Understanding and Utilizing DaVinci Resolve's Node Interaction Types for Advanced Color Grading

Cullen Kelly

Introduction [0:00]

The video clarifies that despite common terminology, DaVinci Resolve essentially has only one type of node for image manipulation: the corrector node. What are often referred to as different "node types" (e.g., serial, parallel, layer mixer) are actually different ways these corrector nodes are set up to interact with each other and process image data. Understanding these interaction methods is key to advanced color grading.

Deep Dive - Understanding Nodes [2:34]

The core concept is that a "node" in Resolve, for the purpose of image manipulation, is always a "corrector node."

• Other structures like "parallel nodes" or "layer mixer nodes" are not different node types but rather different configurations of corrector nodes designed to interact in specific ways.
• This perspective helps simplify the understanding of Resolve's node graph, focusing on the relationships between operations rather than inherent differences in the nodes themselves.
• The video uses a color-managed setup and a pre-applied look (Contour, Film Print 1) to demonstrate node interactions in a visually appealing context.

Node vs Node Interaction (Serial Nodes) [8:03]

The most basic interaction is serial processing, where nodes operate in a sequence.

• Serial Processing: Adjustments are applied one after another (A then B then C). The output of one node becomes the input for the next.
• Changing Order: Rearranging serial nodes changes the order of operations, which can sometimes lead to subtle or significant differences in the final image. However, for simple tonal adjustments, the visual difference might be minimal.
• Simplicity: Serial nodes are the simplest form of interaction and should be preferred if no greater complexity (like parallel or layer interactions) is required.

Parallel Nodes Deep Dive [11:49]

Parallel nodes offer a more complex interaction where adjustments are applied simultaneously.

• Mathematical Logic: Parallel mixers operate by taking the difference between a node's output and its input for each parallel branch, summing these differences, and then adding the original input back to the total.
  • Output P1 = (Output of Node 1) - (Input of Node 1)
  • Output P2 = (Output of Node 2) - (Input of Node 2)
  • Final Output = Input + Output P1 + Output P2 + ... (for all parallel nodes)
• No Hierarchy: Unlike serial or layer nodes, the order of nodes within a parallel mixer does not affect the final result because all adjustments are processed in parallel against the original input.
• Per-Pixel Operations: Parallel nodes are ideal for per-pixel operations (e.g., primaries, HDR palette, RGB mixer, custom curves, color slice tool), where the adjustment to one pixel is independent of its neighbors.
• Limitations (Textural Operations): Parallel mixers should not be used for textural or spatial operations (e.g., noise reduction, blur, midtone detail, grain, halation) because these adjustments depend on adjacent pixels and can lead to problematic results in a parallel structure.
  • Dedicated serial nodes are recommended for textural adjustments.

    

Layer Mixers Deep Dive [20:24]

Layer mixers introduce a hierarchical processing model, similar to layers in software like Photoshop.

• Hierarchy: In a layer mixer with the composite mode set to "normal," the lowest node in the stack takes the highest priority, effectively overlaying the nodes above it.
• Key Output Gain: Adjusting the "key output gain" in a layer node can introduce transparency, allowing the underlying nodes to contribute to the image.

  

• Composite Modes: Layer mixers leverage blend modes (e.g., Add, Multiply, Subtract, Divide, Difference, Luminosity, Color) to define how layers combine mathematically.
  • These blend modes often align with common blend modes found in other image editing software, and their mathematical formulas can be found online (e.g., Wikipedia).
  • Visual Math Example (HSV Saturation):
    • This technique demonstrates how to isolate and manipulate saturation using layer mixers.
    • Convert the image to HSV color space using a Color Space Transform node.
    • Use a splitter node to isolate the 'V' (Value/Brightness) channel.
    • Set up a layer mixer:
      • The bottom layer receives the original RGB image.
      • The middle layer receives the 'V' channel from the splitter.
      • The blend mode of the middle layer is set to "Subtract" to remove luminance, leaving only color information.
    • Apply desired saturation adjustments in a new node after the subtraction.
    • Add another layer mixer:
      • The bottom layer receives the saturation-adjusted image.
      • The top layer receives the original 'V' channel.
      • The blend mode of the top layer is set to "Add" to restore the luminance.
    • This complex setup visually demonstrates the underlying math of HSV saturation, even though Resolve offers a simpler single-node HSV saturation adjustment.

