OctaneRender™ Standalone 3.06 TEST 4

A forum where development builds are posted for testing by the community.
Forum rules
NOTE: The software in this forum is not %100 reliable, they are development builds and are meant for testing by experienced octane users. If you are a new octane user, we recommend to use the current stable release from the 'Commercial Product News & Releases' forum.
Post Reply
calus
Licensed Customer
Posts: 1308
Joined: Sat May 22, 2010 9:31 am
Location: Paris

Zay wrote:
calus wrote:(also why did you remove the render target from the orbx ? was not easy to find back your camera and settings)
Click the RenderTaget Node button.
lol
Again, there is no rendertarget node in your ORBX, no camera node, no kernel settings.
octane_2017-03-24_17-43-13.png
But nevermind, As I said I get around that and found that as your lights overexpose the scene a lot,
you must have lower the exposure without lowering expected exposure at the same time,
the reason why you have this result.
Pascal ANDRE
Zay
Licensed Customer
Posts: 1123
Joined: Sun Jan 17, 2010 2:53 am

calus wrote:Again, there is no rendertarget node in your ORBX, no camera node, no kernel settings.
Yes there is :) Click on "Geometry Group" to start the render. Then you have all the settings in the Node Inspector.
Win 11 Pro | i5 12600K | 32GB ram | 2x GTX 1080Ti + 3080Ti - studio driver 560.94| Modo/Blender/ZBrush/Daz/Poser
calus
Licensed Customer
Posts: 1308
Joined: Sat May 22, 2010 9:31 am
Location: Paris

Zay wrote:
calus wrote:Again, there is no rendertarget node in your ORBX, no camera node, no kernel settings.
Yes there is :) Click on "Geometry Group" to start the render. Then you have all the settings in the Node Inspector.
So you don't use any rendertarget but the preview of the Geometry Group node, ok why not.

Finally this has nothing to do with expected exposure, but I don't see any problem either,
you have high noise treshold0.05 and a gamma of 0.9
So the result is as expected, a part of the noise you see is the limit you defined and is made more visible because of the tone-mapping,
Also most of the noise is just because you need a higher max samples than 7500,
to finish to clean all the non-green area of the noise map.
Pascal ANDRE
User avatar
mojave
OctaneRender Team
Posts: 1340
Joined: Sun Feb 08, 2015 10:35 pm

BorisGoreta wrote:One of the nodes won't authenticate.
Capture.JPG
I have just sent you a PM with some instructions to either confirm what I think is happening or try to figure out.
vijay
OctaneRender Team
Posts: 196
Joined: Sun Jul 24, 2016 8:42 pm

mikinik wrote:vijay_thirukonda, this happens also with pathtrace kernel, I do not show it because I have only a slow gpu.
There is no error if I disable alpha shadows, the noise threshold is reached first at the light areas of the image as it should be. With portals adaptive sampling does not work properly?
alpha shadows on
07.jpg
alpha shadows off
08.jpg
11.jpg
orbx - https://www.dropbox.com/s/bs1n54x4ur28k ... _.zip?dl=0

path tracing kernel:
12.jpg
13.jpg
I think that it should always be, regardless of alpha shadows:
14.jpg
Path tracing kernel alpha shadows off.

Hi Mikinik,


1. we are planning to give the option to switch on/off the usage of expected exposure value used in the noise estimate. Incorrect use of this setting will give some unexpected result. eg: when a scene is overblown with gamma. Also, we are planning to tweak the expected exposure calculation a bit.

2. The camera response curve already has a default gamma in it, On top of the default gamma, this scene's gamma is set to 2.2. Ideally, the response curve should be set to linear/off or gamma settings should be 1.0. Due to overblown gamma, the noise near the window is not visible and the direct light contribution is fairly good in the interior than some parts near the window. Thus it is looking like adaptive sampling noise threshold is reached in the interior before the clean near the window.

I will try to post some picture later.

Thanks
Vijay
vijay
OctaneRender Team
Posts: 196
Joined: Sun Jul 24, 2016 8:42 pm

justix wrote:Here something that I'm trying to understand since Test 3. Although Every scene is different and so should AS values I got a very easy and little scene were the noise just won't clear easy, Now.. the image is only 5000 samples 2 light sources and it took 37 min to leave it quite noisy and you can see the Noise map a bit empty.

