Compositing Nodes#

Clarisse has been designed to offer a set of specialized nodes meant to eliminate the mandatory back-and-forth between the lighting and a compositing application needed to visualize the render in context when working with layers, live action plates and AOVs.

Clarisse compositing engine has also been designed with quality control in mind, so that lighting artists can make sure final compositing artists have all the different render passes/AOVs they need to reconstruct the beauty in the compositing package they use for example.

Compositing nodes are a set of nodes related to compositing. Most of these nodes process something from input connection(s) and output an image. The 3 most common compositing nodes you will be likely using are Image Node Read which reads an image from disk, Image Node Render which renders a Render Scene and Image Node Write which write an image to disk. With the exception of a few nodes such as Image Node Write, that inherits from the base class Process, most compositing nodes inherits from the base class Image Node.

Common Attributes#

Nodes inheriting from Image Node class can all be displayed by the Image View. To display an Image Node in the Image View, either directly drag its output to the Image View or while hovering in the Build View press X to initiate a link and plug it to the Image View Node or alternatively press 1, 2, 3, 4 to automatically plug it to the desired Image View slot.

!!! Tip New to Clarisse? The Image View in Clarisse is the dedicated user interface element that displays images or renders in your builds. When the current node isn’t evaluated (for example an Image Node Render or an Image Node Read), the Image View will automatically request an evaluation to display it. For more information about the Image View, please refer to Image View.

Image Node Constant#

The Image Node Constant, generates an image color matte.

Image Node Render#

The Image Node Render is used to render an input Render Scene. In other words, the Image Node Render translates a input 3D Scene to a 2D image through rendering.

Deep Rendering#

It is possible to compute Deep data while rendering in output to output a Deep EXR. By default, deep rendering is disabled. It can be enabled using Enable Deep Rendering attribute.

Deep Rendering Options#

You will find quite a few attributes to control deep rendering.

Deep Output#

Deep Render output is performed using a process Image Node Write Deep on which you connect the Image Node Render. For more information, please refer to Image Node Write Deep.

Image Node Write#

Unlike most compositing nodes, an Image Node Write is actually a process node. The role of the Image Node Write process is to save an input image to disk. To learn more about processes and rendering in using the build assembly workfow, please respectively refer to process nodes and setting up renders.

To write an input Image Node in an interactive session, you have to select the Image Node Write and press Run Process in the Attribute Editor.

Image Sequence File Pattern#

The Filename attribute supports a special tag # to format the output image sequence pattern. This special tag allows you to specify where the current frame should be located in the output image filename. This tag is also used to format the number of digits of the output frame number.

For example:

• /path/to/my/seq.## outputs to /path/to/my/seq.01 for frame 1 and /path/to/my/seq.159 for frame 159
• /path/to/my/seq_####_final outputs to /path/to/my/seq.0001 for frame 1

Image Node Write Deep#

Image Sequence File Pattern#

The Filename attribute supports a special tag # to format the output image sequence pattern. This special tag allows you to specify where the current frame should be located in the output image filename. This tag is also used to format the number of digits of the output frame number.

For example:

• /path/to/my/seq.## outputs to /path/to/my/seq.01 for frame 1 and /path/to/my/seq.159 for frame 159
• /path/to/my/seq_####_final outputs to /path/to/my/seq.0001 for frame 1

This node status bar displays useful information in the Build View.

Image Node Read#

The Image Node Read is used to read a still image or a sequence of images from disk. While it may sound as its main purpose is be to bring live action plates into your compositing graph, it does actually a lot more than its obvious usage.

If you carefully look at Image Node Read attributes, you can notice that it can take as Input an Image Node Write process. It may look curious at first glance but it is actually a key feature to improve the flow of your builds and add the support of image disk caching.

Consider the following example. Let’s say you have an Image Node Render representing a render layer that you’d like to save to disk. Naturally, you would connect it to an Image Node Write. However, there’s a good chance that you would like to merge it with others in order to get a full precomp of your layers. As a result you node graph would look like the following one.

Let’s say that you already partially or entirely rendered your render layers. You may want your compositing graph to directly read the rendered frames from disk instead of having the renderer re-rendering them and the read node allows just that. This is where the Input Image Node Write ability of the Image Node Read becomes very handy. Let’s consider the following graph.

