The making of Hammerson House resident accommodation
The making of Hammerson House resident accommodation
Nightingale Hammerson is a multi-million-pound redevelopment of Hammerson House, a residential care centre of 116 bedrooms for the Jewish community. The bedrooms will be part of households, each with dedicated facilities and comfortable spaces to make residents feel at home. https://nightingalehammerson.org
The interior design team at P+HS has worked consistently with Nightingale over several months producing in depth mood boards for various design schemes for the resident bedrooms. The comfort of the resident was always at the forefront of the scheme proposals, creating a luxurious room that they would call home.
The residential aspect of the project consists of 6 households set out over 4 storeys and it was important that the households had their own identity. The designers worked closely with the Hammerson team to establish how best to achieve this.
Coupled with the wayfinding and arts strategies discussed early in the project it was felt that the use of colour would be the most effective and user friendly option. The choice of colours in fabrics and wall coverings were then chosen to build on the character of the households. Three colour themes where chosen by the Hammerson Team and the process of creating mood boards began. These included key images and words, fabrics and furniture choices that the design team felt best conveyed the essence of the Nightingale Hammerson brand.
Nightingale contracted P+HS to produce CG imagery to help with marketing the bedrooms to potential residents. As of any standard marketing CG visuals, the completed imagery needed to engage the potential resident and allow them to emotionally invest in the bedroom and household.
Workflow: Bringing the idea to life
A successful CG image workflow consists of six key areas. Each area is equally as important as the other in contributing to a photo realistic image. Within each area is a list of criteria which needs to be met to allow successful progression onto the other stages. However workflow of the stages isn't always linear, and it is often the case that changes in one of the key stages effects another, so further work is required. The visual team worked closely alongside the interior designers throughout all key areas to ensure that the final result portrayed the desired ambience.
Modelling: Developing the base for photo realism
P+HS had a working Revit model for Hammerson House as the project was being delivered using BIM. A shell existed for the bedrooms so it was simple to extract the data and import it for further development into 3D Studio Max.
Each piece of furniture within the room was carefully examined and modelled in detail to real life form. Many of the furniture providers had BIM information accessible online to download. Time was spent developing these models into a more detailed representation of the shape. Ripples and creases in the fabric were modelled into the furniture. It was easier to apply this level of detail in the structure of the model, rather than to bake it in to textures. It proved quicker to create and less complicated than the process of developing textures to suit each individual furniture model would have been.
Once the main pieces of furniture had been created, it was vital to introduce placement objects to fill space, add further detail and bring life into the scene. The model needed to emphasise that the bedroom was a personal space, but more importantly a home.
Lighting: Setting the mood
The bedroom was designed to be a calm, relaxing space with a warm welcoming homely ambience. The large window near the bed offers views out to beautiful greenery and the opportunity to steal a vast amount of sun light.
In 3D Studio max the bedroom scene was lit using VRAY. This is the engine we consistently use when creating high resolution detail renders, it offers immense capabilities. In the bedroom scene, the primary light source comes directly from added sun light. The sun casts light through the bedroom window creating soft highlights on the carpet and over the bed. The bed is then lit by a secondary light from the bedside lamp, generating an ambient glow. A HDRI (High dynamic range image) map was incorporated into a dome light placed in the scene to generate realistic light and shadows. Initially the lighting is setup on a white model. Only once the lighting is complete, we would look to introduce materials.
Materials: Embrace photo realism
Good texturing is a powerful technique and if successful can emotively involve the viewer, helping them to understand the story that a still image is trying to tell. Materials set the colourway and ambience, ultimately they are responsible for an image/animation being successful. Even if the scene is stunningly lit and models realistically detailed, if the materials are not correct the sense of photo realism is lost. Portraying materials precisely involves viewing and figuring out the property values of each real material sample. Every material holds values for light being reflected, smoothness, glossiness and roughness. Those values need to be examined and re created in 3D to appropriately mimic it's true form.
Careful consideration and discussion had been taken over the choice of fabrics to use within the resident accommodation at Hammerson House, from the heavy draped curtains, lightness of the voile and softness and velvet feel of the chairs. As a whole, the careful chosen range of materials was to create a luxurious high end finished environment.
