of us are sitting on already-purchased Max 6 upgrades -- until DR works in 6, I can't go back to using one machine for a render. I need them done at the very least overnight, and sometimes it takes at least five machines to do even that! Leaving one rendering for a week when you have deadlines is not an option.

Render striping is a possible workaround, but there have been negative posts about using it on this board. And yes, there is ALWAYS region render, which I suppose would work well enough for finals -- but I use DR in every day of my production work. Interiors that take 15 minutes with five machines just to render a decent preview are impossible if you can only render to one machine.

Anyway, good to hear of DR's "functioning" status at least. I wonder how it compares to 1.09.x -- can we cancel DR renders without having to reboot render nodes? I know if I time it just right and disable all the machines in the list in between buckets kicking back pixels, it seems to recover fine... but with 10-12 buckets going you screw up eventually. And then it's off to reboot the office. Yahoo!

I really don't want to speculate about dates for Vray anymore. I just want to wake up one morning and have it, no warning. We were all told to count on May. We did -- dropped cash on all new max licenses to migrate to 6. Summer's almost over and we're still on 4.2, those shiny new discreet boxes still shrinkwrapped, so I'd rather just forget about the whole thing.

Shaun

[bojo5150]

To: Da_elf

To answer your question, I can't wait several hours to render high res images, and days even weeks to render animations. We would go out of business if it took that long.

[cpnichols]

Distributed rendering will not help your animations. You should be using traditional network rendering, which works GREAT with Vray. Also MAX 6 has tools that allow you to distribute your still rendering to many computers. I think it is called scanline strip. From what I was told, it will work with Vray and MAX 6. Someone correct me if I am wrong. It is not as efficiant as the distributed rendering system that you find in Vray or mentalRay, but it will still work.

It seems strange that you can't wait hours for a still. Most people have no choise. As far as animations are concerned, most people, even with a large farm of 100+ compyters, still have to wait several days for them to get all there frames. That is unless the still or renderign is extreemly simple.

Distributed rendering is not the answer to everything.

Are you sure you are not confusing Distributed renderign with network rendering?

[tom schuelke]

Hi

Well of course the distributed thing would do a fewthings better.

but at last the stripeset option is a thing we now use for many jobs, and we didnt have any problems thinking of speed breakdown on single machines so far, as some have told.

the other thing about dr is the possibillity to do fast previews.

well this is throu if you only want to do previews with many glossys and so on.

but !!! if you use your previews most time for a better lightsetup, the new lightmap is a ckickass timesaver that saved our live in many situations.

for example we rendered about 20 scenes in animation for a huge Aida Ship interior with about 4000 000 to 7000 000 polygons and about 400 lightsources. I could do previews in Pal in about 2 minutes.

you only switch of reflections and take lightmap for secoundary and first lightbounces, and this rocket takes you away.

at last, this and also the posibillity to use qmc settings for your previews allready on single machines, in my opinion makes 1.4570 a much better solution as 1.09 ever was. And we now use it for about 2 month in production on several projects, and all bugs we found could be managed.

for us there´s no way back to 1.09 even if we also would like to see dr soon.

[Long John]

Distributed rendering will not help your animations. You should be using traditional network rendering, which works GREAT with Vray. Also MAX 6 has tools that allow you to distribute your still rendering to many computers. I think it is called scanline strip. From what I was told, it will work with Vray and MAX 6. Someone correct me if I am wrong. It is not as efficiant as the distributed rendering system that you find in Vray or mentalRay, but it will still work.?

if this is correct then please someone read the bug forum und tell us how to solve the irradiance map problem.

the problem that sometimes the irradiance map is not distributed.

[tammo]

whats the advantage of DR?
in the earlier versions it wasnt possible to distribute IR map calculation.
this takes the most time in my scenes. has this already been changed?

even if you have a scene without IR map you will have to send the whole scene data across the network to each slave. on most of my scenes this took longer then rendering the scene right away from ram.

so what do you use it for?
maybe very large stills?

[tom schuelke]

Hi tammo

Well if your distribution takes longer than the rendering, hmm, i dont know, this sounds very unusual.

of course, when you render a scene and use split scanline, the rendering that usualy would take abut 50 hours, with 5 machines can be rendered in only one night. for example u precalculate the l-map, save it on the server, and the split scanline option with backburner, distributes several stribes over the network and i-map gets only calculated for those single stripes. this realy speeds up your work.

dr if it allready worked like announced also would give you the opportunity to do previews with several machines. which also could make things a bit easier.

also it is like long john told a kown problem, that in animations, with precalculated imaps, imap sometimes doesent get loaded on each machine propperly, so that you sometimes loose frames. nevertheless dr wasnt a solution for this i think, and there seem to be some workarrounds arround the corner. (don´t remember the thread right now).

but at last i hope 1.5 realy is arround the corner, and untill this the beta still is the better solution compared to 1.09.

Tom

[Long John]

bingo tammo.

DR is for stills that need hours to render with only one system.
because they a big in resolution or heavy to render.

imagine having not one bucket but 10, rendered at the same time with slave systems.

[Long John]

so that you sometimes loose frames. nevertheless dr wasnt a solution for this i think

i read that it should be a problem with read/write access to the irradiance map file. not sure.

i hope this does not mean that vrays DR will have the same problem?

i expect that DR IS a solution to this problem.

[tom schuelke]

Hi Longjohn.

your right.

it was sushidelic who seemed to have a workaround for this.

We also ran into trouble when netrendering with saved irradiance maps via backburner and UNCs. It basically seems, that irradiance maps have first of all to be writable, and when a client loads it, no other client is able to get this map at the same time, as it seemed to be flagged "in use" or something (this might be completely wrong, but that would explain the symptoms). That could also be the reason for getting more dark frames with irr. maps that are bigger in size than with small ones. With what we ended up was a utility that copies the irr. map locally to c:/global on every slave, that's where it is linked in 3dsmax as well. Result - no errors.

Greetings,
Michael

he seems to have a utility to copy imaps to the local renderslaves, and he told me, in sometime, his team maybe will delever a less specific script for free. he told, the imap seems to be flagged as used, while other slaves try to load it. and this seems to make problems. they solved it by copying the precalculated imaps to the several slaves with some sort of utility, so they where able to load it after this.

untill 1.5 does its job, this may save for us some time for sleep at night.

tom

[tom schuelke]

by the way, of course distribution also would solve this problem, but i think it also could work without dr single frame option.

it rather is a backburner and imap writing problem.

tom

[egz]

I'm not sure DR would fix this.

This is pure speculation, but I suspect the problem lies in the part of VRay that loads irradiance maps from disk. Right now it seems the problem is that if the file is not accessible, probably because other computers are accessing it at the same time, VRay will just load a "blank" irradiance map and keep rendering. Instead it should wait and try again and again untill it is able to get the irradiance map loaded.

Both DR and normal netrender would be using the same loading code, so unless it is fixed DR will have the same problem. I suppose