Skip to main content

Combine Page

The Combine page generates unique images by mixing layers from each trait. This is where your collection comes together.

Page Layout

  • Canvas (center) - Preview of selected combination
  • Right panel - Combination settings and layer overview
  • Gallery (center/grid) - Browse all generated combinations
  • Thumbnail navigation - Page through combinations

Understanding Combinations

A combination is one specific selection of layers:

eyes: original + hat: variation_1 + background: variation_2

The total possible combinations is calculated by multiplying:

(trait1_layers) × (trait2_layers) × (trait3_layers) × ...

Layer Overview

The right panel shows each trait and its available layers:

TraitLayersFavorited
eyes42
hat33
background53

Total combinations: Based on favorited layers only (if favorite-only mode is on)

Generating Combinations

Configure Options

  1. Use favorite layers only (default: on)

    • Only includes starred layers from Variations page
    • Reduces total combinations to your best work
  2. Include base image (default: on)

    • When on: base image is used as the background layer
    • When off: combinations render on a transparent canvas (layers only)
    • Useful for creating transparent overlays or layer-only compositions
  3. Limit (optional)

    • Set a maximum number of combinations
    • Useful for very large collections
  4. Shuffle (optional)

    • Randomizes which combinations are generated
    • Use with Limit for random sampling

Color Presets

If you've created color presets on the Colorize page, they're applied automatically during combination. Each preset generates an additional colored version of every combination:

Total = shape_combinations × (1 + number_of_presets)

No extra configuration needed here — colors are set up entirely in Colorize.

Generate

Click Combine All to start generation:

  1. VariantLab calculates all combinations
  2. Composites layers in order (back to front)
  3. Saves each result as a new image
  4. Progress updates in real-time
Mana Cost

Combinations cost 1 Mana per 100 images. Large collections can add up quickly.

Browsing Combinations

After generation, browse your combinations in the gallery:

  • Thumbnail grid shows all results
  • Click any thumbnail to preview in the canvas
  • Use pagination for large collections

Combination Details

Each combination shows:

  • Preview image
  • Display name (encodes which layers were used)
  • Creation timestamp

Managing Combinations

Deleting Individual

  1. Select a combination in the gallery
  2. Click the trash icon

Batch Delete

  1. Enable selection mode
  2. Check multiple combinations
  3. Click batch delete

Delete All

Click Delete All to remove all combinations and start fresh.

Irreversible

Deleted combinations cannot be recovered. You'll need to regenerate them.

Layer Order

Combinations use the layer order set in the Detection page:

1. background (rendered first, at back)
2. body
3. clothing
4. accessories
5. eyes (rendered last, at front)

If combinations look wrong (layers hiding each other), check your layer order in Detection.

Combination Naming

Each combination gets a display name encoding its layers:

final_eyes_v0_hat_v2_back_v1
  • v0 = original layer
  • v1, v2... = variation layers by prompt index

See Display Names for details.

Tips

  1. Start with a test batch - Generate 10-20 first to check quality
  2. Check layer order - Wrong order causes compositing issues
  3. Review favorites - Make sure you've starred your best layers
  4. Use limits - For 1000+ combinations, work in batches
  5. Preview before export - Catch issues before downloading

Troubleshooting

Layers are hiding each other

Check layer order in Detection. The last layer renders on top.

Some layers are missing

Ensure those layers are favorited in Variations (if using favorite-only mode).

Wrong layers are showing

Check that you're looking at the correct combination. Each has a unique layer set.

Combinations look wrong

The base image may have changed. If you modified masks or the base image, regenerate combinations.