Vision Canvas: Blender’s UI, On Your Terms

Build a Blender control panel that answers to you: targeted properties, camera widgets, Python buttons, reusable animation and asset favourites.
Blender’s dark-themed interface shows a low-poly orange humanoid model standing in the central 3D viewport on a gray grid floor, with timelines and object lists spread across the bottom and right panels. A colorful pose-picking panel fills the left side, and a webcam overlay in the lower right shows a man in a cap speaking into a microphone.

For those who don’t know the tool: Vision Canvas is Maciej Gliwa’s configurable control editor, built into a custom Blender 5.2 LTS application. It combines scene selection, property controls, Python actions and pose or animation recall in a workspace the artist arranges. The current beta is a portable Windows 64-bit build.

The wish is simple: finally, Blender’s UI behaves the way I want it to. The controls I use every five minutes live together. The buttons stay where I put them. Adjusting a camera does not require a small pilgrimage through the Properties editor, with a detour to ask the Outliner for directions. Vision Canvas gives that wish a dedicated editor inside a custom Blender build.

Its useful trick is letting you design a working surface around a task. A rig picker, a lighting panel or a camera console can expose the controls that matter in that context. The attraction is less interface archaeology and more deliberate access to the scene. Blender keeps its considerable machinery; you get to rearrange the dashboard.

A button should know whom it is bossing around

The property-button tutorial explains the mechanics. In Canvas Edit mode, create a Prop element, then configure it in the N-panel’s Property Widget settings. Eyedroppers identify the target object and property. Targets include a specific object, selected objects, the active object, the active camera and all objects of a chosen type. A fixed-object binding keeps that object’s control available when something else is selected.

For groups, Match assigns the same value to every target; Relative applies the demonstrated offset. Gliwa’s light-power example starts at 100, 200 and 300. Raising the active light from 100 to 150 produces 150, 250 and 350. That is an additive change, not proportional scaling. Match, by comparison, can set every light to 5,000. The lights have received a memo, and the memo has numbers.

The camera gets its own little bridge

The available control shapes include sliders, wheels, joysticks, trackpads and switches. The developer lists light strength, camera settings, modifier amounts, shader inputs and rig controls among their uses. Those are interfaces to properties and actions; a more agreeable knob does not secretly introduce a new rendering algorithm.

In the camera demonstration, focal length and depth of field share the canvas with movement controls. A wheel handles boom, a trackpad operates pan and tilt, a joystick allows finer movement, and another trackpad combines tracking and booming. A dolly control moves towards the 3D cursor, while buttons switch cameras. A Python action focuses the active camera on that cursor.

That gives layout work a coherent control surface: the lens, framing and focus operations used together occupy the same small patch of screen. It is the appeal of a cockpit where someone has finally asked the pilot which switches belong within reach. The video illustrates the arrangement; it provides no measured productivity benchmark.

Python, wearing a respectable button

The Python-button tutorial supplies the practical recipe. Copy code to the clipboard and paste it onto the canvas in Edit mode: Canvas creates a button with that code attached. Switch to Pick mode to execute it. Alternatively, create an empty Python button, then add its code. The context-menu command Edit Code in Editor lets you revise the script, save and close; the button can also be renamed.

Your poses can bring their animation

Vision Canvas includes an additional pose and animation library, separate from Blender’s Asset Browser facilities. In the animation demonstration, Gliwa recalls a stored walk cycle onto his character, then adjusts the pose across selected keyframes while playback continues. The example changes the body height throughout that selection instead of altering only the current pose.

For that operation, the artist works through the character controls rather than first locating and offsetting the relevant curves in the Graph Editor. The selected keyframes define the scope; the demo does not establish automatic retargeting between unrelated rigs. A reusable walk is welcome. A universal translator for every skeleton in the building would require rather more evidence, and possibly a diplomatic passport.

Favourites without reorganising the entire cupboard

The asset workflow demonstration addresses another familiar nuisance: repeatedly finding the handful of assets needed for one sequence inside a much larger library. Items can be dragged from the Asset Browser onto the canvas, arranged as a working set and then dragged into the 3D viewport. Gliwa describes this as keeping favourites without changing the underlying library organisation, including when a team shares that library over a network.

The demonstration also shows movable, collapsible frames and text labels for grouping items. This is a useful distinction for production: a personal or sequence-specific selection can sit above the shared collection without demanding that everyone adopt your filing philosophy. The robots you need today can stand together; the master library need not undergo municipal rezoning.

Elsewhere, selection buttons address objects, collections or bones. The documentation index distinguishes individual and group selection, collection and visibility controls, and tools for distributing and aligning buttons. Images, labels, frames and multiple tabs provide organisation, and the store describes canvases that can be saved, shared and updated. The supplied Orange/Blue theme supplies the livery. Sensible grouping still supplies the navigation.

The orange icon is not the whole deployment plan

The installation video shows the actual delivery path: download the ZIP, extract its folder and run the included blender.exe. You are launching a separate Blender application, rather than adding an extension to the executable already approved for your pipeline. Keep that distinction visible in workstation shortcuts and production instructions. Identical orange icons are perfectly capable of concealing two different answers to “which build are you running?”

The listed base is Blender 5.2 LTS. Blender’s LTS programme provides critical fixes over two years, but that upstream commitment does not establish how this third-party beta receives fixes or support. Source is included with the download; the public listing does not settle the precise licence or future update entitlement. Those are questions for the developer, not promises that can be extracted from the letters LTS with sufficient optimism.

https://maciejgliwa.gumroad.com/l/wkeygd

Availability

The official store currently lists the Windows 64-bit beta at US$39.99, with an EARLYBIRD offer from US$24.99 and a US$15 discount applied at checkout. No promotion end date is shown. The video documentation remains a work in progress; some sections are explicitly marked as coming soon.

ProductVision Canvas
DeveloperMaciej Gliwa
Application baseCustom Blender 5.2 LTS; beta
PlatformWindows 64-bit
DeliveryPortable ZIP download; source included
Interface themeVISION Orange/Blue, pre-applied
Listed priceUS$39.99; EARLYBIRD offer from US$24.99
DocumentationOfficial video documentation