Dev Log #19

Welcome to 2024! We have some exciting plans for this year, but before we can make any grand announcements we need to finish what we started and polish off our previous projects 🙂

The Hidden Alley

Another wildly successful community project! You can read about it here, head straight to the assets page, or download the final scene file.

We’ll certainly set up another community project later this year, but for the next few months we need to focus on some other popular projects…

Little Falls Verdant Trail

Our latest environment project is just about complete! We decided on a final name for the collection – “little falls” was simply a working title and no longer suitable since there are no falls in our finished scene – instead we decided “Verdant Trail” is more in style with the theme.

Here’s a small sneak peek, we’ll share more in a few weeks once all the assets are uploaded:

Fabric Plans

Last time I shared a bit of our big-picture ideas for scanning fabrics at scale.

Since then we’ve been researching some other approaches and testing out a few different workflows that might work at the scale we want, without compromising on quality, nor spending 99% of our effort on 1% gains.

We’re still ironing out the kinks (though not yet literally), but our plan is to construct a variant of the Cornell box so that we may capture solid physical reference of each fabric sample and use this to aid in the replication of material properties in Blender after we’ve digitized it with the photometric stereo workflow.

This may not be the final design, but the idea is to keep it simple while still providing as much information as possible to aid in discerning the subtle view-dependant material properties that are critical for creating a convincing fabric material such as sheen weight, roughness, subsurface scattering, etc.

The fabric sample will be wrapped around a polystyrene sphere and impaled with an aluminum rod to support it on the back wall of the box, like a large leathery lollypop.

A two-point light setup gives you a strong, but not too harsh, key light and a rim light in predictable locations.

We forego the red and green walls of the traditional Cornell box to avoid confusing the hue of the fabric sample itself. Instead, half the walls are painted white to bounce some light to the underside of the ball, and the rest are dark to keep enough contrast for the diffuse gradient to remain visible.

The 18% grey sphere and Macbeth color chart provide an anchor point for exposure and white balance respectively, ensuring we can calibrate the surface albedo accurately.

#TheMoon

Prior to the Blender Conference, we visited a lunar simulant facility in Rostock, Germany.

You can hear more about the project in my lightning talk below, but in a nutshell, the aim is to create relatively small-scale textures of the lunar surface as accurately as we can without actually visiting the moon ourselves.

We’ll likely only get to processing these scans later this year, as we want to first complete the Verdant Trail and Namaqualand collections.

Q&A Stream

Finally, we hosted a stream at the end of the year to answer some community questions and share some updates and plans: