Fighting vs fireflies

Post your tests, experiments and unfinished renderings here.
lighting_freak
Posts: 69
Joined: Thu Jan 18, 2018 6:02 pm

Re: Fighting vs fireflies

Post by lighting_freak » Fri Apr 06, 2018 8:01 pm

Hi,

First I'd like to mention that I totally agree with your basic approach to render real world scenario without adding some imaginary lights. I'm trying to do the same during my work.

Unfortunately this usually means that those renderings take a lot more time (samples) since realistic light path are harder to trace.

In your last post you've mentioned that you could turn some LED representing faces into a light emitting face.
That will definitely help tomake the lit lamps visible, but please be aware that the light cone is shaped from your reflector afterwards... it might generate strange illumination of the office room, since the surfaces aren't defined in an optical way (I guess).

One more thing that I'd like to mention. If I have to deal with lighting systems working in a very different way, I'm used to create single renders using individual settings for each system. I'm your case maybe sky with path trace and 80 ceiling spots with BiDir.
This makes your results independent of lighting strategy or path length (very interesting if you deal with translucent sub surface scattering volumes).
This will require an postpro step on top, but it can be set with blenders nodes technology.

BR
OS - Windows 7 X64
CPU - Intel CORE i7
GPU1 - Variants of notebook card from nVidia
GPU2 - Variants of notebook onboard card from Intel
Lux - Latest possible relaease

david57
Posts: 103
Joined: Mon Jan 22, 2018 11:30 pm

Re: Fighting vs fireflies

Post by david57 » Sun Apr 08, 2018 6:47 am

Ive been trying to work with this scene but its not stable. Some times it works fine but some times
all sorts of errors pop up. Lights work sometimes and sometimes they dont.
Another strange thing just popped up. Under the shelves are what looks like leaks.
Running late here right now, its 12:45 am. Goodnight and sleep tight.
Attachments
Errors.jpg

marcatore
Posts: 84
Joined: Wed Jan 10, 2018 8:04 am

Re: Fighting vs fireflies

Post by marcatore » Mon Apr 09, 2018 6:53 am

Dade wrote:
Fri Apr 06, 2018 3:21 pm
Yes, this may be considered somewhat a bug or a counter intuitive behavior, the halt condition is always evaluated for the complete image even when rendering a small region. So if you render 50% of the image with an halt condition of 1700 samples is like rendering 100% of the image with an halt condition of 3400 samples.

This because to achieve an average of 1700 samples per all pixels when rendering only half of them, you have to render 3400 samples.
Ok. good to know. I think that this should be explicited somewhere because , as it shown now, it seems an absolute number that is valid for both region and full render but it's not.
So..I'm thinking that should be explicited in the UI or in the tooltip or in the wiki or make an " automatic sample calcolator" that reduce the samples for region render according to the % of area rendered.
what do you think?

Anyway, I've tried multiplying the samples for the whole image but the final quality seems different (lower for the whole image).
This is a 22h render result with 8600 samples.
test_luxcore_8600samples_BiDir_21h50min.jpg
david57 wrote:
Sun Apr 08, 2018 6:47 am
Ive been trying to work with this scene but its not stable. Some times it works fine but some times
all sorts of errors pop up. Lights work sometimes and sometimes they dont.
Another strange thing just popped up. Under the shelves are what looks like leaks.
Running late here right now, its 12:45 am. Goodnight and sleep tight.
About the stability of the scene I really don't know what could be. It's quite an easy scene without particular modelling/shading solutions.
Are you able to post here the log errors?
About the light under the furniture I've already seen and I can't understand why. I've a suspect about the floor material. I'll try to make some tests with a simpler glossy material.
Thank you for your test.

EDIT: I did a quick test and it's not the floor material.
it seems that it happens when there is faces overlapping.
test_luxcore_overlapping.jpg
In this image the furniture on the left has raised by 1mm . The right one has the face on the botton that is perfectly over the floor
Is it planned to work in this way?

User avatar
Dade
Developer
Posts: 1144
Joined: Mon Dec 04, 2017 8:36 pm

Re: Fighting vs fireflies

Post by Dade » Mon Apr 09, 2018 9:27 am

It is simpler to just "fix" the halt condition to use the pixel count of the current rendering region instead that of pixel count of the complete image.

https://github.com/LuxCoreRender/LuxCore/issues/93
marcatore wrote:
Mon Apr 09, 2018 6:53 am
In this image the furniture on the left has raised by 1mm . The right one has the face on the botton that is perfectly over the floor
Is it planned to work in this way?
This may be a bug, try to isolate the problem, can you reproduce the problem with just a plane and cube ?
Support LuxCoreRender project with salts and bounties

marcatore
Posts: 84
Joined: Wed Jan 10, 2018 8:04 am

Re: Fighting vs fireflies

Post by marcatore » Mon Apr 09, 2018 12:29 pm

Dade wrote:
Mon Apr 09, 2018 9:27 am
It is simpler to just "fix" the halt condition to use the pixel count of the current rendering region instead that of pixel count of the complete image.

