RGBSplit takes an RGB color input and outputs separate heightfields derived from the Red, Green, and Blue channels. It is the inverse of RGBMerge: where RGBMerge packs three heightfields into a single color image, RGBSplit unpacks a color image back into usable scalar maps.

This is primarily a utility node for workflows where important data is packed into channels, for example masks or control maps authored in an image editor, generated by another tool, or exported from Gaea or other DCC apps as a single RGB texture to reduce file count.

When to Use It

Use RGBSplit when you need to unpack a channel-packed RGB texture into separate usable masks or control maps.

Common Uses

• Unpacking channel-packed masks. Common packing patterns include:

  • R = flow or rivers
  • G = deposition or sediment
  • B = wear, curvature, or a general mask The meaning is entirely up to your pipeline.

• Recovering separate maps from baked textures. If you received a single control map from an artist or another tool, RGBSplit lets you break it into independent heightfields for further processing.

• Converting a color texture into grayscale control data. In addition to channel extraction, RGBSplit can produce grayscale outputs based on standard luminance conversions. This is useful when you do not care about individual channels but want a consistent intensity map.

Channel Behavior and Interpretation

Each output is a heightfield derived from the corresponding channel intensity, typically treated as a 0-1 scalar range. Once split, you can feed those maps into standard terrain and mask workflows: remap, clamp, blur, terrace, erode, or use them to drive selectors and blends.

A frequent gotcha is that the source image may not be linear. If the input is an sRGB color texture, the channel values may not represent clean data, especially around midtones. For channel-packed masks, you generally want the input to be treated as linear data end-to-end.

Methods and Why They Matter

RGBSplit supports multiple splitting or conversion methods depending on what you are trying to extract:

• Average: Produces a grayscale map from the mean of R, G, and B. Use this when the image is essentially colored grayscale or when you want a quick intensity proxy without caring about perceptual correctness.
• BT.601: Converts RGB to luminance using BT.601 weights, common in SD or broadcast contexts. This tends to emphasize green more than red or blue and often looks more natural than a plain average for typical imagery.
• BT.709: Converts RGB to luminance using BT.709 weights, common for HD video and modern display assumptions. If you are converting modern color imagery to a grayscale control map, BT.709 is usually the more contemporary default.
• Single-channel extraction (for example RedChannel): outputs just one channel as a grayscale heightfield. This is the typical choice for packed-mask workflows where each channel has a specific meaning.

Autoleveling Channels

When splitting channels, it is common for each channel to have a different usable range. One channel might be very subtle, with values clustered in a narrow band, while another has full contrast. The automatic channel leveling option can normalize each channel independently so you get usable contrast immediately after splitting.

Use it when:

• You are dealing with unknown or inconsistent input textures.
• The channels look flat after splitting.
• You want faster iteration and do not want to manually remap each output right away.

Avoid it when:

• The channel values are calibrated data, for example if you encoded exact thresholds or weights.
• You need the outputs to remain numerically faithful to the original packed texture.

In calibrated pipelines, a better approach is usually to keep the split values untouched and then apply explicit Remap or Levels nodes downstream where needed.

Workflow Tips

• Treat packed masks as data, not color. If the source was authored for channel packing, ensure your import and export pipeline preserves linear values and avoids unintended sRGB conversions.
• Name and route outputs immediately. It's easy to forget what each channel represents. Rename or label the downstream nodes, or keep a small note macro, so the intent remains clear.
• Post-split cleanup is normal. Typical steps after RGBSplit include Remap, Clamp, Blur, or a light Gamma to tune mid-response, especially before using a channel as a selector mask.
• Prefer single-channel extraction for packed workflows. Luminance methods such as Average, BT.601, and BT.709 are best when you want image intensity, not when channels are carrying separate meanings.

RGBSplit is a small node, but it's a key enabler for channel-efficient pipelines and interoperability with external tools.

Properties