Arion by Random Control

[Cesar R]

Sorry to post on the FryRender forum but I m not sure where to ask.

I am looking at Arion as well as "its competitor" and was wondeing if anyone here is usin it. One of the things I like about Arion from what I have been reading is that is uses CPU + GPU (CUDA) which can provide a much faster render / feedback? ( I am assuming).

 

Thanks,

[RayOfLight]

which competitor of Arion do you mean?

I'm using Arion, and am not convinced (yet) that it has competitors. I can use it to rapidly render using CPU & GPU and I can easily change materials, lights, render elements, thus creating nice realistic results!

[Cesar R]

I was refering to Octane.

 

I was just wondering, In Arion, can you ADD or REMOVE or even move light sources? I think for architectural viz you would have all you fixtures in place and have rigged the entire scene, but what about for a studio setup? where you would need to interactively move the lights and such? Right now it seem that you do the work in your modeling app, go to arion and if you need to make any of the above changes you need to go back to the original model?

[RayOfLight]

You can't move anything in Arion (yet?). But you can change color and intensity of your lights. Which is the same as turning them on or off, and you can do this with an animation too. (animating the lights going out)

[Cesar R]

so what happens if: You have a scene. You export it as an OBJ? bring it into ario and assign materials. All of the sudden you realize that you need to move something or better yet you need to add one more light somewhere. You obviously have to go back to you modeling application (3dsmax for instance or Rhino). Make the changes, re-export and start over again?

[RayOfLight]

yes.

Arion is only for rendering and applying materials. It is not a modeller, nor is it some kind of "ActiveShade" (3dsMax uses this for "rendering" inside a viewport).

As with all renderers, as soon as you need to change the geometry, you have to go back to your modeller application (3dsMax/Rhino etc.) and change it, and render again. (Exporting is part of the render process for Arion).

For instance, when using Vray, you would have to start the render again too. Not with Vray RT, but that is running inside 3dsMax.

Maybe a future version of Arion will allow objects to be moved. RandomControl used to have the name VirtualEditor on their website, which is gone now and replaced by Arion. No idea what the future might bring

 

also, rendering in Arion in mostly fast enough to not care about the re-rendering

Edited September 12, 2010 by RayOfLight
additional info

[Jan]

There are more engines than octane and arion. In example, smallgpulux:

 

and only with CPU, thea:

(from 0:56)

 

http://www.thearender.com/resources/Thea_interactive_render_web/index.html

Edited September 12, 2010 by Jan

[Fran]

so what happens if: You have a scene. You export it as an OBJ? bring it into ario and assign materials. All of the sudden you realize that you need to move something or better yet you need to add one more light somewhere. You obviously have to go back to you modeling application (3dsmax for instance or Rhino). Make the changes, re-export and start over again?

 

Hi Cesar,

 

Since you have Rhino and 3ds Max, and both apps have supporting plugins, you would not need to import OBJ scenes into Arion. That is mainly meant as a courtesy for people without a supported host application. You can save any changed materials to a .frm file and reload them into your host scene.

[RayOfLight]

exactly like Fran said, you can assign materials in Rhino or Max (or any of the other supported programs).

You would only have to restart the rendering if you want to add/move/change geometry.

All material/camera/environment settings can be changed in Arion.

[Cesar R]

I just downloaded the Fry demo. I tried one of the other suggested programs and I did like that you can control all the geometry (posing) but I dont like the interface and it doesnt seem very supported. I remeber that maxwell would also let you move things if needed but once again I feel confortable with the fry interface. Is the main advange of Arion over Fry the speed and ability to use the GPU ?

 

I wrote and e-mail to support asking who are they targetting with Fry verus Arion. I am awaiting a response.

[RayOfLight]

Arion and Fryrender are still quite different, but it's in the details.

The biggest thing, in Arion, you render with CPU and GPU (and possible other CPU's and GPU's on the network), which results in a tremendous speed increase. You can therefor render again and again at low time costs. You can change materials, camera settings and environment in Arion, but also can you change the position of your camera!

In Fryrender, you can stop a render, and continue it later on. Also you can render once, and during or after the render is done you can change the light settings (color & intensity) in almost realtime (without rendering again).

Personally I like to use them both, and quite a lot of our customers too. They use Arion to test materials and several light setups and camera positions. Then they save that and use Fryrender in a renderfarm (or on multiple computers in their network) to render very large formats (i.e. 15k x 9k).

