Preview of OctaneRender for Blender

[kavorka]

I'm excited to see this in person.
Being patient, but also excited.

[patlecoq]

There is already a node API in Blender.
LuxRender is using it:
http://www.luxrender.net/forum/viewtopi ... =11&t=7931

Jim, are you aware of this API?

[JimStar]

There is already a node API in Blender.
LuxRender is using it:
http://www.luxrender.net/forum/viewtopi ... =11&t=7931

Jim, are you aware of this API?

No, what LuxRender uses is not even close to what Octane nodes network needs to become fully integrated. This is not external node types managing API (creating own types and its instances, creating own groups of types, building own nodes network, managing own connectable pins and its custom types, etc etc...). Removing some little feature from this list - and there will be no sense to develop integrated Octane plugin, as anyway it will have all the same limitations as already existing scripting exporters... And the users will complain "why I still can't do this, why I can't do that, standalone can do it...". LuxRender just uses the existing nodes types and adds some simple attributes to them (not even network pins). It's not the way for getting the full maximum from integration of such cool engine as Octane into such cool modeling software as Blender. We need to do this integration at maximum quality to match the quality, power and flexibility of Octane itself. It is the reason I don't use such compromises as using a few generic Blender nodes (already having few own pins unusable for Octane) with a bunch of external-attached not truly connectable Octane attributes to hurry-scurry represent "Octane nodes" and convert all this junk on the fly every time the user hits "render". One more reason I want to develop this plugin ideally - my wife is skilled Blender professional too, and she loves them both: Blender and Octane. So, I have a lot of feedback from her after years of her use of Octane through exporter script...

So, it is not just "for no reason" Cycles uses the same "internal module architecture" as the Octane plugin is forced to use at the time...
Don't worry, when the usable external nodes fully API will be present in Blender - Octane will immediately start use it to become fully external. Before it will happen - be sure, the "internal module architecture" Octane Blender plugin will be released as binary and as patch for new Blender releases, don't worry about it too.

[kavorka]

I like that your wife is a professional Blender user. I'm sure she will really help you out during the alpha phase.
Hopefully, soon after the open beta, the Blender Foundation will see that they should work on the API you need (admittedly, I have no idea what you are talking about, having no coding background at all). I'm sure other programs would then use Blender to integrate and that would lead Blender into being a much more professionally used program.


Looking forward to seeing more
btw, are PT and PMC working in the plugin right now? It looked like just DL in the preview.


Thanks for all your work!

[JimStar]

btw, are PT and PMC working in the plugin right now? It looked like just DL in the preview.


Thanks for all your work!

Yes, all kernels work, and all settings you see in the user-interface work too. I've just not used them all in video capture...

[daveyt]

Great stuff, keep up the good work

[ROUBAL]

@Jimstar : Great work ! Congratulations and many thanks !

I have got a question about the memory necessary amount when using Octane with the Blender plugin :

I assume that the scene has to be loaded twice in system RAM : once for Blender (the Blender file itself) and once in an other format for Octane (voxelisation/loading on GPU). So, using the plugin should imho increase the required RAM amount compared with Standalone, not needing to have Blender opened at the same time.

Am I right ? In which proportion ? Same as Standalone plus usual Blender scene ? Less ? More ?

As I am currently often short with my 8GB of RAM for 3GB of VRAM, I wonder if Plugin release time will have also to be the time of renewing my hardware...

[abstrax]

@Jimstar : Great work ! Congratulations and many thanks !

I have got a question about the memory necessary amount when using Octane with the Blender plugin :

I assume that the scene has to be loaded twice in system RAM : once for Blender (the Blender file itself) and once in an other format for Octane (voxelisation/loading on GPU). So, using the plugin should imho increase the required RAM amount compared with Standalone, not needing to have Blender opened at the same time.

Am I right ? In which proportion ? Same as Standalone plus usual Blender scene ? Less ? More ?

As I am currently often short with my 8GB of RAM for 3GB of VRAM, I wonder if Plugin release time will have also to be the time of renewing my hardware...

Yes, you would need the required memory for Blender plus Octane. Fortunately, since 1.11, building the acceleration structures (aka voxelizing) requires now less memory than before. And RAM is cheap these days. Upgrading to 16GB doesn't cost an arm and a leg, but really helps when you are doing 3D stuff in general.

Cheers,
Marcus

[ROUBAL]

Thanks for the answer !

Upgrading to 16GB doesn't cost an arm and a leg, but really helps when you are doing 3D stuff in general.

Unfortunately, My old motherboard doesn't support more than 8GB, the casing can't host bigger motherboard, and the PSU couldn't feed GTX or Titan cards. So it means upgrade for a new workstation... when I will be money-ready !

[JimStar]

btw, are PT and PMC working in the plugin right now? It looked like just DL in the preview.


Thanks for all your work!

So, I've just made some additional sample to make it clear. Don't worry, all the features Octane has are present in the Blender plugin.
Yes, perhaps this plugin may be a little slower in real-time viewport movements (as it uses network to get the render-buffer), but anyway you can see, that this is fast enough to enjoy using Octane fully integrated into Blender...