Video VFX breakdown: Mavericks VFX on s3 of ‘The Handmaid’s Tale.’
For season 3’s ‘Household’ episode of The Handmaid’s Tale, Mavericks VFX turned the Washington Monument area into a Gilead cross. Scenes in the ep required a mass of handmaids, and for several shots this was achieved with digi-doubles.
In this article, Mavericks VFX visual effects supervisor Leo Bovell details the work, and shares a VFX breakdown of the process.
Leo Bovell: One of the more challenging parts of the sequence was our digi-double handmaids. In total we had approximately 22,000 plus digi-doubles. Large CGI banners and flags. This all required the Mavericks VFX team to come up with a realistic way to manage a lot of large detailed cloth simulations in a timely and flexible manner.
After a few days of R&D, the team settled on a combination of Maya and Houdini to pull off the crowds. All of the hero detailed cloth simulations were handled by Houdini, this included the large Banners and flags which were handled by our FX lead Stephen Wagner. We really focused on the look of the cloth as Stephen Lebed, Brendan Taylor and myself became very sensitive about the look of CG cloth. If done wrong, the audience would be able to tell and that would subtract from the sequence.
On the day of the shoot, it was rather windy which presented a complexity to our simulations. I took a lot of on set reference of the handmaids to capture the details of the cloth interaction i.e (how the cloth compressed and folded when the actors kneeled, how the wind would ruffle the gown, etc). All of this reference was then passed to the CG team.
The scene was all assembled in Maya for lighting and shading. James Minett would set up principal lighting to ensure continuity between scenes working with myself as I created a few shaders that would smartly diversify the skin tones of the handmaids. The team then added additional digi-doubles using Mayas’ MASH network and then rendered our CG crowds and environment using V-Ray.
The compositing team was already working feverishly to remove all spectating crowds from the plates. This was quite challenging as the set could not be closed since it is a national park. It is the first time I recall being on set and having a few people (civilians) follow me around asking me questions about my job while I was scrambling to get HDRI data and set reference! After, the comp team was able to complete the challenging paint outs and reconstructing areas of the set. The CG flags banners and digi handmaid crowds were then seamlessly integrated by the compositing team lead by Rob Greb.
Question on the wide shots: What’s the benefit of splitting the extras up into separate tiers like that, versus pushing all the extras to the front and filling in the back rows with the digital doubles?
I wanted to have real women in the front row of each block. That way if we couldn’t get the look we were after, we would still have the real women covering the doubles. Thankfully it all worked out.