      

Splitter Combiner [27:29]

Splitter/Combiner nodes allow for channel-specific manipulation.

• Splitting Channels: A splitter node separates the three color components (e.g., Red, Green, Blue; or Hue, Saturation, Value) of an image into individual monochrome channels.
• Individual Channel Adjustments: Each channel can then be processed independently (e.g., adjusting the gamma of only the red channel).
• Recombination: A combiner node reassembles these separated and potentially modified channels back into a full-color image.
• Creative Uses:
  • Channel Swapping: Connect green output to blue input, and blue output to green input for a unique color shift.
  • Masking: Use one channel (e.g., the red channel) as a mask to limit adjustments in another node. This allows for highly localized effects.

Keys and Alpha Channels [36:40]

Keys and alpha channels define the areas where adjustments are applied.

• In-Node Qualifiers: A qualifier within a node (e.g., a luminance qualifier) limits the adjustments made in that node to specific conditions (e.g., only bright areas).
• External Key Inputs: A key (mask) can be generated in a separate node (e.g., using a qualifier or roto-shapes) and then "piped" into another node's blue triangle input. This limits the effect of the second node's adjustments to the area defined by the key, allowing for modular mask creation.

  

Using the Splitter Combiner to Check Noise [44:24]

The splitter combiner can be a diagnostic tool.

• Noise Analysis: By splitting an image into its RGB channels, colorists can visually inspect each channel for noise. Often, the blue channel exhibits the most noise, especially in underexposed areas.

  

• Targeted Noise Reduction: This allows for more precise noise reduction, applying more aggressive settings only to the noisiest channel (e.g., the blue channel) without over-processing other, cleaner channels. This can lead to a more natural and less noticeable noise reduction overall.

Luminosity Blend Mode and Contrast [46:34]

The Luminosity and Color blend modes offer control over tonal and chromatic aspects independently.

• Luminosity Blend Mode: When applied in a layer mixer, the Luminosity blend mode allows an adjustment to affect only the tonal (contrast) aspects of an image, leaving the color unaffected.
  • This is useful for making contrast adjustments without altering saturation, which often happens in a "normal" blend mode.
• Color Blend Mode: Conversely, the Color blend mode allows an adjustment to affect only the color of an image, preserving the original luminosity.
• Orthogonal Adjustments: These blend modes enable "orthogonal" adjustments, meaning changes to one property (e.g., contrast) do not inherently affect another (e.g., saturation), offering greater creative freedom and precision.

  

Creating an IR Look [51:41]

Simulating an Infrared (IR) look on non-IR footage is primarily a creative exercise.

• Lack of IR Data: Since standard camera sensors do not capture infrared light, it's impossible to "extract" IR information from regular RGB footage in a rigorous, scientific way.
• Creative Approach: To achieve an IR style, one must rely on creative grading choices that emulate the visual characteristics of IR photography (e.g., bright foliage, dark skies, specific skin tones).
• RGB Mixer for Monochrome IR: A practical approach involves converting the image to monochrome using the RGB Mixer and then manipulating the relative weights of the Red, Green, and Blue channels.
  • By adjusting these weights, you can make certain elements (like green foliage, which strongly reflects IR) appear brighter or darker, mimicking the contrast shifts seen in true IR images.
  • The viewer, not having seen the original IR, accepts these creative choices as the "IR look."

    

Node Composite Mode vs Layer Mixer [55:36]

The composite mode available in individual serial nodes is a simplified version of a layer mixer.

• Functionality: A composite mode on a single serial node can perform the same blending operations (e.g., Luminosity, Color) as a two-layer mixer node.
• Limitations:
  • It's limited to blending only two sources.
  • The "bottom layer" is always the image input to that node without any modifications within the blending operation itself.
  • Only the top layer can have its adjustments directly controlled within that node; the bottom layer remains static as its input.
• Use Case: This single-node composite mode is efficient for simple two-source blending when those limitations are acceptable, saving space in the node graph. However, for more complex multi-layer blending or independent control over multiple inputs, a dedicated layer mixer node structure is necessary.

Conclusion [1:01:20]

Understanding how nodes interact—through serial sequencing, parallel mixing, layer blending, splitting channels, or keying—is fundamental to mastering DaVinci Resolve. These interaction types provide the tools to precisely control image adjustments and achieve complex, nuanced looks in color grading.