Goxel

Voxel art sits in an odd corner of 3D. It takes the grammar of pixel art (a grid, a color, one block at a time) and adds a third axis. That makes it approachable for illustrators and indie developers who would never open a polygon modeler. Voxel tools have mostly lived on desktops, where a mouse can pick a single cube out of thousands. Goxel, from developer Guillaume Chereau, is a serious attempt to move that workflow onto phones and tablets.

Where it comes from matters. Chereau also builds the desktop Goxel, a free, open-source editor for Windows, macOS and Linux. The iOS/iPadOS and Android releases are the paid branch of the same project. The mobile app therefore arrives with a feature set inherited from desktop and export targets that match what game developers and 3D-printing hobbyists already use. Mobile buyers pay for portability and a touch interface. They also support, indirectly, an open-source project whose desktop users pay nothing.

That arrangement shapes where Goxel fits. On mobile it works as a sketchbook and workbench for people already invested in voxel pipelines. It is not a casual building toy. The natural audience includes MagicaVoxel and Qubicle users, indie developers who need glTF2 or OBJ assets, and anyone who wants to block out a model on an iPad and finish it elsewhere.

Public reception is mixed. The listed store rating is 3.8 out of 5 from 50 ratings. Praise centers on its clean look and ease of use; complaints focus on saving, exporting and selection. A long stretch without updates once had users worried the app was abandoned. A recent update brought the mobile build in line with desktop and puts that worry to rest, at least for now.

• Sparse-matrix scenes: Scenes are stored as sparse matrices, so there is no fixed limit on scene size and no bounding box to outgrow.
• 24-bit RGB color: The full RGB range is available for every voxel.
• Unlimited undo: The undo history has no ceiling, so mistakes stay cheap to reverse.
• Layers with per-layer materials: Scenes split into separate models that can be edited, hidden or combined independently. Each layer can carry its own material, including transparent and emissive (glowing) ones.
• Three render modes: Classic cubes, smooth marching-cubes surfaces, and a physically based path tracer for realistic final images.
• Broad export: MagicaVoxel, Qubicle, glTF2, OBJ, PLY and Build Engine formats, covering games, 3D printing and other 3D tools.
• Desktop-parity update: A better selection tool, a reorganized interface with movable panels, and improved glTF export.

The Building Model

The premise is direct: place colored cubes on a grid until a shape emerges. The pixel-art comparison holds because the mental model is the same. Artists think in units instead of vertices, edge loops or normals. That lowers the barrier for 2D artists and game developers who already think in grids, and a touch screen suits this kind of discrete, tap-to-place construction better than freeform mesh sculpting.

Two architectural decisions widen the canvas. Sparse-matrix storage records only the voxels that exist, so a prop, a building or a full diorama can grow in any direction without hitting a wall. 24-bit RGB color lets artists pick any shade instead of rationing a short palette. Both choices favor exploration over constraint, which is the right bias for a device people reach for when an idea strikes.

Unlimited undo completes the case for experimentation. In voxel work, one misjudged fill can wreck a section of the model, and an undo stack with no ceiling makes those mistakes free. The weakness is how hard undo is to reach. A two-finger-tap undo, the gesture tablet art apps have trained people to expect, is absent. On a touch device, how easy undo is to reach shapes how boldly people work. A missing shortcut taxes exactly the experimentation the unlimited stack is meant to encourage.

Layers and Materials

Layers do two jobs. For organization, they split a scene into separate models (a character, a prop, a base) that can be edited, hidden or combined on their own. This matters most once a model gets dense enough that exterior walls hide interior detail. For expression, each layer carries its own material, including transparent and emissive options.

Binding material to layer is a smart choice. Glass, water and glowing signage each get a clean home, and the structure maps neatly onto separate meshes and materials when assets move into a game engine. It is a small decision that reflects an editor designed by someone thinking about where the model goes next.

Three Ways to See a Model

Each viewing mode serves a distinct stage of the work:

• Classic cubes are the working view, with every voxel visible and every edge legible.
• Marching cubes smooth the grid into continuous surfaces, which helps when judging the silhouette of organic forms or previewing how a model might look once printed.
• The path tracer handles presentation, producing physically based lighting for finished images without a separate renderer.

