WEBVTT 1 00:00:00.010 --> 00:00:05.160 Sampling is a setting for all three render engines in Blender at the moment, but in this 2 00:00:05.160 --> 00:00:08.820 video we'll be specifically talking about sampling in Cycles. 3 00:00:08.820 --> 00:00:13.700 You can argue that out of the three render engines, Cycles relies on sampling the most. 4 00:00:13.700 --> 00:00:17.620 The higher the samples, the more accurate the light calculations will be. 5 00:00:17.620 --> 00:00:21.420 But instead of explaining it, let's just demonstrate this real quick. 6 00:00:21.420 --> 00:00:26.000 If we go into our Cycles render engine, we can play around with these sample settings. 7 00:00:26.000 --> 00:00:30.720 For the sake of demonstration, let's bring the render samples down to something low. 8 00:00:30.720 --> 00:00:32.440 Let's say 10. 9 00:00:32.440 --> 00:00:38.280 Now we can render out a test image by going to the render menu and selecting render image. 10 00:00:38.280 --> 00:00:40.160 Or you can press F12. 11 00:00:40.160 --> 00:00:44.560 If you look very closely, you can see that the image is very grainy, and has a bit of 12 00:00:44.560 --> 00:00:46.360 noise. 13 00:00:46.360 --> 00:00:50.080 Now what I want to do is preserve this render so I can compare it to the new render I'm 14 00:00:50.080 --> 00:00:52.980 about to make with some tweaked settings. 15 00:00:52.980 --> 00:00:56.000 We can do that by simply changing the slot number. 16 00:00:56.000 --> 00:01:00.540 You can change your slot number by selecting a new slot from this drop down menu here, 17 00:01:00.540 --> 00:01:04.100 or pressing J, or any number on the number row. 18 00:01:04.100 --> 00:01:08.920 I'm going to go ahead and go to slot 2, which will save our previous render in slot 1 as 19 00:01:08.920 --> 00:01:11.220 we render a new test. 20 00:01:11.220 --> 00:01:16.960 So now we can go ahead and go to our sample settings again, and change it to 400. 21 00:01:16.960 --> 00:01:21.400 Now if we render it, you'll notice first of all that it will take a lot longer. 22 00:01:21.420 --> 00:01:24.400 This is the byproduct of increasing the number of samples. 23 00:01:24.400 --> 00:01:28.160 However the trade off is that our new render is much much smoother. 24 00:01:28.160 --> 00:01:31.940 There is very little noise and it looks and feels great. 25 00:01:31.940 --> 00:01:35.400 This is the difference between low samples and high samples. 26 00:01:35.400 --> 00:01:39.200 However not every person or project will be able to afford the longer render times for 27 00:01:39.200 --> 00:01:40.700 higher samples. 28 00:01:40.700 --> 00:01:45.740 So let's say if we absolutely have to render a project with lower samples that has noise, 29 00:01:45.740 --> 00:01:47.760 what would that look like? 30 00:01:47.780 --> 00:01:51.700 Well we can simply shorten our playback region to be about 10 frames or so and render out 31 00:01:51.700 --> 00:01:52.700 some tests. 32 00:01:52.700 --> 00:01:59.000 Let's bring our samples back down to 10 and hit ctrl f12 to render an animation. 33 00:01:59.000 --> 00:02:02.340 As you can see there is definitely noise in this animation. 34 00:02:02.340 --> 00:02:06.680 But there's something interesting about it because it's the same noise in every frame, 35 00:02:06.680 --> 00:02:09.320 even when the camera and objects are moving. 36 00:02:09.320 --> 00:02:14.160 This can be a bit jarring and very distracting in animations, but there is a value under 37 00:02:14.160 --> 00:02:17.640 the advanced options called seed. 38 00:02:17.680 --> 00:02:21.960 Seed is basically just another word for randomizer as it provides random algorithms that can 39 00:02:21.960 --> 00:02:24.680 be re-referenced with a seed index. 40 00:02:24.680 --> 00:02:29.400 But anyway, this seed is what drives the random pattern of our noise in our render. 41 00:02:29.400 --> 00:02:33.720 If you prefer not to have the same pattern of noise in every frame of your render, simply 42 00:02:33.720 --> 00:02:37.120 press the clock next to the seed value. 43 00:02:37.120 --> 00:02:42.360 Let's try the render again but with a different filename output to compare the difference. 44 00:02:42.360 --> 00:02:47.200 As you can see, the noise is now different for every frame, giving it a more natural 45 00:02:47.200 --> 00:02:49.360 look even if it's grainy. 46 00:02:49.360 --> 00:02:54.840 Now this is great, but ideally we still don't want any noise at all, but we still want to 47 00:02:54.840 --> 00:02:56.880 have short render times. 48 00:02:56.880 --> 00:03:01.640 Believe it or not, this is actually possible with the denoiser. 49 00:03:01.640 --> 00:03:06.200 Cycles actually comes with a denoising function that will help remove a lot of the noise left 50 00:03:06.200 --> 00:03:08.720 over from lower sample renders. 51 00:03:08.720 --> 00:03:12.920 This setting is not actually located in the render settings, but rather in the view layer 52 00:03:12.920 --> 00:03:13.960 settings. 53 00:03:13.960 --> 00:03:18.720 Simply go there and scroll to the very bottom to check the denoising checkbox. 54 00:03:18.720 --> 00:03:23.280 Naturally there are more settings under the denoising checkbox that you can tweak if necessary, 55 00:03:23.280 --> 00:03:26.280 but default settings work pretty great already. 56 00:03:26.280 --> 00:03:30.360 Sampling is a very important part of the Cycles workflow, and hopefully this video helps you 57 00:03:30.360 --> 00:03:34.000 understand it enough to get started with making your renders even better.