Statement Of Intent

For my dissertation project, I want to focus on two major things.

1. To create a micro-narrative short led by composited CG characters in a live action environment that meets a standard seen in cinema.

2. To collect specific data, perform look development tests, and follow current VFX pipeline techniques.

Over the course of my time at University, I have learned a number of VFX techniques, but I’ve never brought everything together. I’ve composited with a cinema level camera, with less than ideal camera work and no planned shot. Similarly, I’ve had a good camera with difficult CG. Or maybe I’ve had CG with many AOVs that would be great with compositing, but the original designs and intent restrict the final image from looking as good as it could.

This is all well and good for gaining problem-solving skills, and as I eventually want to be a compositor, a lot of factors I won’t be dealing with and just need to find compositing solutions to things. While I’m confident that this project will further strengthen my passion for compositing, I believe that gaining a broader perspective on how artifacts, assets, and ideas are exchanged and developed, particularly in the realm of VFX, throughout the course of a project, is extremely valuable knowledge.

Appreciating and understanding the work of other artists will only have a positive impact on your own. A skilled compositor excels at compositing, but they also possess a deeper understanding of the craft. I want to comprehend the “why,” and I consider shooting with a high-level camera as an advantage, as comprehending the capture process and fighting the high density pixels to achieve true photorealism will only aid me in merging the fantastical with reality and developing my skills further.

A must for the project is a high level camera, I want to use the Arri Alexa Classic as it’s no secret that Arri cameras have superior dynamic range and colour science which is why they are the go to for the majority of films nominated for best cinematography. However, a cinema grade camera that shoots in 4k will also suffice. I want something where you have to get it perfect, because when shooting at such high fidelity, you can’t hide little mistakes.

For this project, I also want to create my own HDRI’s. This is a fundamental aspect of marrying CG with the real world environment, especially when it comes to reflections, which are going to be important due to the material used for robots. I am likely going to need to do a fair amount of comp grading, but if I can light as close as possible in Arnold, I am going to end up with a better result, and using HDRI’s from the actual location rather than something found online that is “close-enough”, just makes sense to me.

On the compositing side, I attended a talk at the VFX festival in London where certain AOV’s were mentioned.

1. Cryptomattes.

2. Utilities (N, Z, UV, P)

3. Diffuse, Specular, Shadow.

4. Motion Vectors

This was for a shot done by Union VFX for a Disney+ show, so this is likely the kind of things I would expect in industry. Whilst I would have chosen the majority of these anyway, the UV AOV is something I have less experience with. But I foresee it being highly useful for adding any extra textures later on, or at least creating specific 3D masks.

The rest I have used before, but for documentation sake I will explain how I intend to use them. Cryptomattes will be used a lot for controlling what effects I want to apply where, certain parts of the robot may need grading but not others, I have complete flexibility with them there. N and P worlds I will likely use for 3D mattes, again for specific grading, but they can also be useful for if I need to do any relighting. Motion vectors mean I can have a bit more control over the CG, as it can be easier to grade and edit without any motion blur being applied. And finally the light passes are useful because I can independently control certain factors of the CG like the shininess, without impacting others.

Additionally, in order to make the whole process easier for me, I will be collecting lens data using a lens gird, reflection and light reference using chrome and matte balls, and using a Macbeth chart in order to get accurate colours. This project is me showing off what I can do, but there are a number of things I can do to make it easier for myself later on down the line that are a necessity in the VFX pipeline.

I will have certain restrictions though. If this was a larger budget film shoot then there would probably be model versions of the robots built to get even more accurate size comparisons and reflections that could be used as reference in comp. I simply do not have the resources for that, so will have to make do with what I have. I will however be giving the robots specific heights and bringing a tape measurer to my shoot in order to keep everything consistent throughout the shoot, I will essentially be filming nothing, but I can’t have it feel random. To also assist in this I will have to place physical tracking markers, this both means multiple takes always focus on the same area and also means I can track and position CG in the exact places I want with an obvious point of reference.

Really my project is as about how preparation beforehand should mean that by the time I’m using Nuke I can be efficient, accurate, and also completely on top of any possible issues, because they have been thought out beforehand. I fully expect to still run into problems, but even allocating time for things like that is a necessity when it comes to planning a VFX project.