That last mode carries particular weight on mobile. A shareable, realistically lit render straight from a tablet completes a job that would otherwise need a desktop.

The Pipeline Is the Product

Export is where Goxel earns a place in real workflows. MagicaVoxel and Qubicle cover voxel-native tools. glTF2, OBJ and PLY reach general 3D software, game engines and 3D printing. Build Engine support adds a more specialized target. The recent update’s improved glTF export is the right priority for an app whose value depends on handing models to other tools.

Breadth means little without reliability, and here the public record is uncomfortable. Saving and OBJ export are recurring concerns, and the workflow needs to make confidence easy. A creation app can survive clunky menus. It cannot survive users doubting that their work exists. Requests for a clearer new-file and save workflow point the same way: file handling needs work even where nothing outright breaks.

Two more gaps limit production use. Image import is absent, which closes off tracing reference art and building from imported images. Non-square voxels are also absent, which production artists need when uniform cubes don’t match a target art style or engine grid. Neither omission matters for sketching. Both matter for anyone trying to ship assets.

Selection and Manipulation

Box selection and moving objects are central to any edit beyond placing single cubes. Repositioning a limb, duplicating a window or shifting a wall all depend on them. The recent update’s better selection tool aims directly at this weakness. Public complaints don’t say which build they describe, so whether the update resolves those concerns remains open.

The movable panels are a sound call for touch. On a tablet, every panel competes with the model for screen space, and letting users push tools out of the way matters more there than it does on a desktop monitor.

Goxel’s visual identity lives mostly in what it renders, not in its menus. The interface is described as clean, and the design rightly keeps the model in the foreground. The reorganized layout with movable panels shows an interface still adapting from desktop conventions to touch. The direction is right, though how far the adaptation has come is hard to judge from the available evidence.

The rendering range gives the app a high visual ceiling for a mobile editor. Blocky cubes serve the work, marching cubes serve form, and the path tracer serves the finished piece, with transparent and emissive materials adding glass and glow.

On performance, sparse-matrix storage is the right foundation, since memory tracks what has been built rather than an empty volume. The path tracer is the heavy component by design. Physically based rendering is expensive on any hardware, so on phones and tablets it fits best as a final pass, not a live working view. The cube view should be the lightest of the three modes.

• Pricing: Both the iOS/iPadOS and Android versions are paid. The desktop version for Windows, macOS and Linux is free and open source. That caps the mobile app’s value: anyone with a computer can get the same project’s editor for nothing. The mobile price buys portability and a touch interface.
• Input: Touch replaces the precision of a mouse cursor. Picking one voxel out of a dense model with a fingertip is harder than clicking it, which makes selection reliability and undo access more important on mobile than on desktop. The app makes the most sense on iPad and Android tablets. On phones, a panel-heavy interface competes for far less space.
• Parity: The recent update aligned the mobile app with the desktop version. The feature list is presented as shared between iOS and Android, and nothing in the published feature set separates the two.
• Reception: The listed store rating is 3.8 out of 5 from 50 ratings.
• Cross-tool workflow: The export list is the bridge. Models started on a tablet can move into MagicaVoxel, Qubicle or glTF2-based tools on any platform, provided the export itself succeeds.

Goxel brings the right ideas to a tablet: a grid without walls, color without a palette, undo without a ceiling, materials tied to layers, and a renderer that can finish the job on the device. Its export list connects it to the tools where voxel models get used. On paper, it makes a case for doing serious voxel work away from a desk.

Trust holds it back. Saving and export are the weak points, and selection concerns hit the next most important part of the workflow. The recent parity update targets the right problems and answers the abandonment worry, but the public reception has not yet caught up with it.

We can recommend Goxel only to voxel artists and indie developers who want a portable front end for an existing pipeline and who export and back up often. Anyone who needs production guarantees should start with the free desktop version, which offers the same project’s feature set at no cost. We rate it Mixed.