OctaneRender™ 2019.1 XB2

[FrankLIU]

thanks a lot for your informative reproduce details
it will help a lot to spot the bugs

[calus]

Some observations about layered materials :
- maybe the material layer pin should not be exposed in the base material of a layered material as it seems highly redondant in this case ?
- what is the reason to not make the "thin film layer" attributes a layer node type on itself ?

[PolderAnimation]

it looks like 'Affect alpha' does not work when there is no background.

[Goldorak]

Some observations about layered materials :
- maybe the material layer pin should not be exposed in the base material of a layered material as it seems highly redondant in this case ?
- what is the reason to not make the "thin film layer" attributes a layer node type on itself ?

I want to be able to add layers to existing materials in standalone without swapping in a new layered material which could break things (including loading in older octane scenes). So we have both dedicated layers materials (where the base can be swapped out easily) as well as layers on top of the existing materials in cases where you may not want to have the core material be a pure layered material.

[Notiusweb]

Some observations about layered materials :
- maybe the material layer pin should not be exposed in the base material of a layered material as it seems highly redondant in this case ?
- what is the reason to not make the "thin film layer" attributes a layer node type on itself ?

I want to be able to add layers to existing materials in standalone without swapping in a new layered material which could break things (including loading in older octane scenes). So we have both dedicated layers materials (where the base can be swapped out easily) as well as layers on top of the existing materials in cases where you may not want to have the core material be a pure layered material.

Thank you for taking this approach with the implementations made in Octane.

[calus]

Some observations about layered materials :
- maybe the material layer pin should not be exposed in the base material of a layered material as it seems highly redondant in this case ?
- what is the reason to not make the "thin film layer" attributes a layer node type on itself ?

I want to be able to add layers to existing materials in standalone without swapping in a new layered material which could break things (including loading in older octane scenes). So we have both dedicated layers materials (where the base can be swapped out easily) as well as layers on top of the existing materials in cases where you may not want to have the core material be a pure layered material.

I understand that but I was speaking only of hiding the pin when in a base material in a group layered material, indeed seems not possible in Octane to have different display of the same node type.

But what about extracting "thin film layer" attributes as a separated additionnal layer node ? besides It already has the perfect name: thin film LAYER
( this won't break previous octane scene as they don't use layer nodes)

[miko3d]

Away from PC but I will post a screenshot and pm a scene.

I had the chance to mess with the up & sub sampling and it is simply Awesome.
Some of the scenes now, even at 4K, just render so fast, I am blown away.
I see uses for both sub sampling and up sampling...
many times a video is compressed so the speed you can get on up sampled frames is wonderful.
But the Sub has me totally enthralled....
what kind of wizardry did you guys pull with the sub sampling...the speeds and detail are awesome.
I can render 8K images in under a minute with no denoiser???!!!....What the F is happening!

I can see now- this PLUS an RTX speedboost?....Holy SHiiiiiiiiiiiiiizzzz....

****APPLAUSE****

These are my thoughts:

sub sampling: exactly the same as it was in previous octane versions, , its just for quicker scene interaction at resolution expense.No AI magic here.No image improvements
Upsampling: the new upsampled AI mode.renders 2x/4x/ lower res and upsamples to the scene resolution.So far not very impressed to be honest, but just had a quick play with it.
Hope it helps
Mikel.

[Kalua]

Away from PC but I will post a screenshot and pm a scene.

I had the chance to mess with the up & sub sampling and it is simply Awesome.
Some of the scenes now, even at 4K, just render so fast, I am blown away.
I see uses for both sub sampling and up sampling...
many times a video is compressed so the speed you can get on up sampled frames is wonderful.
But the Sub has me totally enthralled....
what kind of wizardry did you guys pull with the sub sampling...the speeds and detail are awesome.
I can render 8K images in under a minute with no denoiser???!!!....What the F is happening!

I can see now- this PLUS an RTX speedboost?....Holy SHiiiiiiiiiiiiiizzzz....

****APPLAUSE****

These are my thoughts:

sub sampling: exactly the same as it was in previous octane versions, , its just for quicker scene interaction at resolution expense.No AI magic here.No image improvements
Upsampling: the new upsampled AI mode.renders 2x/4x/ lower res and upsamples to the scene resolution.So far not very impressed to be honest, but just had a quick play with it.
Hope it helps
Mikel.

I was not much impressed either. Maby I'm just not using it correctly... and maybe some user will post a tutorial showing how to make the best use of it.

[Notiusweb]

I find on my rig with 8 GPU that sub sampling starts the image render's top speed faster and reaches completion faster than with no sampling, with seemingly no loss of detail.
Then up sampling, I really like 2x2, but probably won't use 4x4 in all situations, because of the detail loss. But on less detail-intensive scenes, like a blob or water, awesome for this.

I am impressed by both, but OTOY I think this will make a lot of people interested in it and happy-
**Have a 1.5 x 1.5 Up-sampling also**.

Anything that maintains detail and up speed is great. The 2x2 does it, and I think a 1.5 mode, starting off at a higher res, also, would be wonderful because it would actually pay off for lower res final resolutions as well.
Alot of other products that do up-sampling have that - like a 200%, a 400%, but also a 150% mode.

Thus, on bigger final resolution scenes you could opt for 2x2 or 4x4, but then now for lower res final scenes it could even pay off if you could use 1.5x1.5, getting the speed boost and keeping more of that original detail at start.
Any further modes on the AI upsampling 'checkbox' itself would be interesting too, applying less or more amounts of that checkbox's sharpening/smoothing, tailored to the specific scene, could be cool to have.

THX!!!!

[wallace]

But what about extracting "thin film layer" attributes as a separated additionnal layer node ? besides It already has the perfect name: thin film LAYER
( this won't break previous octane scene as they don't use layer nodes)

Because ultimately thin film depends on an underlying BSDF, in which case is either specular or metallic.

Each layer represents one BSDF exactly currently.

Why do you want thin film in a separate layer? Can you please provide a use case?