Arion needs CUDA, so you need a nVidia card, for example a Geforce 470, which has 1.2Gb of memory and 448 CUDA cores. This means that when rendering with that card, the whole scene will have to be loaded onto the card...you can load quite a lot of data in 1.2 GB (I've seen scenes of 15M polygons with 400Mb of textures, being rendered on Arion at 5000x5000...I think that's nice...)

[Cesar R]

I notice you mention what I understand at test rendering in Arior and then moving over to Fry for the final? - Is this because of Fry's ability to render on a farm (is this is something arion cannot do?)

[RayOfLight]

Arion can also render on a farm, however you would have to add nice CUDA cards to your renderslaves. When you have renderslaves without CUDA cards, you can use Fryrender on it. Besides, there are some things Fryrender can render and Arion cannot (yet), like coloured glass and Subsurface Scattering for instance.

[Cesar R]

um.. good to know. Its a shame the you have to buy both program seperatly also.

[Cesar R]

I have a question about the GPU's. Can I have more than one Graphics card to render in Arion? Let's day that today I get an FX550 because of buget but in the future I can affort a much better card or a GTX4000 series. Can I use more than one card to push the rendering and do they have to be the same or SLI ?

[AJLynn]

Don't get an FX550 if the plan is to use CUDA. Or the more current 580. If that's your budget, get a Geforce 450.

[RayOfLight]

indeed, better use the geforce card, 450s, or 470s are nicely priced (at least I think so ) It has 448 cuda cores...and over here it costs €350,- (or something). Probably a lot less in the US :S

 

Also, you can buy Arion and Fryrender as bundle, saving €300,- (with which you can buy an extra GTX470 if your MB can handle that..)

You cannot use SLI, but you can use different types of cards, although I'd stick with one brand.

You could use a geforce220GT for your OS and use one to three GTX470 cards just for Arion.

[Cesar R]

I cannot use SLI and is - Do not buy an SLI motherboard, or Arion just does not care if its an SLI motherboard or not. Or just not connect them with the SLI bridge?

[AJLynn]

For CUDA you don't need to SLI the multiple video cards, you can just stick them in.

[RayOfLight]

exactly what AJLynn said, CUDA doesn't use SLI. Just plug in an extra CUDA card (preferably the same type or at least brand to avoid driver conflicts) and Arion will detect it.

[AJLynn]

Does brand actually matter? nVidia has unified drivers these days. I should think that the most important thing would be having the same core type (GF100 vs GF104 vs G92 etc.) so you know they support the same version of CUDA... really, to be safe, you want the same GPUs and RAM configs and it's better if you can make the cards all be identical.

[RayOfLight]

you are correct, what I meant is that I'd stay away from using nVidia and ATI (you can use ati as primairy card, but I wouldn't do that)

[Cesar R]

^

I wanted to post the following on the Random control board, but although I am registered I am not yet a customer and I cannot read or post on some subforums.... Something that I think should be changed to help inform potential customers such as myself.

 

Anyway, If someone can answer the following it would be great:

 

What is the effect of the PCI slot speed with Arion? I am looking an some motherboards but most of them only have 1 PCIe X16 along with 2 PCIe X8.

If I were to place 2 graphic cards one of the X16 and one on the X8, how does Arion handle this?

 

Can I have a dedicated car for video on the X16 slot which Arion does not utilize during rendering and rather would use additional cards (GTX470's) installed on the adjacent X8 PCIe slots?

 

I have the otption to purchase an micro-ATX motherboard with 2 PCIe X16.... if that is better. In which case both cards would be used for rendering.

[Fran]

What is the effect of the PCI slot speed with Arion? I am looking an some motherboards but most of them only have 1 PCIe X16 along with 2 PCIe X8.

If I were to place 2 graphic cards one of the X16 and one on the X8, how does Arion handle this?

 

Can I have a dedicated car for video on the X16 slot which Arion does not utilize during rendering and rather would use additional cards (GTX470's) installed on the adjacent X8 PCIe slots?

 

I have the otption to purchase an micro-ATX motherboard with 2 PCIe X16.... if that is better. In which case both cards would be used for rendering.

 

The bus speed of the pci slot only affects the amount of time it takes to upload the scene to the gpu and that is only done once, when the scene is opened. Actual render time is not affected.

 

If you have the space, it's a good idea to have a lesser card for handling the UI.

[AJLynn]

So basically, since these MBs that can handle 4 PCIE cards at x16 because they have added PCIE bridges are expensive, one strategy would be to get an MB that can handle one at x16 and two or three at x8, use the x16 slot for a card that drives the display (say, a FirePro card or something) and the others for dedicated CUDA cards, since the one for the display needs to trade data with the system constantly and the CUDA cards don't need as much bandwidth.