Blender 4.2: Precise Modeling Workshop
Learning Resources →

Architectural visualization artists working with Blender and rendering in Cycles knows that image noise is a common problem. If you start your render and choose a low amount of samples for the render, by the end of the process, you will have an unusable image with lots of noise. To get rid of noise you have to raise the sample value, which will result in higher render times or use a technique like a Denoising tool.

An average amount of samples for an architectural visualization scene will stay between 1500 and 2500.

What if I told you that a recent research found a way to generate noise free images with only one sample. It is not a typo; I mean one sample.

The technique has a name of Spatiotemporal Variance-Guided Filtering. Here is a video showing a demo of the process.

As a way to produce images quickly, it is an impressive technique! You will get clean images instantly. On the downside of the process, you will also notice that images lose a lot of detail. In a very similar way of the denoising tool in Blender 2.79.

How to get that working in Blender?

Someone would have to take the paper and implement the technique in the Blender source. Since it is still quite recent, look to the date of the paper pointing to last month.

The good news is that a significant amount of features from Blender came from papers just like this one. All you need is a volunteer to implement the technique.

By the time we have a feature called Spatiotemporal Variance-Guided Filtering in Blender, you will be able to render scenes in Cycles with one sample and no noise!

If you want to learn about the Denoising tool in Blender, take a look on our premium subscriptions. The Blender 3D Architect subscribers receive downloadable tutorials explaining how to use features like Denoising. Here are the available plans:

Previous

Free bathroom furniture download

Next

Free download of a bathroom in Blender and Corona

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Check Also

Clicky