2
$\begingroup$

I was wonder how I could make the camera in my scene look "low quality". Think like potato cameras from ghost videos. I'm currently using the noise from cycles, but someone said it would look better with something done in the composition. I used 10 samples so it would all render faster, so the noise hid all of that. Is there a way I could make everything look "low quality" with denoising and keeping the samples as low as possible for render times?

If you need anymore info or anything added let me know.

I want to add fuzz and some grain I guess.

enter image description here

$\endgroup$
4
  • 3
    $\begingroup$ Please show some pictures of what you call low quality images. Low sampling will give highly grainy images, this is a way to make them look low quality. $\endgroup$
    – moonboots
    Feb 1, 2021 at 5:13
  • $\begingroup$ @moonboots ive added an example image, i guess i want to add some noise and fuzz. When i use a denoiser with low samples, about 10 i used originally, it looks spotty and weird. I used 10 before because it was covered by the noise and saved a lot of time. $\endgroup$
    – Todahmoon
    Feb 1, 2021 at 13:18
  • $\begingroup$ In camera settings set depth of field to an empty, and experiment with the empty's position until you get desired blur. You can get noise from low samples count or add it in post-processing. $\endgroup$ Feb 1, 2021 at 13:24
  • $\begingroup$ @MarkusvonBroady what if I lowered the resolution of the camera from 1080p to something like 360p and added grain? $\endgroup$
    – Todahmoon
    Feb 1, 2021 at 18:38

1 Answer 1

2
$\begingroup$

There's many ways to achieve a low quality camera look, and as already covered in the comments, the most convenient ways to do so are low quality settings of the render: low resolution, low FPS if it's an animation, low sample amount, high compression of the image, resigning from filmic color space, and choosing many other sub-optimal or outright incorrect settings.

The actual challenge is to simulate a low quality camera realistically.

Motion blur

The easiest aspect to explain is exposure time: when you render an image in Blender, even in an animation, for each frame, all objects stay perfectly still and are rendered as such. Therefore even a flying helicopter's propeller will look still on a single image. Meanwhile, no camera is stopping the time while taking the picture, and most cameras will produce a visible motion blur:

You can enable Motion Blur in Render Properties tab, but contrary to the advises in the first paragraph, it will increase the render time.

Aliasing

Somewhat similarly, a digital camera has a sensor, which doesn't react to every single photon hitting it, but rather has a well for each subpixel, which collects photons, generating electrons inside, and then an electronic circuit calculates the subpixel value based on the charge density. This produces an averaged value of brightness for a given region. Meanwhile Cycles renderer works in an opposite way: not only it shoots a light ray outwards from the camera, but, more importantly, it's a single ray. A real camera recording a checkerboard moving away, will eventually perceive it as gray, as light coming from both black and white squares will hit the same well on the sensor.

2

I was able to render it with Cycles, because the default number of samples was enough to average the colors. However, watch what happens if I render with just 1 sample for each pixel:

3

The takeaway here is that the noise in Cycles is not the same as noise in a cheap camera - you won't get some artifacts like fireflies in a real camera. If you want to get a realistic grain effect, you have to use a high sample count and wastefully add the grain in post-processing. As far as I understand, the grain is a result of dithering data into a limited color space and then increasing the contrast, which moves these colors apart.

$\endgroup$

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .