1.5 preview: DR max6!

[triar]

Vlado, specify please and please us new FAST Fake Glossy - you promised all of us to remove the most important brake in current beta versions 14570. There is a progress in this direction or version 1.5 will not leave until you will not write new algorithm of imitation of this process?

[cpnichols]

Hey thanks for all the great info guys.

Chris, maybe I'm not understanding what you're saying but wouldn't crease information just translate to camera dependant displacement?
PS. Do you have popups blocked? I sent you a PM a while back.

--Jon

Well if you take a cube and subD it it turns into a sphere... if you take the top and bottom edges and define them as creases, it becomes a cylinder. The way that it is done in MAX is to take an edge and define it as a crease or partial crease. Will Vray take that data and use it as well? Or will it only take the raw polygons and smooth them all over. If so, one would need to add several edges close to eachother to make a crease. Not an uncommon way of working, but an important part of the workflow.

PS... yeah I have popup blockers... check your messages...

[J_Bug]

Ok I see now, I was misunderstanding creases

--Jon

[egz]

Vlado, specify please and please us new FAST Fake Glossy - you promised all of us to remove the most important brake in current beta versions 14570. There is a progress in this direction or version 1.5 will not leave until you will not write new algorithm of imitation of this process?

Glossy highlights are in 1.45.70 already, and they are quite fast

[Da_elf]

Paul Oblomov every software company has their team of testers. People who have been using the software so long that sniffing the bugs seems pretty second nature. And the testers form their own little community to discuss bugs, work around to current bugs and mistaken bugs (user errors). before the bug is rushed off to vlado its investigated thoroughly that makes vlados job a whole lot easier. you said yourself that you dont have time to go thorugh all of this and you just want a copy to play with however the alpha process doesnt work that way with any of the softwares i know.

[Robert]

How does: Rendertime subdivision surfaces works?

Does it support Meshsmooth (edge creasing?) or is it an own modifier with no viewport preview?

Meshsmooth rendertime itteration is one of the most encountered reason for Vray to crash (I think it has to do with a memory problem or so.).

robert

[Da_elf]

it will subdivide whats needed when its needed just the same way the displacement modifyer works

[egz]

How does: Rendertime subdivision surfaces works?

Does it support Meshsmooth (edge creasing?) or is it an own modifier with no viewport preview?

Meshsmooth rendertime itteration is one of the most encountered reason for Vray to crash (I think it has to do with a memory problem or so.).

robert

Right now in this beta it only has basic rendertime SubD functionality, as part of the displacement modifier. Vlado has talked about ways of improving this though, so the shipping version will most likely be a lot better workflow wise. The current version doesn't support creasing, the final version may or may not supoprt this, I don't know yet.

Also adding a MeshSmooth with render iterations set to 0 works quite well for viewport previews. One could even make a scripted modifier that uses VRay SubD when rendering, and MeshSmooth/TurboSmooth in the viewport.

[gustojunk]

Hey EGZ,

I have a DR question for you. Since now you can distribute the IR pass and the whole setup on slave machines is much simpler. Will it be possible to use a remore farm for DR?

So lets' say I'm working on my laptop in a client's site and I want to do some rendering changes, tweaking llghts, materials etc. Can I tap remotely to a farm in my office?

That would be an extremely useful feature for me, especially since allows people working in different locations all use a single farm resource. If that's not possible, what would it take for something like that to be doable? IS it a banwith problem? do you need to setup a VPN? Am I out of mi mind even thinking this would be possible?

Thanks,

Gustavo

[egz]

DR uses TCP/IP so it should be possible. As long as you set up the network so that the correct ports are mapped to the correct servers. With the previous DR in VRay 1.0 Da´Elf and I did a test where we rendered a picture with one computer in Canada and one in Norway.

Bandwidth will be a problem if your scene is large though, since the entire scene will need to be sent from the client to each server.

[gustojunk]

mmh, this sound interesting. So you are saying that if I have a 1MB scene in the client side, and I use a farm remotely that has 10 machines

a) do I have to send the 1MB files remotely to each of the 10 machines?being a transfer of 10MB?

b) or the 1MB file travels over the remote distance once and then one of the 10 machines acts as a server and shares the data locally to its 9 other slaves?

Also for test renderings of lets say 800x600 single frame test, it seems to me that the rendered frame of pixels uncompress is about 1.4MB, so to do a test rendering via a remote farm would be the time of uploading a scene (lets' say 1MB Max file) , have the local machine plus the 10 remote machines rendering buckets and then return back 1.4MB of pixels (minus the buckets that were done locally).

If a local client has an upload speed of 1.1 minutes for upload and then there is some rendering overhead plus the time you get the buckets rendered back (another 30 seconds) it looks too me that you are talking about a 2minutes toll just to make the trip and then you can start taking advantage of the network rendering.... Does that make sense? Or I'm missing something else

So I'm not sure how practical his is for tweaking if no matter what you never going to be under 2 minutes ugh?

Unless we upgrade the home cable modems in the clients side to something faster, then it would start to make more sense ?

Basically I'm trying to cut down render time for setting up, I'm cool with leaving final renders churning overnight or having a renderfarm that can do those, but still out biggest problem is the time spent in front of the tube making small changes, etc. It would be nice to cut this time down and being able to do so no matter where you are working

Gustavo

[circus]

Egz,

how fast is your network? mine is a 100meg network and I want to know if this will be fast enough for DR... (not the loading of a scene more the bucket speed)

Also does the DR app hold max open for as long as you want on the slaves? It doesnt need to restart max for every render does it?

[egz]

how fast is your network? mine is a 100meg network and I want to know if this will be fast enough for DR... (not the loading of a scene more the bucket speed)

Also does the DR app hold max open for as long as you want on the slaves? It doesnt need to restart max for every render does it?

DR works fine on a 100 Mbit network. The spawners will keep max open on the slaves, if it had to restart max every time you pressed render it would be too slow.

[egz]

mmh, this sound interesting. So you are saying that if I have a 1MB scene in the client side, and I use a farm remotely that has 10 machines
a) do I have to send the 1MB files remotely to each of the 10 machines?being a transfer of 10MB?

As far as I can tell it sends it to every slave.

If a local client has an upload speed of 1.1 minutes for upload and then there is some rendering overhead plus the time you get the buckets rendered back (another 30 seconds) it looks too me that you are talking about a 2minutes toll just to make the trip and then you can start taking advantage of the network rendering.... Does that make sense? Or I'm missing something else

That sounds about right

So I'm not sure how practical his is for tweaking if no matter what you never going to be under 2 minutes ugh?

Even though it works, DR isn't really meant to be used this way, mainly due to the bandwidth restrictions you mention. When the internet is all 100Mbit someday though.....

[f97ao]

This subD sounds very very exiting to me at least.

As some of you know my work on PolySpeed should let the user use meshsmooth on just about any sized object. I've been doing some tests on 500k objects, and it's no big problem modeling with iterations 2-3 on them. (this is for the detail areas, and then you have max 1 iteration on the parts that are not where you are working). Problem is rendering!

Rendering a 500k object with say 3 iterations will create a 32mil poly object, and well max doesn't like that at all. So if I understand it correctly these kind of objects should not be a problem to render with Vray then?

Some of you may say that 500k objects for the base mesh is madness, but I myself know about many things that I would like to model that have this kind of detail and is not easily turned converted into displacement mapping.

/Andreas