How to Make a Vtuber Avatar on Procreate
StyleFrame Team7 min read
To make a VTuber avatar in Procreate, design the character on a large canvas, paint it as separated layers (hair, eyes, mouth, body, accessories), and export it as PSD or PNG files. Then bring those files into a rigging or streaming tool such as Live2D or a PNGtuber app. Procreate handles the illustration, but it does not rig or track your face, so the avatar only moves once you assemble and rig the layers elsewhere. The main constraint is Procreate's layer limit, which depends on canvas size and your device, so most artists split a model across several files.
Decide What Kind of Avatar You're Making
The workflow depends on how much movement you need.
- PNGtuber or pose-swap avatar: You draw a handful of full-character states, such as neutral, talking, surprised, shy, and eyes closed. You export each as a transparent PNG and switch between them in streaming software. This is the fastest route and needs very few layers.
- Rigged 2D avatar (Live2D style): You draw the character split into dozens or hundreds of parts so each piece can bend, blink, and sway. This is far more demanding on layer management, and it is where Procreate's limits matter.
Beginners should start with a chibi or half-body design. Fewer parts means fewer layers and fewer hidden areas to paint.
Step 1: Design the Character and Collect References
Before opening Procreate, settle the essentials: silhouette, hair shape, color palette, outfit, and signature props. Decide whether the avatar is face-only, half-body, or full-body, because that sets your canvas proportions and part count.
Quick thumbnails in a sketchbook or Procreate are enough. Gather references for the hair in particular, since hair strands are usually the most layer-hungry element on a rigged model. Also plan the expression set you want, like a blush, a tilted head, closed eyes, or a surprised mouth. It is much easier to design for those states now than to retrofit them after the art is finished.
Step 2: Set Up the Canvas
Canvas size controls both quality and how many layers you get.
- Create a canvas at least around 3,000 px wide. Some Live2D artists go taller, using 4,000 to 5,000 px depending on character height, because the model needs to hold up when scaled.
- Check how many layers the canvas allows in the Layers panel before you commit. If you can't get a comfortable number, reduce the dimensions or plan to split the model.
Because Procreate's layer cap depends on factors like resolution and your device, a very large canvas gives you detail but fewer layers. You'll be balancing both throughout the project.
Step 3: Sketch With Symmetry Assist
Most VTuber faces and bodies are close to symmetrical, so use Drawing Guide in the Canvas settings and turn on Symmetry. Mark the layers you want to mirror as Drawing Assist. Mirrored strokes make the eyes, bangs, and shoulders consistent and save a lot of time.
To make it faster, assign Drawing Assist to a shortcut in the gesture controls so you can toggle it per layer without digging into the panel.
Finish the sketch with more detail than you'd normally need. Include accessories, hair layering, and color notes. Changing proportions after you've split the model into parts is painful, so settle the design first.
Step 4: Block In Color on a Flat Reference
Before separating anything, flatten your sketch into one reference layer and block in base colors. Working on a single layer makes global adjustments easy: you can shift hue, saturation, and value across the whole character before committing to parts.
One practical note from artists who rig in Live2D: imported PSDs can look slightly less saturated than they did in Procreate. If you notice that, push the saturation up a little on your reference before painting final parts, then verify in your rigging software.
Keep two copies of the reference: a low-opacity one for tracing and a full-opacity one for picking colors.
Step 5: Paint the Model in Separated Parts
This is the core of the process. Instead of one illustration, you're painting a kit of pieces that can move independently. For a rigged avatar, typical separations include:
- Face: base skin, blush, nose, and shadows under the bangs
- Eyes: upper lash, lower lash, iris, highlight, and eye whites as separate pieces, left and right
- Brows and mouth: each in its own layer, plus alternate mouth shapes if you want them
- Hair: front bangs, side locks, back hair, and any ahoge or tied sections, each split so it can sway
- Body: neck, torso, arms, hands, and clothing layers
- Accessories: earrings, ribbons, headphones, and other props that need their own motion
Paint the hidden areas too. If a bang covers part of the forehead, draw the forehead underneath. If the eyelid closes over the iris, the iris needs to extend beyond the visible area. Any movement later will reveal gaps that weren't drawn.
Use clipping masks and separate shading layers where it helps, but be careful, since each of those counts against your layer budget.
Step 6: Work Around Procreate's Layer Limit
A fully rigged model can easily need more layers than a single large Procreate canvas allows. The standard workaround is to split the model across multiple files:
- Duplicate your finished sketch file.
- Crop each duplicate to one region, such as the head, torso, or legs.
- Paint that region's parts in its own canvas.
Cropping out empty space shrinks the pixel area, which in turn lets you hold more layers. Many artists use two to five files and reassemble them later. When you crop, keep a note of each file's offset from the original canvas so the pieces line up cleanly. Using the snapping or guide tools to center and crop consistently helps.
Step 7: Export and Reassemble
For a pose-based PNGtuber avatar, export each state as a transparent PNG and you're done with Procreate.
For a rigged model, export each part-file as a PSD. Then:
- Open the PSDs in Photoshop or Clip Studio Paint.
- Combine the layers into one document at the original canvas dimensions, aligning each piece according to your notes.
- Check layer names, because clear naming (such as "eye_L_iris" or "hair_front_01") saves time during rigging.
- Confirm that transparency, layer order, and blend modes survived the export. Blend modes and effects may not carry over exactly, so merge or bake problem layers if needed.
Do a test import into your rigging software early, before you finish every part. A single export quirk is far cheaper to fix on a half-finished model.
Step 8: Rig, Track, and Stream
With the assembled PSD, you can:
- Rig in Live2D to create blinking, head turns, breathing, and hair physics.
- Use a simple tool (a PNGtuber app or pose-swap overlay) to switch images when you talk or press hotkeys.
- Bring the avatar into OBS or your streaming software with face tracking.
Procreate's job ends at the artwork. Rigging quality depends on how cleanly you separated and extended your parts.
Common Mistakes to Avoid
- Starting too small. Low-resolution canvases blur when the avatar is scaled for streams or thumbnails.
- Skipping hidden areas. Missing paint under hair, eyelids, and clothing creates visible gaps once parts move.
- Over-layering early. Spending layers on fine shading before you've finished the base parts can leave you stuck at the cap.
- Changing the design after splitting. Lock the proportions before you divide the model into files.
- Waiting to test export. Run a PSD test early to catch color shifts, blend mode changes, and alignment problems.
- Naming layers carelessly. Rigging hundreds of unlabeled layers is slow and error-prone.
Quick Workflow Recap
- Choose avatar type: PNGtuber or rigged.
- Design the character and plan expressions.
- Create a large canvas, around 3,000 px wide or more.
- Sketch with Symmetry and Drawing Assist.
- Block in color on a flat reference.
- Paint separated parts, including the hidden areas.
- Split across multiple files if you hit the layer limit.
- Export PSD or PNG, reassemble, and test.
- Rig, track, and stream.
Procreate is a capable place to draw an avatar as long as you plan your layers around its limits. Get the separation and export workflow right, and the rest of the pipeline, from Live2D onward, becomes much easier.