Page 2 of 3
Re: first test with water
Posted: Sat Jan 20, 2018 5:16 pm
by blibli
Dade wrote: Sat Jan 20, 2018 9:33 am
Anyway, I will work on this topic after the release of alpha2, it shouldn't be hard to fix.
thanks
Dade wrote: Sat Jan 20, 2018 9:33 am
you are using auto-linear tone-mapping and it was affected by the presence of fireflies (i.e. the image brightness was changing during the animation); without fireflies, the auto-linear use a more constant brightness and the pattern may be easier to spot.
Is there a way to have get stable values, like a pre pass, running the pathtracing for 10sec, getting good enough values for the tonemapping and then save it in scene custom properties or so? Or do you know another way to get good values quickly for videos?
Regarding fireflies, they could be used to make some nice bloom effect like that with an appropriate convolution filter:
Edit: how do I add images that are displayed in the message? It just write "image" for me.
Re: first test with water
Posted: Sat Jan 20, 2018 5:46 pm
by B.Y.O.B.
blibli wrote: Sat Jan 20, 2018 5:16 pm
Is there a way to have get stable values, like a pre pass, running the pathtracing for 10sec, getting good enough values for the tonemapping and then save it in scene custom properties or so?
This is basically like I implemented it.
You start a rendering and after a warmup of 10 seconds it the optimal clamp value is computed.
You can then just press a button in the render settings (below the clamping settings) to apply and use this value.
blibli wrote: Sat Jan 20, 2018 5:16 pm
Regarding fireflies, they could be used to make some nice bloom effect like that with an appropriate convolution filter
Support for bloom will be in the next alpha release, but I doubt you want it with fireflies, it will cause a lot of flickering.
Re: first test with water
Posted: Sat Jan 20, 2018 5:48 pm
by Dade
blibli wrote: Sat Jan 20, 2018 5:16 pm
Dade wrote: Sat Jan 20, 2018 9:33 am
you are using auto-linear tone-mapping and it was affected by the presence of fireflies (i.e. the image brightness was changing during the animation); without fireflies, the auto-linear use a more constant brightness and the pattern may be easier to spot.
Is there a way to have get stable values, like a pre pass, running the pathtracing for 10sec, getting good enough values for the tonemapping and then save it in scene custom properties or so? Or do you know another way to get good values quickly for videos?
As soon as BlendLuxCore supports tone mapping you will be able to pick the tone mapping you prefer, adjust the parameters, etc. At the moment the image pipeline used is just hard coded so it is not very flexible.
LuxCore has several image pipeline plugins, including a bloom filter:
https://github.com/LuxCoreRender/LuxCor ... ne/plugins
(like tone mapping, they are not yet supported by BlendLuxCore at the moment but it is only a matter of time)
blibli wrote: Sat Jan 20, 2018 5:16 pm
Edit: how do I add images that are displayed in the message? It just write "image" for me.
You click on the "Attachments" tab just on the left of the "Options", under the post you are writing, than click on "Add files" and "Place inline" to include them in your post.
Note: the current version of the forum software has a know bug when you re-edit a post with attachments:
viewtopic.php?f=10&t=49
Re: first test with water
Posted: Sat Jan 20, 2018 7:02 pm
by B.Y.O.B.
BlendLuxCore supports tonemapping (and the bloom plugin) since a few hours

Re: first test with water
Posted: Sat Jan 20, 2018 8:04 pm
by Sharlybg
B.Y.O.B. wrote: Sat Jan 20, 2018 7:02 pm
BlendLuxCore supports tonemapping (and the bloom plugin) since a few hours
Thanks you so much !
Re: first test with water
Posted: Sun Jan 21, 2018 7:53 am
by blibli
B.Y.O.B. wrote: Sat Jan 20, 2018 7:02 pm
BlendLuxCore supports tonemapping (and the bloom plugin) since a few hours
BYOB, the one coding faster than his shadow!
Re: first test with water
Posted: Sun Jan 21, 2018 8:21 am
by blibli
B.Y.O.B. wrote: Sat Jan 20, 2018 5:46 pm
blibli wrote: Sat Jan 20, 2018 5:16 pm
Is there a way to have get stable values, like a pre pass, running the pathtracing for 10sec, getting good enough values for the tonemapping and then save it in scene custom properties or so?
This is basically like I implemented it.
You start a rendering and after a warmup of 10 seconds it the optimal clamp value is computed.
You can then just press a button in the render settings (below the clamping settings) to apply and use this value.
Ok, I had the alpha1 version, with latest github, there is this optimal value button, however, it seems to set very high values. Here is a blend that still get fireflies after setting the optimal value from the warmup:
http://pasteall.org/blend/index.php?id=48770
I already said it, but the error log is really helpful, it was really easy to rebuild the broken node trees after the addon update.
The Nvidia kernel compile time is impressive... slow. But it's NVidia's job.
Re: first test with water
Posted: Sun Jan 21, 2018 10:45 am
by B.Y.O.B.
blibli wrote: Sun Jan 21, 2018 8:21 am
however, it seems to set very high values
The formula is: optimal_clamping_value = (average_film_luminance * 10)²
https://github.com/LuxCoreRender/BlendL ... er.py#L170
I just copied it from Dade and did not test it so far.
Maybe it does not perform well on some conditions, e.g. if the values are very high.
By the way, in your test scene disabling specular light visibility on the sun light removes all fireflies. But I realize this is a different matter from the sub-optimal clamp value.
Re: first test with water
Posted: Sun Jan 21, 2018 11:24 am
by Dade
The "* 10" is intended for normal renderings (with 2000-10000 samples per pixel). If you are in a corner case like rendering an animation preview and you want a strong clamping just reduce the value (but you have now a starting point, before you were totally blind).
The "* 10" could be transformed in a "Strength bar" with values, for instance, between 2 and 25

Re: first test with water
Posted: Sun Jan 21, 2018 7:46 pm
by blibli
Particles are enabled

I found 2 problems, they seem to be shift on Z when using emitter type. With hair, it makes an "list index out of range" error.
The file is here:
http://pasteall.org/blend/index.php?id=48776
Result with wireframe over to see the difference: