Buffers

Altus needs two input images for each scene, which we call buffers b0 and b1. Each of these buffers must be rendered with a different sample seed such that noise varies between images. Altus will use the difference in noise patterns to remove noise while preserving detail. Two buffers are required.

Internally Altus will combine the two buffers, providing a denoised output that matches the quality level of both buffers added together. For example; if you want to denoise a render at 64spp (samples per pixel), then you can render twice at 32spp and Altus will output the equivalent or better of a 64spp render, in a fraction of the time.

Here is an image that alternates between two rendered images with a different seed. Notice the noise pattern differs between the two images.

../../_images/buffer_seed.gif

Altus needs this variance in noise to work effectively. If you are getting noisy results from Altus then check that all AOVs have 2 buffers with different noise patterns.

If your output still looks noisy, check Altus’ logs to see if there any warnings: Altus will warn you if images don’t appear to have enough variance. Note: if you feed Altus black AOVs, you’ll always receive warnings about lack of variance.

If the inputs are given correctly to Altus, you should end up with this denoised result:

../../_images/ChaplinCamera.denoised.png

Note

For more information on generating/using two buffers correctly visit our pages:

To pass two buffers visit: https://help.innobright.com/usage/basic-usage/

To pass two buffers at once by using side-by-side images. please see: Side-by-side images.

For more information on how to generate two buffers with your favorite 3d package and renderer see Integration with Third Party Software.