Then without changing the Threshold leaving it at 0.030 I used the Region Render and just after kind of 7 minutes I got so much noise cleared out and wondering why it hasn't detected in the first place not even a bit, region render obviously takes all the power need to quick render up but no noise is detected at 0.010 or should I increase samples at 10000 .. I though that it could kick in and start clear fast..instead I'm curious to get things cleared out on the matter.

edit: Tested at 10000 SAmples AS threshold 0.020 and almost no noise detected from stairs and around the lamp. twice the time and AS did not kick in almost nothin..basically same as 5000 samples.. Pease I really don't get how this is happening. very simple scene just for test but intriguing.. :(
Render target10000noise.png
Render target10000beauty.png

Hi,

Looking at the images your scene still noisy at 10k sample. Need more samples to get down the noise for this scene. Noise map at the top of the stairs is dark so it should be reaching the threshold soon but it is quite noisy at the bottom And similar happening around the lamp that not all samples find the light source. I see you set the expected exposure to 14, I guess you should be matching it with imager setting, with the high exposure the noise becomes visibly lesser after tone mapping, but it will still exist in our calculations. You can increase noise threshold if you are happy with the noise in the scene or increase the max samples. Not all the scene could get benefits from adaptive sampling.

Note: Only the most cleanest pixels will reach 0.01 threshold.

Cheers
vijay
User avatar
justix
Licensed Customer
Posts: 585
Joined: Sat Feb 27, 2010 8:01 pm

vijay_thirukonda wrote:
justix wrote:Here something that I'm trying to understand since Test 3. Although Every scene is different and so should AS values I got a very easy and little scene were the noise just won't clear easy, Now.. the image is only 5000 samples 2 light sources and it took 37 min to leave it quite noisy and you can see the Noise map a bit empty.

Then without changing the Threshold leaving it at 0.030 I used the Region Render and just after kind of 7 minutes I got so much noise cleared out and wondering why it hasn't detected in the first place not even a bit, region render obviously takes all the power need to quick render up but no noise is detected at 0.010 or should I increase samples at 10000 .. I though that it could kick in and start clear fast..instead I'm curious to get things cleared out on the matter.

edit: Tested at 10000 SAmples AS threshold 0.020 and almost no noise detected from stairs and around the lamp. twice the time and AS did not kick in almost nothin..basically same as 5000 samples.. Pease I really don't get how this is happening. very simple scene just for test but intriguing.. :(
Render target10000noise.png
Render target10000beauty.png

Hi,

Looking at the images your scene still noisy at 10k sample. Need more samples to get down the noise for this scene. Noise map at the top of the stairs is dark so it should be reaching the threshold soon but it is quite noisy at the bottom And similar happening around the lamp that not all samples find the light source. I see you set the expected exposure to 14, I guess you should be matching it with imager setting, with the high exposure the noise becomes visibly lesser after tone mapping, but it will still exist in our calculations. You can increase noise threshold if you are happy with the noise in the scene or increase the max samples. Not all the scene could get benefits from adaptive sampling.

Note: Only the most cleanest pixels will reach 0.01 threshold.

Cheers
vijay
Thanks, as a matter of facts the 0.030 Threshold value is the one that works the most out of many scenes, I agree it is quite of an odd scene but that is a stress test ..somehow..The exposure was set to 12-13 , my question was actually how the render region be so effective in those areas just after few passes while the noise cleared out by it is not yet detected.. So if I set the same to 15000 I won't get the noise greening out like if I'm using the Region tool.. I t could well be a technique I see..if only we could split the render with a snapping grid...
Win 7 64 | 2 X MSI AERO GtX 1070| Intel I7-6850K| 32 GB DDR4 RAM | Asus X99 II-A
coilbook
Licensed Customer
Posts: 3032
Joined: Mon Mar 24, 2014 2:27 pm

alternatives to noise threshold in AS???

It seems that areas that are directly lit should get a separate noise threshold I think.
Because 0.1- 0.2 is more than enough for direct lit areas but indirectly lit areas especially dark interiors would need around 0.04 or less.

But right now if we set it to 0.03 we waste a lot of time waiting for direct lit areas to reach their noise levels even though they look done way before the system colors them green.

But if we set to 0.1 - 0.2 direct lit areas looks fine but indirect still look grainy because system uses one threshold for both direct and indirect.

Thanks
coilbook
Licensed Customer
Posts: 3032
Joined: Mon Mar 24, 2014 2:27 pm

exposure in Imager don't forget to also lower the "expected exposure" in Adaptive sampling.


can we get a sync check box next to as exposure not to forget to sync it to camera exposure
coilbook
Licensed Customer
Posts: 3032
Joined: Mon Mar 24, 2014 2:27 pm

will normal mapping be fixed soon. Where we have to set it to none? Unfortunately we have 100s of textures and cannot use 3.06 because of normal mapping bug
Post Reply

Return to “Development Build Releases”