https://github.com/LuxCoreRender/LuxCore/issues/93
Ok.
Dade wrote:
Mon Apr 09, 2018 9:27 am
marcatore wrote:
Mon Apr 09, 2018 6:53 am
In this image the furniture on the left has raised by 1mm . The right one has the face on the botton that is perfectly over the floor
Is it planned to work in this way?
This may be a bug, try to isolate the problem, can you reproduce the problem with just a plane and cube ?
I've quick tried a new scene but it seems i can't replicate. I'll investigate more...

lighting_freak
Posts: 69
Joined: Thu Jan 18, 2018 6:02 pm

Re: Fighting vs fireflies

Post by lighting_freak » Mon Apr 09, 2018 12:52 pm

Hi,

I made some tests based on your scene:

Environmental lighting:
RGB_IMAGEPIPELINE_0 (3).png
Ceiling area light:
RGB_IMAGEPIPELINE_0 (2).png
Ceiling spot lights - turnes into face emitters:
RGB_IMAGEPIPELINE_0.png
The renders ran on my notebook only.
So there are only few samples per image done.

BR
OS - Windows 7 X64
CPU - Intel CORE i7
GPU1 - Variants of notebook card from nVidia
GPU2 - Variants of notebook onboard card from Intel
Lux - Latest possible relaease

marcatore
Posts: 84
Joined: Wed Jan 10, 2018 8:04 am

Re: Fighting vs fireflies

Post by marcatore » Mon Apr 09, 2018 1:19 pm

lighting_freak wrote:
Mon Apr 09, 2018 12:52 pm
Hi,

I made some tests based on your scene:

So there are only few samples per image done.

BR
Nice tests.
Some random thoughts:
- what about render times and samples?
- I can see that the bright pixel under the furniture is gone. Not really clear what should be the reason.
- what renderegine did you use? path for environment only and bidir for the others?
- how did you blend the 3 images to have all-in-one image?

Thanks in advance.

lighting_freak
Posts: 69
Joined: Thu Jan 18, 2018 6:02 pm

Re: Fighting vs fireflies

Post by lighting_freak » Mon Apr 09, 2018 9:12 pm

Hi,

I can't say anything about the rendering time.
My machine is very weak and I'm running more tests parallel. I started all your renders yesterday morning.
The good thing about the step by step method is that you can stop the one calculation (e.g. the sky one) if you think it's clean enough and put more power to remaing (harder) sessions.

There are the settings, I'm using:

Sky light source:
*Path CPU (don't have necessary GPU)
*SOBOL Sampler
*11 maxmimum bounces
**11 diffuse
**11 glossy
**10 specular
*1500 samples last image
RGB_IMAGEPIPELINE_0.png
Ceiling area source
*BiDir CPU
*METROPOLIS Sampler
*5 eye path
*5 light path
*460 samples last image
RGB_IMAGEPIPELINE_0.png
Ceiling spot sources
*BiDir CPU
*METROPOLIS Sampler
*10 eye path
*5 light path
*447 samples last image
** eye and light path might be better vice versa (more light paths for reflector interactions)
RGB_IMAGEPIPELINE_0.png
There is a very simple node setup to add the single images in the end:
composite.blend
(532.88 KiB) Downloaded 22 times
I hope this is helpful.

BR
OS - Windows 7 X64
CPU - Intel CORE i7
GPU1 - Variants of notebook card from nVidia
GPU2 - Variants of notebook onboard card from Intel
Lux - Latest possible relaease

marcatore
Posts: 84
Joined: Wed Jan 10, 2018 8:04 am

Re: Fighting vs fireflies

Post by marcatore » Tue Apr 10, 2018 7:19 am

Thank you @lighting_freak

I've downloaded the composite file and I've understood. I'll try as workflow.

david57
Posts: 103
Joined: Mon Jan 22, 2018 11:30 pm

Re: Fighting vs fireflies

Post by david57 » Mon Apr 16, 2018 1:22 am

I moved all the furniture away from the floor and walls.
Trying to render sections because it takes to long to render with Bidir.
This is just 1/3 of the whole scene because it takes to long to render.
This is 6100 samples and yet to grainy or blotchy.
Time was 2 hrs 40 minutes.
Maybe its all the lights that is slowing the render.
Attachments
Lights.jpg

Post Reply