In-Canvas Render

Overview#

In-canvas render turns the main viewport into a progressive ray-traced preview. When you stop moving the camera, the image starts noisy and gradually refines. It lets you see the real effect of appearances, environment and camera changes without launching a full render job.

It is a preview tool with an output option. It is not the same as the Render command. Autodesk's tutorial describes it as real-time ray tracing on your computer's CPU that starts noisy and clears as it iterates. Think of it as ideal for decisions, limited as a deliverable. The Capture Image command saves what the canvas currently shows.

Where to find it#

• Render workspace > In-Canvas Render panel: In-Canvas Render, In-Canvas Render Settings and Capture Image. Autodesk's help also words the first step as Render > In-canvas Render.
• Switching to the Render workspace does not by itself start in-canvas rendering. Toggle it on explicitly.
• Quality controls live in In-Canvas Render Settings, which has a Fast tab and an Advanced tab.

Step by step#

Turn it on#

1. Open the Render workspace.
2. Click In-Canvas Render in the toolbar to switch it on. The canvas begins to refine.
3. Wait a few seconds for the noise to settle. Large scenes with glass need longer.

Adjust settings#

1. Click In-Canvas Render Settings.
2. On the Fast tab the materials and lighting are simplified; the time and iteration count show at the bottom right of the window. The Limit Resolution slider trades sharpness for speed (20 percent is blurry, 100 percent is sharp).
3. On the Advanced tab, drag the quality slider; rendering stops when it reaches the level you set (the slider runs from below Excellent up to Infinite, with Final along the way).
4. Close the dialog.

Capture an image#

1. Frame the model using the camera page techniques.
2. Wait until the sample count (shown in the on-canvas status or by visual noise) is acceptable.
3. Click Capture Image.
4. In the Capture Image dialog set the image size and resolution, enable or disable Transparent Background and Anti-aliasing, click OK, then name the file and choose PNG, JPG or TIF.

Switch it off#

1. Click the In-Canvas Render toggle again. This is wise before heavy modelling because rendering consumes resources.

Options reference#

Tips and pitfalls#

Tip: Use in-canvas rendering as your look-development tool. Change a roughness value and see the highlight spread immediately. This is much faster than repeated local or cloud renders.

Tip: When tuning an environment, hold the camera still and move only the Rotation slider. Because the geometry is unchanged, refinement resumes quickly.

Warning: Every time the model or camera moves, the calculation restarts. Do not judge noise level while you are still moving things.

Warning: In-canvas render depends heavily on your hardware. On integrated graphics or laptops, it may be sluggish or may reduce modelling performance. Turn it off for modelling.

Warning: The capture is only as clean as the preview. A capture taken early shows grain. Wait for the sampling to settle, or use a local or cloud render for a final image.

Tip: Geometry changes in the Design workspace do not update the render until you return to the Render workspace. If a part does not appear, check that the component is visible and that you are viewing the correct workspace state.

Warning: The in-canvas image can differ from a final render because of different sampling settings, particularly for glass and subsurface effects. Treat it as a preview.

Additional notes:

• A system with a dedicated GPU provides the best experience. Fusion's help lists graphics requirements; consult them for your version.
• Large assemblies increase memory use. Use Display Settings to hide components that do not appear in the shot, or create a simplified design for rendering.
• If the canvas goes black or shows artifacts after enabling, update the graphics driver and toggle the render off and on.

Note: not confirmed in Autodesk's documentation: Personal-use differences for in-canvas or cloud rendering.

• Autodesk's Animation help does not describe ray-traced output; see the animation pages.
• The preview respects section analysis and hidden bodies. That makes it easy to produce cutaway shots.

Choosing between capture and render#

Speed tricks#

• Hide small hardware (fasteners) when they are not visible.
• Lower the Limit Resolution slider while exploring, then raise it for capture.
• Keep the view still while waiting for convergence; any view change restarts the render.
• Close the browser tree and other panels to give the canvas more room only if your resolution depends on window size.

Common symptoms#

Worked example: how long should you wait?#

Progressive rendering converges roughly with the square root of the number of samples: to halve the visible noise you need about four times as many samples. In practice the first ten seconds remove most of the grain, the next minute removes about half of what remains, and the last bit is very slow. A useful working rule is to judge at three checkpoints.

