OctaneRender™ Standalone v1.0 RC 3 (obsolete)

[pixelrush]

More manual stuff:
p63 should mention the preferences/controls panel for multiplier, mouse settings/app presets
p66 actually deep channel is first in the UI dropdown
p76 PMC items out of order compared to UI screenie
p77 tangent vector missing
p79ish should mention sRGB colour picker space I think and discuss gamma etc in octane...and btw no mention anywhere of right click with material picker to select hidden mat. - and also btw missing is clicking on bar for texture numbers etc
p96 might mention import glossy specular scale
p114 daylight power missing
p120 mention change speed and invert via preferences
p121 doesnt say where to enable these modes - show icons. Might be out of order placed here.
p125 missing near clip depth
p127 missing re camera is - min display samples, dithering and saturate to white
p130 missing - can add a coloured note, renaming nodes..

- also missing is setting the image resolution, viewport rulers, viewport resolution lock, how to clear recent files list.
- btw license mentions 1 hour reactivation period

[Synthercat]

Is it me or is the new glossy / reflection system bad?

Check out this comparison of 2.58e vs RC3
Let me also include the Red Africa granite I made...

I tried to change the node for the new RC3 to make it look like the old one but failed
I can't get it to gave this nice reflection at the yellow parts (especially around the ball)

[abstrax]

Is it me or is the new glossy / reflection system bad?

Check out this comparison of 2.58e vs RC3
Let me also include the Red Africa granite I made...

I tried to change the node for the new RC3 to make it look like the old one but failed
I can't get it to gave this nice reflection at the yellow parts (especially around the ball)

With beta 2.59 or beta 3.x we introduced correct handling Fresnel reflections, as well as allowing you to disable Fresnel completely. You now control the Fresnel reflection by the index input pin. A value <= 1 disables the Fresnel effect. Unfortunately it's pretty hard to automatically convert old materials and to calculate the correct index of refraction. Because of that the index value is set to 0 by default, which makes the specular reflection independent of the viewing angle. This affects all glossy materials that were created in older versions.

In your case maybe start with an index of 1.5 and then try to tweak both the specular pin as well as the index.

Cheers,
Marcus

[manalokos]

I've found a shading bug related to opacity, that happens only in RC3, and has already caused problems in the facade of a building in one render.
I've attached an image and a sample scene showing the bug.

The bug happens in near tangent surfaces that have lower than 1 in opacity but become black and opaque.
I've opened the scene in RC1 and it doesn't happen there.
Can you replicate this problem?

Best regards
Filipe

[ROUBAL]

I am encountering the same trouble : fully transparent surfaces seen with a low angle appear pure black, with all Kernels.

An other bad thing occurs when the object or the camera is far from the center of the scene.

On my current London street scene, requiring a very large scene (currently more than 600 meters of length, the buildings far away from the center (more than 200 meters) show black artifacts, weird shadows, black strips (horizontal and vertical black strips, visible on the sides of the buildings, between two buildings)... and the ground green surface shows from high altitude weird patches like overlaping faces, but there are no overlapping faces.

in a serie of instanced buildings, only the buildings far from the center of the scene show the artifacts. If I move them closer by modifying their own placement values, under around 200 meters, the artifacts disappear.

This problem is also present in RC1.

Proof screens :

[suvakas]

What? Problems start already at 200 meters?
That's not far at all by regular standards. I have large city scenes that are over 2 km's from origin. Those are Mental ray scenes though. Never tried to render them with Octane. Seems like we have a serious precision issue here.

Suv

[ROUBAL]

You are right, in the past (before instances) I made a large mountain scene with snow of 4km x4km, but the landscape was simple and there were no sharp edges. However, I didn't encounter artifacts.

About precision, I also noticed and reported that it is not possible to enter accurate values in placement nodes : you type for example 7.0 in X location, and when you open the field later, you read 6.99623 or something like that. Values never stay as typed ! This occured already in RC1, but I didn't get response for this report.

[roeland]

I've found a shading bug related to opacity, that happens only in RC3, and has already caused problems in the facade of a building in one render.
I've attached an image and a sample scene showing the bug.

The bug happens in near tangent surfaces that have lower than 1 in opacity but become black and opaque.
I've opened the scene in RC1 and it doesn't happen there.
Can you replicate this problem?

Best regards
Filipe

We can try to reproduce this issue here but then we need the actual geometry. If you send an OCS file you also need to send the obj file you're using.

You are right, in the past (before instances) I made a large mountain scene with snow of 4km x4km, but the landscape was simple and there were no sharp edges. However, I didn't encounter artifacts.

Does this landscape still render properly?

About precision, I also noticed and reported that it is not possible to enter accurate values in placement nodes : you type for example 7.0 in X location, and when you open the field later, you read 6.99623 or something like that. Values never stay as typed ! This occured already in RC1, but I didn't get response for this report.

Octane internally works with floating point numbers, which are stored in binary format instead of decimal. There is no 100% exact conversion between these floating point numbers and the decimal numbers shown in the interface. The precision of this floating point numbers is about 7 digits, so it can happen that 7.0 gets shown as 6.9999995 or so. There is no way it would change to 6.99623 though.

On my current London street scene, requiring a very large scene (currently more than 600 meters of length, the buildings far away from the center (more than 200 meters) show black artifacts, weird shadows, black strips (horizontal and vertical black strips, visible on the sides of the buildings, between two buildings)... and the ground green surface shows from high altitude weird patches like overlaping faces, but there are no overlapping faces.

There is a problem with planes that are exactly axis-aligned which is hard to solve without a major speed drop and increase in memory usage. The artifacts on the grass may show up because the rayepsilon value is set too low. Do these artifacts also show up in RC 2?

--
Roeland

[manalokos]

Hello Roeland,
I sent the OCS file in my last post, it is inside a .rar file as the forum server wouldn't allow to upload .ocs files directly, I forgot to put the .obj file, sorry, here it is again.

Best regards
Filipe

[ROUBAL]

@Roeland !

Does this landscape still render properly?

It is a scene that required 40 min to load on my system with 8GB of RAM only.
I finally got it : no noticeable artifact, but the scene is very different : no flat surface.

Tested with Rayepsilon =0.0001 instead of 0.00001 : the artifacts have disappeared ! Thanks a lot ! I don't master every parameter yet !

Octane internally works with floating point numbers, which are stored in binary format instead of decimal. There is no 100% exact conversion between these floating point numbers and the decimal numbers shown in the interface. The precision of this floating point numbers is about 7 digits, so it can happen that 7.0 gets shown as 6.9999995 or so. There is no way it would change to 6.99623 though.

I wrote it from memory... maybe I did an error : I have checked, and a value of -7.2 gives -7.1999998. So, it is because the first number after the point had changed that I thought that the error is bigger ! In fact, you are right, the error is at digit 7 after the point !

Thank you !