DR slave renders with no texture

[puha0941]

Hello,

I have set up a DR and it seems that for some reason, the buckets from the slave machine renders without textures?!. So the final image is there are buckets with no texture (just grey). This happens as soon as the slave machine starts to kick in...

Im using 1.5 vray Rhino 5 64 bit win7.

Any suggestions...as to why this is happening?

Thank you

[puha0941]

Is this coming from all the textures saved on the local machine only?
Should they be located on the server instead for the DR to work? (If so, is there a way to move all of them at once and re-link? I have lots of materials so to re-link all would be very time-consuming.)

Thanks

[Micha_cg]

enable "Asset collection" + "send to each node" should do it

[fooprobe]

Micha's right ... But I find that sometimes this still happens if the network slows down or one of the DR slaves is unhappy ... And sometimes it can be if the local machine is unhappy! If Micha's solution doesn't work, try restarting the slaves and make sure to close other apps on local machine.

[puha0941]

Ah, great Micha... Thanks again !!

enable "Asset collection" + "send to each node" should do it

[puha0941]

Thank you fooprobe so much! Yes, I think I have not restarted the machine in a while I will try that as well.

Micha's right ... But I find that sometimes this still happens if the network slows down or one of the DR slaves is unhappy ... And sometimes it can be if the local machine is unhappy! If Micha's solution doesn't work, try restarting the slaves and make sure to close other apps on local machine.

[puha0941]

Micha, By the way, is there a rule as to performance of machines to be the host and the slave? I read one of your post saying that the host works 'harder' than slaves because it divideds buckets and collects them back.

My question is can a slave be much weaker than the host? For instance, a workstation host and a, say, laptop slave. Would it still help?
Another question is, if the host is struggling with RAM, would using a slave help reduce the burden on the host machine's RAM?

enable "Asset collection" + "send to each node" should do it

[Micha_cg]

I wrote this? I don't know. I used slower slaves, because my master machine is my working machine.

RAM - at the master machine you need more RAM, because Rhino need additional to hold the NURBS model + the rendermesh + Vray rendering. At the slaves there is only the Vray rendering.

[puha0941]

Thank you for the reply. I think I did not word my question clearly.

I have a scene where my host pc cannot render properly because the RAM gets full. If this is the case, if I use DR with a very weak laptop with some RAM on it, would it help the host pc to be able to reduce the RAM load and be able to render it properly?

I wrote this? I don't know. I used slower slaves, because my master machine is my working machine.

RAM - at the master machine you need more RAM, because Rhino need additional to hold the NURBS model + the rendermesh + Vray rendering. At the slaves there is only the Vray rendering.

[fooprobe]

I doubt it. DR just sends a copy of the scene to the slave and renders buckets there as well. So it's multiple machines all rendering the same model, using the same amount of memory, just rendering different parts of the image. However your host machine also has to manage the slaves, so if you are crashing with one it would probably only get worse with DR

[puha0941]

Thanks fooprobe! It is very clear and now I understand it well.
I have another question about vray proxy. If I turn the whole model to vray proxy, I would not benefit to save RAM usage because proxy only helps when more than one proxy is repeated and it helps machines to save RAM by only using one proxy's required RAM.. Am I understanding the concept here right?

Also, the video said I can use 3ds Max or Maya created vray proxies, do you know if there is a website where I can download proxies (trees or grass etc)?

I doubt it. DR just sends a copy of the scene to the slave and renders buckets there as well. So it's multiple machines all rendering the same model, using the same amount of memory, just rendering different parts of the image. However your host machine also has to manage the slaves, so if you are crashing with one it would probably only get worse with DR

[fooprobe]

This, i don't know much about. If proxies work like blocks, then you are basically right, but I think proxies are actually somehow more efficient than blocks, so there may actually be benefit to making the whole model a proxy. There is a proxy tutorial on this site, you should check it out.

Mich's right, you should try to get access to the whole forum, there's a lot more info here.

[Svetlozar Draganov]

I have another question about vray proxy. If I turn the whole model to vray proxy, I would not benefit to save RAM usage because proxy only helps when more than one proxy is repeated and it helps machines to save RAM by only using one proxy's required RAM.. Am I understanding the concept here right?

You can get benefits with proxy even when you are using a single proxy object, because when Vray loads information from proxy file it only requires information about the current bucket - it doesn't load the whole proxy. The proxy is loaded and unloaded dynamically only for the buckets being rendered.
Another advantage is that the exporting/translation scene geometry before the actual render will be faster -because the proxy is already exported. In scenes with a lot of heavy geometry this could speed up translation a lot.
And one more thing - with proxy viewport navigation is much faster

[puha0941]

This is mind-blowing. This means that if I convert the whole scene into a proxy at the end, right before I render, I would benefit so much using the proxy function!

Thank you kindly for your clear explanations!

You can get benefits with proxy even when you are using a single proxy object, because when Vray loads information from proxy file it only requires information about the current bucket - it doesn't load the whole proxy. The proxy is loaded and unloaded dynamically only for the buckets being rendered.
Another advantage is that the exporting/translation scene geometry before the actual render will be faster -because the proxy is already exported. In scenes with a lot of heavy geometry this could speed up translation a lot.
And one more thing - with proxy viewport navigation is much faster

[d77coll]

After playing around with the proxy, I've done some tests:
1. Render a scene (3 complex trees) as meshes
2. Render the same with proxy (importing the proxy correctly)
3. Render again with proxy but as adding the proxy as copy-paste from a different file
Results:
1. Heavy Rhino file (60MB) but rendered in 29sec
2. Light Rhino file (1.6MB) rendered in 45.3sec and having to relink materials
3. Light Rhino file (1.5MB) rendered in 42.6sec and no material ID issues

Great, the Rhino files difference in size are enormous though why is the render time almost double?
Anybody has any suggestions?