10/7/24
Half way point!
Yesterday I had asked my teammates to give me the most recent versions of their assets that I could work with to put together a comp by 5pm today. Here's what the current comp looks like for shot C02 with the most up to date footage from them.
I have found a bit of difficulty pulling a difference matte from the background of the actress, so I have pivoted and am using a few luma keys and keylights to get a comprehensive key of each part of the actress. As I make more progress on the shots, I will lean more into using roto to isolate specific parts of the body to key those separately where necessary. Here is the script for C02!
10/8/24
Today we had met with our mentors to get feedback. The biggest goal I have for when we meet next from our mentors is to get as much Roto work done as possible and make the edit more cohesive. Overall the ads for each group needed improvement with their pacing, and that there was too much dead space; which I plan on amending for our group.
Additionally, all cg assets except the set extension are now going to be rendered in Arnold and output as ACEScg.
Here's a quick composite of the assets from Kulin.
10/9/24
As there has been a bit of a change of colorspace I went and did some research as to how that will be effecting our compositing and output process. OCIO, Display Transforms and Misconceptions by Chris Brejon does a great job of talking about the complexities of colorspaces that I found very insightful. Relatively for the most part as the base footage is in raw, there just still needs to be cohesion with all of the CG aspects that will be intersecting. Eva and Kulin are working in Arnold for Houdini and Saw is working in Karma for Houdini, and as long as everyone renders their deliverables to me out in ACEScg there should be very little issue with colorspace.
10/10/24
Eva changed the particles according to the suggestions from the mentors, and Saw continued to work on the set extension. Here's a version with and without the purple trim, and with a rough blur.
Saw gave me the first pass with depth, so here is the depth layer, and can start to play around with getting a good dof matching the shot.
10/11/24
I modified the edit in accordance to the mentors suggestions. Removed dead space, and changed the last shot to a different shot that we had shot on the first shoot day to be a larger live action component.
10/12/24
I have started to play around with external lighting in Nuke, throwing a landscape photo of my own in the background taken in Waterton, Canada, compositing the current elements and adding a very slight defocus of the background using the depth channel which is the depth_extra channel in Karma and everything else that is cg is rendered in Arnold.
As Eva has been working on iterations of the particles and pyro, she has rendered the pyro and particles in the same layer, so the interaction of the particles isn't lost in the composite itself.
10/13/24
The lighting of the set extension has changed as we realized the lighting in the set extension was not matching both the canvas that is lit by the live action plate and component. This is very evident by the luminance. It is going to need a bit of iteration change.
10/14/24
Continuation of the depth of field experimentation and working on it. I've worked with the zDefocus, PxF IDefocus, and Boris Continuum(Depth of Field) nodes to create different iterations.
10/15/24
Most recent up to date version of the main hero shot. I chose another temp photo for the window that is taken by me in Lincoln, Montana. For the hair, I've found keying it separate with a difference+ keyer, and then blurring and subtracting it over the previous roto has been successful in getting a good hair key. Roto always ends up taking so much longer then I initially expect it to, I enjoy it a lot regardless.