Wobble Maker Tips for Better Results: Brush Size, Strength, and Image Choice
Practical, technical tips for getting the best results out of Gootaku's free Wobble Maker — working resolution, brush size vs the motion grid, Fill All vs targeted painting, and image types that wobble best.
Wobble Maker is simple on the surface — paint a mask, pick a strength, export a GIF — but a few things about how it renders motion under the hood change what "good settings" actually means. These are the practical choices worth understanding before you paint your first mask.
---
Image resolution: why 480px is fine, not a limitation
Wobble Maker caps its working resolution at 480px on the longest side. Upload a 4000px source and it gets scaled down to fit within 480×480 the moment it loads — this isn't a quality shortcut, it's what makes the tool possible to run entirely in your browser with no server round trip. The motion mesh is a 24×24 grid layered over the image; at 480px, each grid tile covers a 20px × 20px block, which is plenty of resolution for the kind of soft, springy displacement the effect is going for. Running the same math against a full-resolution image wouldn't add visible smoothness to the wobble itself — the grid is still 24×24 either way — it would just mean pushing more pixels per frame for no motion-quality benefit, which is the wrong trade for something meant to render live in-browser.
The practical upshot: don't worry about sourcing a huge image specifically for Wobble Maker. A clean 800–1000px image downscales into a fine 480px working copy. What actually matters more than resolution is line clarity (more on that below) — a smaller, cleanly rendered image will wobble better than a huge image with fine detail that gets compressed away at 480px anyway.
---
Brush size and the grid: why tiny strokes can vanish
The motion mesh only has 24×24 = 576 tiles total across the whole canvas. At a 480px working size, that's a 20px tile. If your brush is set very small and you dab a stroke that's only a few pixels wide, you may be painting inside a single tile, or even only partially covering one — and a partially covered tile gets a partial weight, which can be too subtle to register as motion at all once amplitude is applied, especially at Soft strength.
Practical minimum: aim for brush strokes that are at least as wide as a couple of grid tiles — roughly 40–50px on the 480px working canvas — so your paint reliably crosses full tile boundaries and gives the mesh clearly weighted zones to animate. Going bigger than that is fine and often better for anything meant to read clearly (ponytails, ribbons); going much smaller mostly risks strokes that don't show up in the final motion at all. If you want a very fine, delicate accent to wobble (a single loose hair strand, a small charm), compensate with a slightly bigger brush than feels necessary and let Strength do the "delicate" work instead — Soft strength with a properly-tile-sized stroke reads as more delicate than a stroke so thin it barely registers.
---
Fill All vs targeted painting
Wobble Maker gives you the option to fill the whole canvas with motion weight in one action instead of manually painting. It's tempting because it's fast, but it produces a very different look than targeted painting: every tile in the image gets weighted, so the entire image sways as one loosely-connected surface. That's a legitimate style choice — it can look good on a simple, mostly-flat illustration where "the whole thing gently sways" is the intended effect (think a chibi sticker or a simple icon-style piece).
It's the wrong choice for anything with a clear "should move" vs "should stay put" distinction — a character with a static face and torso but dynamic hair, for instance. Filling the whole canvas there means the face and body get the same motion treatment as the hair, and a wobbling face reads as broken rather than alive (see the companion troubleshooting post for exactly why large uniform areas tend to visibly tear at higher strength).
Use Fill All for: simple, flat, whole-piece pieces where uniform gentle motion is the point — stickers, icons, small props. Use targeted painting for: anything with a character face, solid body, or any part that should read as anchored while another part moves.
---
Why flat line art wobbles more convincingly than photo-realistic images
The displacement math doesn't know what it's moving — it just shifts blocks of pixels by a computed offset each frame. What determines whether that shift looks like "springy motion" versus "visual noise" is how much the image relies on precise pixel-level detail to read correctly.
Flat, clean line art — solid color fills, bold consistent outlines, minimal gradient — tolerates a 20px tile shifting by 5–18px just fine, because the content inside that tile is simple: mostly one color, one clean line direction. The eye doesn't need pixel-perfect alignment to still read "hair strand" after it's been nudged.
Photo-realistic or heavily detailed images pack much more information into the same 20px tile — texture, gradient, fine shadow work — and displacing that tile disrupts all of that detail at once, which reads as smearing or noise rather than a clean shape moving. The same is true for source images that already have motion blur or very fine linework baked in, since there's less solid pixel information per tile for the displacement to carry along cleanly.
Practical takeaway: if you have a choice between a clean-lined manga/anime style export and a photo-realistic render for something you plan to wobble, pick the flat style — it's not just an aesthetic preference, it genuinely holds together better under the tool's displacement math.
---
Always preview before you export
Wobble Maker's preview renders the exact same sine-wave displacement loop you'd get in the exported GIF — it's not a lower-fidelity approximation. Because it's instant and runs locally, there's no reason to export blind. Paint a mask, toggle preview, watch two or three full loops, and adjust brush coverage or strength based on what you actually see before committing to an export.
This matters most when you're layering multiple painted regions with different strengths (say, Soft on the main hair mass and Strong on a single dangling accessory) — the way those interact is much easier to judge by eye in a live loop than to predict in advance. A wasted export costs you nothing but a few seconds, but checking preview first saves you the back-and-forth of re-exporting repeatedly to find the right combination.
---
Summary
- Source resolution above ~1000px doesn't buy you anything — the grid is 24×24 regardless, and 480px is plenty for it.
- Keep brush strokes at least tile-sized (~40–50px on the working canvas) so paint reliably registers as motion weight.
- Use Fill All only for simple, flat, whole-piece art; use targeted painting whenever a face or static body needs to stay still.
- Flat, clean-lined art wobbles more convincingly than photo-realistic or already-blurry source images.
- Preview before you export — it's the exact same render, instant, and free.
Keep Reading
Ready to create your own manga?
Start free — no credit card required. 10 AI generations per month.
Start Creating ⚡Head-to-head comparisons
Related guides
AI GIF Maker for Twitch Emotes — Complete Guide
How to design animated anime-style Twitch emotes with an AI GIF maker — subscriber tier sizing, read…
How to Make a Looping Anime GIF for X/Twitter Profile Headers
A practical guide to creating AI anime GIF loops sized and paced for X/Twitter profile headers, with…
Best AI Anime GIFs for Discord Emotes — Sizing & Style Guide
Learn how to make AI anime GIFs that work as Discord emotes and stickers, with sizing rules, style t…
How to Turn a Novel or Fanfic Into an AI Manga
A practical workflow for adapting a novel, web novel, or fanfic into a manga using AI art — what to …