Description du produit
Give a hair mesh real strand flow without repainting its texture. Nine hair types, all of them the awkward ones to draw by hand: drill curls (vertical ringlets), dreadlocks, cornrows, wet hair, the angel-ring highlight band, the gather point of a ponytail, backcombed messy hair, a wolf cut with wispy ends and thick twintail bunches. Drop one into the Normal Map slot of lilToon (or any shader), set the tiling, and the light starts following the strands. No color is baked in, so the hair keeps its own color — the flow is the only thing that changes. Every map is a periodic function evaluated on a torus, so the tile joins itself exactly and there is no seam at any tiling value. 2048px for PC, 1024px for Quest, height maps included, 27 files. Not tied to any avatar — it also works on world props and cloth. None of these nine appears in our '12 Hair Normal Maps + 8 MatCaps' pack, so the two sets do not overlap.
No generative AI. No image-generation AI was used in this pack. Every map is generated by code: each material is written as a periodic function on a torus, evaluated into a height field, converted to a tangent-space normal map and written straight to pixels — then checked and packaged by hand. That is also why the tiles have no seam: the function joins itself exactly at the edge, so there is no patched-up border to hide. Declared here because both itch.io and Jinxxy require the origin of the artwork to be stated.
What's inside
- 9 hair folders (01-09) — each with the normal map at 2048px (PC) and 1024px (Quest), plus the 2048px height map
- Drill curls (vertical ringlets), dreadlocks, cornrows, wet hair, angel ring
- Ponytail gather point, backcombed messy hair, wolf cut (wispy ends), thick twintail bunches
- 27 files in total
- READ_ME_FIRST.txt — license and usage
Specs
- 9 hair types, normal maps at 2048 x 2048 px for PC and 1024 x 1024 px for Quest, both included
- Seamless at any tiling value. Each map is a periodic function evaluated on a torus, not a photo patched at the edges, so there is no border to give it away however far you crank the tiling
- Not tied to any avatar. A normal map sits in a shader slot, so it works on any model, PC or Quest, and on world floors, walls and props
- No color is baked in. A normal map only changes how light hits the surface, so the outfit keeps its own colors and its own texture
- OpenGL convention (Y+). Unity and Blender take them as they are
- Height maps (grayscale, 2048px) come with every hair type
- No design here appears in our '12 Hair Normal Maps + 8 MatCaps' pack — it is a separate set of hair types, so owning both gets you no duplicate file
- No matcaps in this pack. The same eight matcaps are already in '12 Hair Normal Maps + 8 MatCaps' and in '30 MatCaps', so nothing here is a repeat purchase
How to use
- lilToon: tick Normal Map, drop in the 2048px PNG, then set Tiling for how fine the strands are (2-6 for hair) and Scale for how deep it looks (0.3-1.5)
- In Unity, set the texture's Texture Type to 'Normal map' first
- Quest: use the 1024px folder — the headset downsizes anything larger
- Any other shader (Standard, Poiyomi, Blender Principled BSDF) takes the same file in the same slot — OpenGL convention (Y+), no flipping needed
- Pair it with a matcap for the shine — our '30 MatCaps' pack drops straight into lilToon's MatCap slot
- The height maps are only there if you want to bake at a different strength or drive displacement
Update
2026-09-07 update: the green channel of every normal map was corrected to the OpenGL convention (Y+). Unity, lilToon and Blender now read them as-is. If you had flipped the previous version yourself - with lilToon's "Convert normal map (DirectX <-> OpenGL)", a negative Scale value, or by inverting green in an image editor - please remove that step and re-download the files.
License
Royalty-free. Use them in commercial and non-commercial work — games, TRPG sessions, videos, streams, VTuber overlays, print. No credit required. You may edit, crop, recolor and combine the files freely.
Not allowed: redistributing or reselling the assets themselves, claiming authorship, or using them as training data for AI / machine learning models.