As soon as you connect an Image Node Write to a Image Node Read, you can notice that the Filename attribute of the node is greyed out in the Attribute Editor. By default, the read node will load the image file if it exists on disk, otherwise it requests an evaluation to the image node connected to the write node. The behavior of the read node can directly controlled using its Input Mode attribute.

This workflow is very handy as the flow of your graph is very elegant while benefiting from a controllable image disk caching feature.

Image Node Reformat#

The Image Node Reformat is used to resize an input image to a new resolution.

Resize Type#

Since there's no single way to resize an image, the Reformat node offers multiple ways to resize the input image.

Filter#

It is possible to apply a filtering algorithm to resample the pixels of the resized image.

Image Node Color Space#

Clarisse internally works in linear space. This means that all input images should be transformed to linear prior any compositing operations. The Image Node Color Space allows to transform the color space of input Image Node. For more information about color management in Clarisse please refer to Color Management.

For example, unlike EXRs which are almost always stored in linear, PNGs are likely to be encoded in sRGB. In other words, PNGs must be transformed to linear so that compositing operations and display looks are applied correctly. In that case, you must plug your Image Node Read to an Image Node Color and set Input Color Space to sRGB and Output Color Space to linear.

Image Node Unmult#

This Image Node Unmult divides (unmultiplies) RGB channels of the input image by the alpha.

if (a > 0) {
    r = r / a;
    g = g / a;
    b = b / a
}

Image Node Premult#

This Image Node Premult multiplies RGB channels of the input image by the alpha.

if (a > 0) {
    r = r * a;
    g = g * a;
    b = b * a
}

Image Node Merge#

The Image Node Merge, merges two or more Image Nodes resulting in a new image. This node is typically used to merge render layers together or to merge foreground/background live action plates to 3D renders.

Stacking Multiple A Inputs#

It is possible to specify multiple A inputs. In that case additional A inputs are considered on top. For example A #1 is under A #2 which is under A #3 etc… In other words looking at the Attribute Editor, the top most input A is the last entry of the A list.

read2 is the top most input. read2 is applied on top of read1

Merge Operation Mode#

The type of merge operation can be controlled using the Merge Mode attribute. By default the Merge node performs an A over B operation.

Resulting Alpha#

By default, the merge node merges the input alphas together using the merge operation specify by Merge Mode. However, it is possible to change this default behavior by using the Alpha attribute.

Varying Input Image Window Behavior#

You may have input images with different image window sizes, in that case the merge node resulting image window is, by default, an union of all input image windows. In other words, the resulting image window will grow to the input image window maximum extent. However, you may want to change this behavior and in that case you can use the Window attribute.

Image Node Grade#

The Image Node Grade is used to perform basic grading operations on the input image.

Image Node Reorder#

The Image Node Reorder is used to redirect an image layer to an another one. This node is meant to reshuffle image layers and image channel layers.

Reshuffling Image Channel Layers#

Using Destination Channel Order it is possible to reshuffle the source channel layer to the destination one using the following characters:

Examples:

Image Node Multiply#

Multiply the input image by a specified color.

Image Node Blur#

The Image Node Blur, blurs the input image.

Blur Filter Type#

Image Node Diff#

The Image Node Diff, return a specified color if the difference between input A and B is higher or equal to a specified threshold. This node is extremely useful when comparing two images together. For example, you can use this node for quality control to see if the beauty you reconstructed using AOVs perfectly matches the rendered beauty pass.

Background Mode#

Using the Background Mode attribute you can control the color that is returned when the difference between A and B is lower than the Difference Threshold.

=== "Background Color (default)

Varying Input Image Window Behavior#

You may have input images with different image window sizes, in that case the resulting image window is, by default, an union of all input image windows. In other words, the resulting image window will grow to the input image window maximum extent. However, you may want to change this behavior and in that case you can use the Window attribute.

Image Node Offset#

The Image Node Offset, offsets the input image in X and Y. Please note that the offset node doesn’t perform subpixel transformations.

Image Node Optix Denoiser#

The Image Node Optix Denoiser denoises an input render image using NVIDIA OptiX Denoiser. This node requires a NVIDIA based graphics card. For more information about the denoiser its hardware specifications please refer to Optix Denoiser.