Texturing the furniture was a complex process. All fabric materials were made to be procedural, meaning that they were seamlessly repeated and could be applied to any object. We used Photoshop to create the seamless fabric patterns which was not always a quick task as the image needs to be repeatable but it doesn't want to look over-repeated when in use as it distracts from the concept of realism. Each 3d furniture model did require texture 'mapping'. Mapping enables us to control the way in which model faces lay over materials. It is a method which technically allows a pattern to be pulled around a 3d surface to look as it should. This was certainly applicable to areas of curvature over the arms and back of the chairs.
The viewing angle
Cameras in 3D work on the same principle as real-world cameras, so it is vital to have some knowledge on camera functions such as aperture size, shutter speed, ISO and F-stop. Acting as the photographer you have the power in directing the viewer towards what is important, what it is you are trying to tell them.
Choosing a good angle can make the difference between selling the idea or not. In this case, the size of the bedrooms was the main message we wanted to get across to the viewer. As the whole of the bedroom scene was built, experimenting with various camera angles was simple. Views were set up and tested until an angle was found to best framed the room.
The two pictures below are taken from the same 3D model but through different camera angles. The primary picture doesn't depict the message; the rooms are outstandingly expansive spaces with a carefully thought out plan. Repositioning the camera facing towards the bed creates a totally different feel to the bedroom and generates the desired message.
Switching from still to animation
You would imagine once the 3D scene is complete and the stills are rendered that setting up cameras and rendering the animation would be a very simple step as everything has been done. Unfortunately that is not always the case!
Static 3D images are one independent still, which means that more time can be dedicated to rendering one single image, to get it as life like as possible and allow many hours of rendering if required. A render can be set to proceed when leaving the office on an evening and on your return fourteen hours later it would be complete, with no issues ideally. Even if errors occur, as it is a single image routinely it would be taken into Photoshop to amend. However an animation involves rendering hundreds of images, normally against a deadline so immediately you have a restriction against the amount of time you can dedicate to rendering one single image. The animation of the bedrooms needed to be approximately 60 seconds long. In each second of animation are 24 separate images. This means that approximately 1440 images need to be rendered. In this case rendering a still which took 3 hours has to be reduced to rendering the same still in under 30 mins, that is a huge difference and several elements have to be re examined to succeed.
The first step in switching this scene from a still to an animation was altering the render engines used under VRay. Using the engine specified to render the static images caused flickering on materials when running a sequence of images back to back. The engine was switched out to 'Light Cache' as this type of global illumination works efficiently with any light type and produces correct results in corners and around smaller objects, but mainly as it eliminated flickering on image sequences.
A light cache is built on light samples, the amount of samples and size of the samples is defined in the roll out settings. Below are 2 images to show a difference in light cache settings from first initial setup to the final light cache calculation. The third image was an animation test created once light cache settings had been finalised to ensure that there was no light flickering.
Following on from this, object materials needed to be 'downsized' in terms of the information used regarding reflections, displacement and bump mapping. This meant prioritising what was important, as for every piece of information a material held, extra time was needed to render. For some materials this meant deleting complete pieces of information, for others it meant readdressing those areas.
All of this was addressed in order to match the same output as the static renders, but to create in under a third of the time. The two images below show the differences between the single static image and a frame taken from the animation.
From animation to VR
Transferring the completed CG model from 3D Studio Max across to our VR software was a simple process. The process took approximately 1-2 days with some additional time for final light baking of the scene. Light baking is a technique used to store data in materials. It is hard to accomplish realistic lighting in real time, it requests high processing power. By 'baking' the data in materials, it allows most graphic cards to run complex scenes.
Once in VR the user can access the application via a web link and explore the 3D environment either on their desktop computer, tablet or mobile device.
In spite of the fact that the 3D model imports across into the VR software easily, object materials don't transfer so need to be reassigned. This is a straight forward process as all materials have previously been created to suit the model so its only a matter of re linking them in the software material editor.
Additionally as it is a VR scene light probes need to be introduced. Light probes capture 360° panorama of its surrounding and are used to generate reflections, to enhance realism in the scene.