Suppose you are capturing a 1920 x 1080 image of a product on a laptop with a mid-range GPU and it still shows grain after 90 seconds. Rather than waiting five more minutes, check the cause: a scene with several transparent bodies or small bright emissive parts converges several times slower than an opaque scene, and the honest fix is a final render with a higher quality preset (see the local versus cloud page). Capture Image is the right tool for most review images, not for every image.

Resolution arithmetic#

Cost scales with pixel count, not with the width alone. 1280 x 720 is 0.92 megapixels, 1920 x 1080 is 2.07 megapixels, and 3840 x 2160 is 8.29 megapixels, exactly four times the pixels of 1080p. If the preview reaches an acceptable grain level at 1080p in 60 seconds, expect something like four minutes for the same grain level at 4K, because each pixel needs its own samples. Use that to decide: explore at a small canvas, raise the resolution only for the final capture, and wait proportionally longer.

The aspect ratio is a separate setting from the resolution. The API documents the Scene Settings aspect ratio as a ratio only (4:3 is the same as 20:15) that lets in-canvas render use a different shape than the window; the in-canvas resolution comes from the screen, and the capture dialog sets the image size.

Edge cases and failure modes#

• Preview looks darker or lighter than the ordinary canvas: the canvas uses fast display effects, while the render uses physically based lighting. Judge brightness only in the render, and adjust Brightness and Exposure in Scene Settings rather than changing appearance colors.
• Part appears in the canvas but not in the capture: it is probably hidden through a Browser light bulb in a different component state, or the capture was taken before sampling restarted after a visibility change. Toggle the render off and on once and recapture.
• Section analysis cut is rendered, but the cut face is a flat color: a section is a display cut, and the capped face takes a uniform color rather than the body's appearance. For a presentable cutaway, model the cut with a real Split Body or a Combine > Cut and apply an appearance to the new face.
• The interior of an enclosure is black: the environment cannot reach enclosed volumes. Open the shell (hide a lid, or cut a window) or add a scene light. Autodesk documents Point, Spot and Photometric lights (Render > Scene lighting); place one inside the enclosure.
• Preview freezes after a driver update: switch the toggle off, go to the Design workspace and back, and restart Fusion if needed. Keep GPU drivers current; most reports of canvas artifacts trace back to old drivers.
• Capture is blurred after upscaling: if you captured at canvas size and enlarged it in an image editor you only enlarged pixels. Set the capture size before you capture.

Workflow advice#

Treat the live preview as three separate passes rather than one continuous session. First pass: materials only, with a neutral environment, so you tune roughness and color without lighting distractions. Second pass: lighting, rotating the environment while the camera is locked. Third pass: camera, composing and saving a Named View. Doing them in that order avoids the trap of tweaking a material to fix a highlight that is really caused by environment rotation.

Keep a short log of values as you go: environment name, rotation, brightness, exposure and the roughness values you changed. When the design changes next week you will want to reproduce the shot exactly, and the captured image files do not record those settings.

Finally, remember the preview runs on your own machine. It does not consume cloud credits and works offline, which makes it the best tool when you are travelling or when credits are low.

Practice exercise#

1. Open a design with at least two appearances (one metal, one plastic) and a ground plane. Save a Named View called "Preview".
2. Enable In-Canvas Render. After 10 seconds, note by eye whether the metal reads as metal.
3. Change the plastic roughness from 0.1 to 0.5 and observe how the highlight spreads. Then set it to 0.3 as a compromise.
4. Hold the camera still and rotate the environment in three steps of 45 degrees, noting which angle gives a single long highlight.
5. Set the Capture Image size to 1280 x 720 and time how long the image takes to reach acceptable grain. Repeat at 1920 x 1080 and compare against the 2.25 times pixel increase predicted by the arithmetic above.
6. Capture a 1920 x 1080 PNG, open it at 100 percent zoom and list two defects (grain, jagged edges, blurred decal text).
7. Expected result: a clean enough capture, two timings that roughly follow the pixel ratio, and a short defect list that tells you whether a final render is justified.

Related#

• Scene Settings and Environment
• Local vs Cloud Render
• Tips for Photoreal Output