Renderpeople Adds Rigged 3D People

A detailed look at Renderpeople’s expansion from ready-to-render scanned people to rigged 3D characters, including topology, skeletons, facial controls and workflow trade-offs.

Archive translation. This article originally appeared in Digital Production 01:2017. Products, pricing, links, software support and availability reflect that period; verify current details before production use.

In brief: Cologne-based Renderpeople expanded its photogrammetry-derived library of ready-to-render people with rigged characters intended for animation and interactive workflows.

This article originally appeared in DP 01:2017.

Last year (in DP O5:2O15), we introduced you to Renderpeople – and now the young company from Cologne has expanded its idea once again.

As a reminder, the range includes “ready-to-render” figures – a type of 3D stock photography that are photogrammetrically scanned and textured and can be integrated into any visualisation ready for use. Anyone who can’t imagine this can find three models to download and try out at renderpeople.com/free-3d-model. If you want to take a look at the history of the company, you can download the article from 2015 for free here: bit.ly/renderpeople2015. This range now includes almost 500 people in various poses, outfits and activities (as of September 2016). But that’s not all: we spoke to Moritz Schneider from Renderpeople to find out what else is going on.

DP: What has happened since last year?

Moritz Schneider: 2016 was a very exciting year. We have doubled our 3D PosedPeople database and would like to double the amount again by the middle of next year so that our database contains more than 1,000 posed people. Furthermore, the models are no longer offered exclusively for V-Ray, but also for the native renderers of Cinema 4D and 3ds Max. Many of our customers also had the problem that they could not load more than approx. 20 figures into a scene due to our large 8K textures. We therefore developed a plug-in for 3ds Max for such large scenes, which reduces all textures to a desired size and links them directly back to the material. Our scanning system has also been properly upgraded so that we now have 196 SLR cameras to produce even better scans. We have also started to work on fully rigged figures.

DP: Rigged figures? Who is that for?

Moritz Schneider: The trend in architectural visualisation is also moving more and more towards films. Even game engines such as the Unreal Engine are increasingly being used for ArchViz in order to present the customer with an even more interactive impression of the building and its surroundings. While green screen shots can still be used for film, there is no such thing for the engines. The perspective, lighting and colour temperature of the people often do not match the surroundings. Rigged people, who walk through the scene and interact with the environment, have the same advantages here as our posed people in the image area. The great effort involved in populating a rendered scene in post-production is therefore completely eliminated. The step from static 3D people to moving or movable scans was therefore only logical.

DP: In which packages are the people currently available?

Moritz Schneider: Our library for 3D rigged people iscurrently still in its infancy. However, we are trying to expand our selection as quickly as possible. Of course, as with our posed people, there is a quantity discount of up to 45%.

DP: To what extent can the individual characters be customised?

Moritz Schneider: For the Rigged People, we have orientated ourselves very much on the advantages of our Posed People. The clothing is a separate geometry to the character. Each item of clothing therefore has its own shader. This makes it very easy to change the colour or properties of the texture of the individual parts. The materials can be adapted and changed in many ways in the 3D programme so that the figure can also be used several times in a scene without it being noticeable. For Maya, we offer scans with a Human IK Rig, while for 3ds Max we offer both Biped and the more modern C.A.T. system. So every user can decide for themselves which system they are most comfortable with.

DP: Rigging bipeds in a chord certainly takes a few attempts: How are the rigs of the renderpeople set up, and where can you still intervene here yourself?

Moritz Schneider: With Biped and C.A.T. for 3ds Max and Human IK for Maya, we offer our customers ready-made control rigs that have long been established on the market and offer great interfaces for motion capturing data, for example. We offer a simple and easy-to-understand custom rig for our Cinema 4D figures. However, it goes without saying that customers can also equip our rigged people with their own control rigs and adapt them to their own requirements. The underlying skeleton is based on common industry standards and does not contain any unconventional functions. This simplifies many work steps when it comes to animation or motion capturing, even with other software such as Motionbuilder.

DP: Would that also be suitable for crowd sim?

Moritz Schneider: Absolutely . In fact, we are working with a partner company in this area, but unfortunately we can’t go any further into it yet. Of course, we see the benefits and are trying to implement this as quickly as possible.

DP: And what other figures are planned?

Moritz Schneider: We have adapted our workflow so that we are now able to create scans in a relatively short time. Business, best agers and casual are currently on the agenda. But children and special occupational fields such as the police or other public professions are also in the pipeline.

DP: What does the creation workflow look like?

Moritz Schneider: First of all , our model is scanned in our scanning system. This is done using photogrammetry with 196 Canon cameras. The images are then merged into a high-resolution 3D mesh in Agisoft. We have already tested several programmes, but in the end we always ended up back with Agisoft. We then process the resulting raw scan in ZBrush. This means that we remove all errors from the scan and record important details of the figure again. We then retopologise the scan so that it is also suitable for animations. This is a time-consuming process and takes some time. However, we have developed a workflow with in-house tools and scripts that supports us very well. The texture is reworked again in Mari and all important areas such as the armpits or face are carefully checked. Particularly in areas where the cameras receive little information, we have to do some extra work. The face is also always very important as it determines whether the character looks likeable or not. The figure is then rigged and skinned by our animator and provided with finished materials for sale.

DP: And anyone can use that?

Moritz Schneider: Absolutely! Our figures are always provided for 3ds Max and Cinema 4D with finished shaders as a scene. This is done for the V-Ray render engine as well as the native renderers of both programmes. For all users of other 3D programmes, we always include the .obj for the posed people and the .fbx for the rigged people so that they can be easily imported into any other programme. Maya is brand new for our Rigged People. As Maya is still the number one tool for animation, we also want to offer our people in this format.

DP: And what’s next?

Moritz Schneider: The next logical step is animated people – our rigged people currently come without motion data, so customers have to use their own motion capture data or animate them themselves. We want to simplify this even further and produce fully animated figures. These can then be inserted and placed in the scene via import. We are not yet entirely sure how we will offer this. The idea is that we create several animations for each character, from which the customer can then select the appropriate ones. Another idea is for the customer to choose both individually and then place the file on the figure themselves. However, this is again associated with more prior knowledge and work on the part of the end customer and can lead to many errors in the retargeting process. But we want to keep it as simple as possible – and we’ll let you know when the animated people are ready.