LuxCoreRender Forums

I would recommend to only compare final renders, not viewport renders.
Both Cycles and LuxCore use rather suboptimal rendering settings/engines when doing viewport renders because they have to be optimized for low latency and fast interactiveness.

B.Y.O.B. wrote: Fri Mar 09, 2018 11:36 am I would recommend to only compare final renders, not viewport renders.
Both Cycles and LuxCore use rather suboptimal rendering settings/engines when doing viewport renders because they have to be optimized for low latency and fast interactiveness.

I Understand but here i want to show the banding problem in the viewport as i think it is something not normal. this banding problem is clearly visible when using generic sunsky.

Also note that the render on the right Top are final render.

Sharlybg wrote: Fri Mar 09, 2018 11:33 am READ NOTE ON IMAGE FOR BETTER UNDERSTANDING:

According the images you are doing the comparison using RTPATHCPU. Sincerely, it doesn't make any sense.

In general, it isn't a good idea to make comparisons with 1-10secs long renderings because the startup time may be too important or measured at different moment by different renders. Something in the 2-5 minutes is probably more accurate, longer is the time and more accurate is the result.

The banding problem is something worth investigating and may be the result of some wrong truncation in BlendLuxCore or LuxCore image related code. It may be also the result of Cycles using some kind of dithering.

I don't know what is going on but remenber that a particular setting greatly improve my render time when rendering interior scene. it was under advance metropolis sampler setting :
Especially the max consecutive rejection ( i like to set it to 128 to avoid lot of noise).
Really hope it'is possible to bring it back

Ok, implemented them.
I'm a bit unsure about the description of the "Image Mutation Rate".
The manual says "Maximum distance over the image plane for a small mutation", but when should I set this to a different value? I guess if I render caustics, I could lower it a bit so the sampler has more focus, is that correct?

B.Y.O.B. wrote: Wed Apr 04, 2018 7:11 am I'm a bit unsure about the description of the "Image Mutation Rate".
The manual says "Maximum distance over the image plane for a small mutation", but when should I set this to a different value? I guess if I render caustics, I could lower it a bit so the sampler has more focus, is that correct?

It is how far 2 consecutive samples can be on the image plane. For instance, a value of 0.1 on a 100x100 image means: not farther than 10 pixels (i.e. 0.1 = 10% of 100).

B.Y.O.B. wrote: Wed Apr 04, 2018 7:11 am Ok, implemented them.

Great!
Thank you

Dade wrote: Wed Apr 04, 2018 8:13 am It is how far 2 consecutive samples can be on the image plane. For instance, a value of 0.1 on a 100x100 image means: not farther than 10 pixels (i.e. 0.1 = 10% of 100).

Yes, but when should I change this value?
Or is it impossible to give general advice for this and we have to try it out for each scene?

B.Y.O.B. wrote: Wed Apr 04, 2018 2:49 pm

Dade wrote: Wed Apr 04, 2018 8:13 am It is how far 2 consecutive samples can be on the image plane. For instance, a value of 0.1 on a 100x100 image means: not farther than 10 pixels (i.e. 0.1 = 10% of 100).

Yes, but when should I change this value?
Or is it impossible to give general advice for this and we have to try it out for each scene?

It is likely large mutation parameter, for instance:

- large mutation = 1.0 and max. image distance = 1.0 and Metropolis works exactly like RANDOM sampler;

- large mutation = 0.01 and max. image distance = 0.01 and Metropolis will obsessively sample paths with more light contribution (up to the point to not even sample some region of the image plane).

So it can intuitively think to the value as a "strength" of how much it should be "not uniform" in sampling. Anyway, it is quite complex stuff, probably not worth changing for most users.

Sharlybg wrote: Fri Feb 16, 2018 3:20 pm Motion blur doesn't work on emitted particles: Alpha5
Blend
particule motion blure.blend

Just remenber about this. Do we support it now ?