Blender Plugin?

[andras]

What you mean not Blender specific? Sorry I dont understand. Blender is an opensource so that can be specific with any software, cant it?

[enricocerica]

What you mean not Blender specific? Sorry I dont understand. Blender is an opensource so that can be specific with any software, cant it?

It means that the content of the obj file should be generated the same way whatever the original application is, so let's say for example that to define the camera you just have to name an specific object 'camera' in your 3D application and then it would be exported with this name so Octane will convert it in a camera instead of the original mesh, this is not specific to Blender.

[Tim Ellis]

What you mean not Blender specific? Sorry I dont understand. Blender is an opensource so that can be specific with any software, cant it?

It means that the content of the obj file should be generated the same way whatever the original application is, so let's say for example that to define the camera you just have to name an specific object 'camera' in your 3D application and then it would be exported with this name so Octane will convert it in a camera instead of the original mesh, this is not specific to Blender.

So you'd need an 'Empty' (Null), or actual geometry object to do this?

I've used this method before with a cone, using copy Loc & Rot from the Blender camera to the cone object.
Not aware of any software that can export a camera in obj format, or even if obj format supports cameras.

Tim.

[jazzroy]

i think he means that if you have a mesh named "Camera" during import it will be substituted with a camera in octane.
and probably if you have cube, cube.001, cube.002 they will be considered instances.

smart workaround, but there's plenty of limitations:
- you lose cameras' parameters, only position and orientation are stored
- you lose light's types and parameters also
- the obj file will have all the duplicates, and so it can become heavy
- you have to put a child object for every camera and light in the scene and name it correctly (quite a waste of time if you're experimenting with different sets)
- you have to name correctly all the objects that you want to be treathed as instances and change the ones you want to be different meshes (that's a huge problem, because every software has a different naming automation and also worse is the fact that every software names duplicates and instances in the same way!)

if the workaround works this way (this is just a supposition based on the info) i still have many doubts in considering it useful for daily work, but it's ok for testing and experimenting with renderer's features.

[pixelrush]

What actually do you guys want to do/expect to do with Octane?
I mean what type of renders do you do that require those heavy polys and instances? Is it arch viz with trees and bushes??

A gpu renderer surely has virtues but I am wondering if people arent expecting miracles and a more aggressive development pace than a small company can reasonably deliver despite their best enthusiastic efforts.

I think they are doing pretty well so far...how about we check out the demo and see how much is too demanding.
Its tough being a pioneer in a field and all too easy for the audience to find fault especially if their beliefs are somewhat idealistic

[jazzroy]

actually i think critics are positive.

they are going commercial, so this rendering engine has to face at least small studios requirements, not only some university research stuff (no offence on university research at all, i work both on private and university labs).

we have 4 nodes maxwell licences and obviously rendering time is an issue, expecially when you have architectural scenes with 20 millions polygons and half of the materials are glossy or reflective, so i'm really interested on seeing octane become an interesting alternative, that's the reason of my posts.

as i said, mine are only suppositions, but i don't think posts like "hey guys, you're incredible" can really be useful, except on the moral side

[pixelrush]

Yeah ok life is tough but it should be fair too.
Soon we get to judge Octane 'naked'...

I am just a bit fearful after reading some posts that after the stampede to try this new technology people will leave in a stampede disappointed because they wanted so much to instantly replace their existing renderer of 20m polys with something that works at warp speed in one quarter of the space.

I wouldnt like to see Octane's and gpu renderers reputation written off because power users cant yet be satisfied.
You are right though as a commercial product it will have to face honest criticism and voting with the wallet.

[radiance]

Hi,

i'll just keep this short,
indeed like pixelrush says,
we're a young startup company, and octane is a very young software.

you can't expect us to develop a renderer in a few months that's on par with existing solutions that have been in development for several years.
we're doing our best, and i'm sure people won't be disillusioned with the demo release,
but keep in mind it will be a 'technology preview' demo, not a complete solution.

however, with a lot of support it won't be long before it is.

Thanks,
Radiance

[jazzroy]

thanks for the answer radiance.

i'm sorry if my words sounded a bit arrogant, surely i don't ask for a tech demo ready for commercial market.
i'm just discussing about import features, simply because the success of a standalone renderer depends 50% on the ability of exchanging data.
so, i want to clarify that my thoughts were towards the approach for import formats, not about tech-demo (or even 1.0 release) rendering features.

and of course i will test the tech demo, as unstable and feature-limited as it will be, without expecting to work with it.

[Tim Ellis]

we have 4 nodes maxwell licences and obviously rendering time is an issue, expecially when you have architectural scenes with 20 millions polygons and half of the materials are glossy or reflective, so i'm really interested on seeing octane become an interesting alternative, that's the reason of my posts.

We run 26 Maxwell V1.7 render nodes here, all day, all night 7 days a week.
Our standard resolution for rendering is 5616 x 3744, with scenes averaging 5 million triangles. Largest renders we can get out on our current farm are 7000 x 4667, but at that resolution, the poly count has to be below 4 million.
Highly detailed arch viz scenes, whether they be interiors, exteriors, for a single office space, whole skyscraper or a complete city, will always push the limits of polygon count. Our nodes have Win x64, 4GB of RAM & they fail if scene ram & Virtual memory push past 6GB together, until user allows VM to be expanded.

If I'm using high resolution texture maps 4096x4096, that can be up to 3GB of texture maps alone for a single scene, then add geometry of 5 million triangles. How can a 1GB GPU possibly render something like this?

Also why limitation on render resolution by Octane? Do renders have to be constrained, ie 1x1 or are aspect ratios possible up to the current limit of 16.7 Mega pixels possible? ie 5616x2988 etc etc

Tim.