What Can T Fly Actually Is
Can T Fly is a lightweight browser-based physics toy where you design and launch paper airplanes using a particle-style rigid body engine. You draw wings, assign mass distribution, hit start, and watch the sim run. That's it. No account, no download, no subscription. The whole thing lives at cantfly.io if you want to just jump straight into it. You don't download anything. It runs directly in Chrome or Firefox. If you're on mobile, it works but the drawing controls are clunky and I'd skip it unless you have to. Desktop is where this thing actually feels usable. The page loads in roughly 3 seconds on a normal connection. No installer, no permissions, no "please disable your ad blocker." Just open the tab. I started using Can T Fly because I was prototyping simple aerodynamic shapes for a game project I had in my backlog. The idea was quick iteration on wing profiles without firing up Unity or any heavy editor. Here's how the workflow actually goes once you sit down with it.
First, you draw a shape. The pen tool snaps to a grid but you can turn snapping off if you want smooth curves. Click to place vertices, double-click to finish the stroke. Make sure your shape is closed. Open polygons won't simulate and will just hang on the loading spinner while the engine throws a silent error. I learned that the hard way on my first attempt. After closing the shape, you assign mass points. Drag the small orange dots that appear on your outline. Move them inward to concentrate weight toward the center, outward for edge-heavy designs. This is where most people make mistakes. They distribute mass evenly and wonder why their plane stalls immediately. A paper airplane needs forward bias, not centered balance. Then you pick a launch angle and force. The default is 45 degrees at medium power. That's fine for testing but it's not realistic for any actual flight distance. I bumped it to 30 degrees and max force. The plane stayed airborne longer and the trajectory data made more sense.
The sim runs at about 60 frames per second on a decent machine. You can pause, rewind, or reset at any time. There's a replay slider that lets you scrub back through the flight path, which is genuinely useful when you're trying to figure out why a design failed mid-air.
Get the Full Details
What the engine actually does
Can T Fly uses a simplified 2D rigid body simulation. Each vertex you place becomes a mass point connected by structural constraints. Air resistance is calculated per-frame based on the surface area exposed to the velocity vector. Lift is approximated using a basic coefficient that increases with angle of attack up to a stall threshold, then drops off sharply. It's not wind tunnel grade accuracy. It's "good enough to see if your design is completely broken" accuracy. That distinction matters if you're planning to use the output for anything serious. The lift model treats the entire shape as a single flat plate. That means winglets, dihedral angles, and cambered surfaces don't behave any differently than a flat rectangle. I found this out after spending about twenty minutes tweaking a complex design only to realize the engine was ignoring every refinement I'd added. The workaround was to simplify the shape to its core planform and adjust mass distribution instead. That actually moved the needle.
Edge cases that will trip you up
Here's one that wasted a solid hour of my time. If your shape has intersecting edges or self-overlapping vertices, the simulation will crash on launch. Not gracefully. The browser tab will just freeze and you'll have to force-close it. I discovered this when I was extending a wing tip and accidentally crossed back over an existing line segment. The fix is to zoom in and check for overlaps before launching. I now redraw any shape that looks too complex rather than trying to edit it in place. Another issue: extreme mass configurations. If you pile too much weight into a single point, the solver can diverge and produce erratic behavior. The plane will teleport across the screen or oscillate violently. Keeping mass points spread out with no single point exceeding roughly 40% of total shape mass prevents this. I use a rough visual rule of thumb. If one orange dot looks like it's pulling the center of gravity into its own corner, I back it off. There's also a limit on how many vertices you can place. The exact number varies by browser but it's somewhere around 50 to 60 points. Beyond that, performance degrades noticeably and the sim starts dropping frames. I've hit this ceiling when trying to draw highly detailed wing profiles. The solution is to approximate complex curves with fewer, strategically placed vertices. You lose some precision but the flight behavior stays stable.
Counter-intuitive things that actually work
Most people try to make their planes look like real aircraft. That's the wrong approach here. The simulation rewards simplicity. A broad, slightly angled rectangle with forward mass will outperform a detailed fighter jet shape every time. The engine's lift model favors large flat surfaces at moderate angles. Thin swept wings that look cool on paper generate almost no lift in this sim because the surface area calculation is too small relative to the weight. Another thing: adding tail surfaces doesn't always help stability. In some cases, a small rear fin or horizontal stabilizer actually makes the plane more tippy. The reason is that the tail adds drag at the wrong point in the flight curve and shifts the center of pressure backward faster than the mass distribution can compensate. I stopped treating tails as mandatory and started testing with and without them for each design. Sometimes the plain wing flew farther.
Known limitations
Can T Fly is not an engineering tool. It's a toy with a physics backend. If you need actual aerodynamic data, use XFLR5 or OpenVSP instead. The sim doesn't model turbulence, thermal currents, or three-dimensional effects like wingtip vortices. It also doesn't save designs natively. If your browser crashes, your work is gone. I keep a folder of screenshots of my better designs and redraw them when needed. It takes about five minutes per shape. There's no multiplayer or sharing system built in. You can't send a link to a friend and have them test your plane. The community workaround is posting screenshots or screen recordings on Reddit or Discord. It works but it's slow. A built-in share feature would have saved a lot of back-and-forth. The mobile experience is the weakest part. Touch controls for precise vertex placement are frustrating. If you primarily use a phone, you'll want to stick to simple geometric shapes and accept that the drawing fidelity will be lower. Desktop is the way to go.
Is Can T Fly worth your time
If you want a quick way to prototype paper airplane designs and see basic flight behavior without installing anything, yes. It's fast, free, and the learning curve is about ten minutes. If you're looking for accurate aerodynamic analysis or professional design output, look elsewhere. The sim is intentionally simplified and that's by design. It's meant for play, not precision. I still use it when I need to brainstorm shape ideas quickly or when I want to show someone the basic concept of mass distribution and lift without pulling up a full CAD package. For that purpose, it does the job. Not perfectly. But adequately.
