When antialiased artwork produces many tiny color layers in Krita, compare higher Fuzziness settings, but do not choose the one with the fewest layers by default. The Split Layer dialog uses Fuzziness to control how precise color matching is: increasing it lowers precision. Choose a setting only if edge colors become manageable while important neighboring color regions remain distinct. The manual explains the control’s role, but does not prescribe a numeric value or guarantee a particular result for a given image.
A controlled comparison
Use a copy of the artwork so you can compare outputs without changing the source. In Krita, open Split Layer on the same target layer and keep the split mode and all other options fixed. The official manual describes both splitting colors into paint layers and outlining them as local selection masks; use the mode that fits the intended result and do not compare different modes as though Fuzziness were the only variable. Krita’s Split Layer manual also notes that some options are unavailable for selection masks.
- Identify the intended major colors before splitting. For a flat icon, write down the four fills that should remain separate and identify the antialiased outline or boundary pixels to inspect.
- Choose three candidate settings: low, middle, and high relative to the control’s available range. These are comparison points, not universal thresholds. Record the displayed values so you can reproduce each run.
- Start each candidate from a separate copy of the same unsplit target layer, rather than splitting a previous result again. Change only Fuzziness; keep opacity handling, output mode, and other dialog options unchanged. Save each output separately and inspect the same boundaries.
- Inspect whether the output shapes preserve the four intended design regions as separately usable results. Do not infer the algorithm’s pixel assignments from similar colors or layer names. Mark uncertain boundary regions for manual cleanup; reject an output that cannot preserve a required region even if its layer list is simpler.
The manual says higher Fuzziness is necessary for antialiased images because otherwise tiny pixels may each receive a separate layer. It does not specify how the algorithm assigns every intermediate pixel, so judge the actual output rather than assuming that all edge pixels will join a particular neighboring color. The documented Fuzziness description supports this precision trade-off, not a fixed recommendation.
Hypothetical acceptance worksheet
The following figures and observations are invented for a planning example, not measured Krita output. Suppose an icon has four intended flat colors and an antialiased outline. Use the worksheet to compare outcomes; do not expect these layer counts from your image.
| Candidate | Hypothetical observation | Decision to record |
|---|---|---|
| Low | Many tiny edge-color results; four main fills appear separate | Reject if fragments make the intended layer set impractical |
| Middle | Fewer edge fragments; each intended region can still be used separately | Consider acceptable after inspecting output shapes; flag uncertain areas for manual cleanup |
| High | Simplest layer list, but the narrow border is no longer separately usable | Reject if the output loses a required design region; this observation does not establish the algorithm’s pixel-assignment rule |
For each run, note the actual setting, output mode, and whether the questionable pixels belong to the outline or to a neighboring fill according to the design. Color appearance is a clue, not a reliable substitute for the layer or mask type label shown in Krita. Record whether the border shape is separately usable, and list any uncertain areas for manual cleanup, rather than recording only the number of outputs.
Limits and next steps
Fuzziness is a matching-precision control, not a general edge-smoothing setting. Raising it can reduce over-splitting, but lower precision can also make distinctions between nearby colors less reliable. The manual does not document a universal safe setting, a guaranteed layer-count reduction, or a rule that determines which adjacent color wins at a boundary.
If every candidate either leaves too many edge fragments or merges an important region, stop increasing Fuzziness. Revisit the artwork’s intended color regions and decide whether those boundary pixels should be treated as part of a fill, an outline, or a separate transition. Then repeat the comparison with the same major colors and the same dialog options. If the split mode itself is wrong for the task, choose between paint layers and local selection masks based on the desired output, rather than trying to compensate through Fuzziness.
Does higher Fuzziness always reduce the number of layers?
The manual explains why greater fuzziness can help with antialiased pixels, but it does not guarantee a particular layer count. Inspect the result.
Should I choose the setting with the fewest outputs?
No. Keep the main color boundaries faithful to the design, even if that means accepting some edge fragments.
Can I use the same setting for every image?
The manual provides no universal value. Compare candidates on the artwork you are splitting.
Sources and next step
This guide interprets public documentation and uses explicitly hypothetical acceptance examples, not application tests. Controls and availability can change; verify the cited manual against your installed version.
For work created in Imagild, carry the input requirements and acceptance checklist into the photo editor. The external software controls discussed above belong to their respective products.