Architectural Design and the Oculus GO : WIP

Project 01
A small art gallery project to determine how viable the Oculus GO is for aiding architectural development work – the building, internal fittings and ground base are completely modelled with ProBuilder in the Unity Editor. Hard/soft landscaping and signage realised in Blender. Posted here in ‘World Building’ rather than ‘AR/VR (XR) Discussion’ as the former is more aligned with my focus.



Image Capture :-
Overview above from the Unity Editor – all subsequent images were grabbed directly in the headset.



Grid :-
0.9m with a 0.3m sub-grid (≈Imperial foot) to aid spatial correlation during walk-throughs.
Lighting :-
Real-time only, no baked Global Illumination or baked Shadows, to facilitate rapid turnaround of concept iterations. 1x Directional Light, 4x Point Lights with cookies to simulate individual Spot-Lighting on 21 paintings, 2 sculptures and 2 signs.
Control :-
Movement is set to a fast walking-pace, around 6km/hour (1.7m/sec) with start/stop via the controller’s trigger and direction via head orientation. Doors are proximity operated.
Performance :-
All screenshots captured in the Oculus GO are stamped with frames-per-second (FPS) values. They vary from around 45 to 72 depending on view frustum complexity. Due to the ‘relaxed’ movement speed the experience was smooth regardless of FPS or rapid head orientation changes. Whilst 90 FPS was considered the ‘minimum’ VR requirement when the Rift and Vive were launched the upcoming Rift S is capped at 80 FPS and the Quest at 72 FPS. These figures are obviously set to accommodate fast-paced action games rather than the more leisurely requirements of building design.
Conclusion :-
The ‘entry level’ Oculus GO is more than adequate for aiding small project development – large scenes could most likely be accommodated by level-loading. Better FPS performance would certainly be achieved by baking lights and reducing soft landscaping complexity but the time penalty for the former is not worth it and minimising the latter would detract from the context. Personally, I prefer the freedom and set-up simplicity of a self-contained wireless VR design aid – so the decision to get an Oculus Quest this spring is a no-brainer. I’ll base my next VR project on the ProBuilder API and procedural generation.
Credits :-
Paintings : Rijks Museum
Sculptures : British Museum

 

So . . . Images not showing but URL's worked fine when creating the thread.
Editing now gives a 'Your content can not be submitted' error.
Any ideas anyone? Maximum file size exceeded perhaps?

Edit:-
Yup - maximum file-size exceeded : 3328 x 2048 = 7.20 MB and 3072 x 2048 = 4.85 MB too large ?!
So reduced to 1664 x 1024 = 2.21 MB and 1536 x 1024 = 1.62 MB

 

Project 02

Overview :-
High-rise apartment building options procedurally generated and iterated on the Oculus GO at run-time. All elements constructed using the ProBuilder API. Maximum allowable number of floors and maximum floor area arbitrarily pre-set : construction terminated whichever is reached first. Building articulated into quadrants on plan – each of which has two full-height orthogonal shear walls to aid structural cantilevering on outward-facing elevations.
Level of Detail :-
The project is structured solely as a procedural test with no constraints regarding a specific site or brief. It is envisioned that the approach would be used at the preliminary design stage, prior to committing to full BIM development. Thus it would augment and run in parallel with conceptual mass studies carried out in Autodesk Revit or similar. Core details : Internal stairs/elevators are omitted as these could be more productively developed in the BIM package. Minimum core size would be set to allow for later integration. The project’s primary purpose is as an ideas generator rather than as a final form finder. Hence the focus is on rapid turnover of a multitude of options from which a few would be chosen for manual adoption in a BIM context.



Image Capture :-
All extracted from videos recorded at run-time in headset
Lighting :-
Real-time only, no baked Global Illumination or baked Shadows, to facilitate rapid turnaround of concept iterations.
1x Directional Light only.
Interface :-
Controller TouchPad –
Divided into five ‘buttons’ as indicated on the image above –
Buttons 2,3,4 could be combined into a single cycle / toggle if further design inputs are required
Controller Trigger –
Start/stop toggle sets movement to a fast walking-pace, 6km/hour (1.7m/sec), in direction of head orientation.

Iterations at 1 : 10 Scale​

Set-out :-
Podium : ‘CreateShapeFromPolygon’ :-

Height varies randomly between 0.9m and 3.6m.
Side slope set to fixed angle of 60° but could be randomly varied.
Extension limits randomly chosen per quadrant per iteration​

Stairs : ‘ShapeGenerator.GenerateStair’ :-

Height varies according to podium. Width randomly selected between 0.9m and Podium Core width – 1.8m
Slope randomly set between 22.5° and 45°. Tread/Riser Dimensions calculated to suit and yield step number.
Four options are generated per design, allowing for choice in a ‘final project'.​

Stringers : ‘CreateShapeFromPolygon’ :-

Calculated separately to give option of full-sided (as per images) or sloped with stair at consistent depth.​

Handrails/Posts : ‘ShapeGenerator.GenerateCylinder’ :-

Height fixed at 0.9m.
Handrail depth scaled to half width.​

Columns : ‘ShapeGenerator.GenerateCylinder’ :-

Height fixed at 3.6m
Diameter fixed at 0.4m
Shape : Circular : 16 sides, smooth, or Square : 4 sides, flat​

Floors : ‘CreateShapeFromPolygon’ :-

Height fixed at 3.6m
Width, depth randomly varied.
Consist of Slab, Walls and Roof in distinct quadrants on plan.
Wall quadrants randomly varied between ‘closed-solid’, ‘open-balcony’ and ‘open-glazed’.
​

Performance :-
Frames-per-second (FPS) values vary from around 45 to 72 depending on view frustum complexity. No perceptible lag when generating new layouts. Whilst the high-rise project has less than half the view frustum poly-count of the art gallery project the ‘live’ nature of the former in terms of layout-change, rotation and scaling precludes using the optimisation techniques available to static structures.
Conclusion :-
The hardware performed as required by limiting the exercise to a conceptual rather than a presentation level. In spite of Oculus GO’s ‘entry-level’ spec it was up to the task at hand. Developing a project in a VR environment has the distinct advantage of being able to appreciate the scale and ‘tactility’ of proposals at an early stage and eliminate the ‘abstraction’ associated with viewing 3D walk-throughs on a conventional 2D screen. The aim is now to revisit the underlying code and generalise the functions in order to provide a stylistically diverse range of options.

 

@gabrielw_unity You’re welcome!

In light of Unity’s ‘democratisation’ mantra I feel that VR needs prioritisation in that regard. Hence my focus on affordable tech. Fatuous terms like ‘gamification’ and ‘serious games’ are now redundant as 3D visualisation and interaction are rapidly becoming the norm across multiple sectors.