Snowy mountains

Hi, I have made this very short animation with cycles:Snowy mountains.
Let me know what you think of it.

Hi !

I think it’s pretty cool ! :slight_smile:
Some tips to make it better would be :

  • Getting rid of the noise is the mountain… Are you are using the denoiser with few samples at render ?
  • Using the bezier interpolation mode for the camera keyframes (it looks like you are in the linear mode, because movement are quite brutal)


The denoising radius is 8; the strength is 0,50 and the feature strength is 0,50. The camera keyframes are vector

Nice work. You did a good job! :slightly_smiling_face:
Just to let you know, if you paste your link on it’s own line, it will embed the video into your post.

Thank you. Here is the embedded video:

I changed the interpolation mode for the camera keyframes, but I can’t get rid of the noise: I rendered the animation with 228 samples as you can see in the embedded video:

Then I tried 528 samples but after 24 frames I stopped it because there was still noise (see the attached animation:

How can I eliminate the noise? (I’m still using the denoiser with the same parameters)

To be honest, I don’t see noises there.

Are you using adaptive subdivision and microdisplacement ?

Yes, I’m using adaptive subdivision and microdisplacement. So you think that the animation with 528 samples is ok?

I think the “noise” you think is not noise due to sampes by pixel value. It’s kind of gemeotry artefact.

Maybe with this video where you can see the same artefacts you’ll understand.

So there is nothing to do?

I would say that there is never “nothing to do”

In fact I agree with warnotte, it looks like it’s not a “render sample” noise. It seems to be geometrical. Your Mountain mesh seems to be so dense and so sharp, that when the camera moves, geometry make some face appear and disappear.

Two ideas :

  • Maybe you messed up with the camera clip planes ? (camera “start” and “end” limits) For example if your scene is “huge” you push up the camera “end” param, but the results is that the precision in depth buffer becomes bad. It means that faces very close to each other will randomly overlap in different orders when the camera is moving. So you have to push up camera start too. (Depth buffer precision decreases depending on how big is “end” and how little is “start”. The bigger “start” the better. The smaller “end” the better.)

  • If the above solution doesn’t work : It could be only due to a mesh which is too high poly and too sharp. Try adding a “decimate” modifier to reduce the number of vertices (If the mesh is textured, the UV re-mapping is automatic with this modifier). A “smooth” modifier to reduce a little bit sharpness might help too.

Let me know

1 Like

I will try a render with 128 samples, the denoiser as above, the camera start limit 5.3 and end limit 100.000; before, the camera start limit was 0.100. Do you think that 128 samples is a too low value ?

I modified the camera clip planes and I have applied the decimate modifier 2 times (in the collapse mode with ratio 0.2) but I have the same problem.