Starting from 09/09/2019, we were assigned a brief consisting of having to develop a horror VR application trailer within a group of 4 people. We have had 4 weeks to do this and it was split up into the following periods of time:
Pre-production (1 week) - All the planning processes such as storyboards and location scouting.
Production (2 weeks) - The main bulk of the project consisting of things such as filming, asset creation and sound design.
Post-production (1 week) - The final part of the project, mainly involves implementing everything together to create a finished product.
My group consisted of myself, Jess, Eduard and Louis. We all had our own set of skills and the ability to coincide what we were able to do with what others could so we were a good match.
Stage 1 - Pre-Production
Storyboard and Planning
We first started discussing potential areas in which we could possibly dwell down. Our first idea consisted of a real estate fiasco into a house which would not end alongside slowly wilting like a dead leaf for instance. We decided to scrap the idea due to it being a seemingly long thing to try and present as a concept in a 30 to 60 second long presentation.
Our next idea followed a similar format, but this time it considered of a baby sitting job, leaving the baby to be the monster reveal at the end. After we had decided what story line we wanted to go with for our trailer, we created a draft storyboard.
After this one I went and made a neater version of it in Photoshop. In these storyboards I've made sure to imply what directions we could possibly dwell down in order to get these different parts of the storyboard alongside small thought processes here and there.
We had decided that for this to fit nicely into what we all thought we were capable of doing we decided to have a main 3D section of footage and a main section of live recorded footage as well as bits of general VFX. The part that would require the most effort and give us the biggest boost in general quality overall was the 3D footage so we started off with developing that. The live footage would come from Jess alone and the group as a whole depending on the footage.
This footage would need to consist of a camera panning towards a cradle through a 3D room to lead up to the final jumpscare which would later be recorded in advanced inside the green screen room, using and old doll provided by Jess.
Whilst Louis was starting in Unreal Engine, Jess was developing conceptual sketches of what the other 3 of us would be able to base our asset development of. I was tasked to model a hanging lamp, a bookshelf and vases. I did them in Maya and sent them over to Louis who would then implement them into the 3D scene within Unreal and apply textures onto them too.
Stage 2 - Production
Asset Development
Before starting with the asset development I needed to look at the concept art Jess created and then develop my own reference portfolio within padlet, that way I had a really good vision of what I had to create.
These really good pieces of Jess' concept art gave me a really good idea of the style I'd have to base my assets off. I made sure to do my research and reference compiling based off of these.
Instead of perhaps taking a direct approach and basing my reference fully off of the concept art I decided I'd best give myself breathing room and make my reference more general. By doing this I allowed myself to think more broadly about what I could potentially do, which ended up working pretty well.
To start the asset development off I decided to go with the hanging lamp first.
I found this image on Google and decided to use it. I have actually created a lamp in a past project so I new the perfect method to construct 3D geometry like this. NURBS curves could be good due to again the fact this was a very symmetrical project.
After some time modelling I had to redo it due to how I at first reversed and converted the mesh. The new lamp had nicer faces so would look nicer textured.
Next I made the bookshelf.
This model was the one I was least proud because I feel it wasn't very planned or thought out well but at the end of the day it looks like a bookshelf so it was ok. The books do look better however and I feel they made it look a lot nicer then perhaps it would've been without them.
I then made the vases using a technique I haven't yet considered up until now, a technique I'll now explain.
So I worked with these modes in smooth mode (3) so that once they were done I could then smooth them knowing they'd look very nice, which I think they ended up looking. For each of these 3 vases I initially knew I'd want them all to look unique in their own way so I decided one would have a handle, one would have a nicely designed tip and one would look unique in the sense that it was generic and perhaps more rounded then the rest if we're looking for more definitive unique aspects.
With each model I started with a basic flat disc (polygon mesh not NURBS) and extruded all edges outwards and upwards, this was I could nicely form them to my liking and also do so with more precision then with another technique. I then went and did the same for the tip downwards but going inside the meshes so that it wasn't a single-sided model.
Group Filming Process
By this point the logo was done and the assets have been done. Next was the main filming stages, one of which was in the green-screen room which me and Jess took charge of and one which was in Epsom common which we all went and did as a group.
The filming processes were very risk potential areas so Jess kindly took charge in assessing the many risks we could potentially run into so she created a risk assessment chart for both filming areas.
The doll filming process was done within the green-screen studio. We were given the idea to keep the doll stable with a sandbag and then we'd move the head with a pole that had bluetac attached so that once it was keyed it would look like it was sitting up by itself. Aside from a few technical hiccups the filming process went very very well yet there may have been room for improvement concerning the colour of the foam pieces because having all of them the same colour would have saved me a lot of effort within keying. Here is a trial piece of footage from my phone to demonstrate how it worked.
The filming process at Epsom Common used a rig you'd keep stable with your shoulder and a camera to walk into a forest clearing so then a house I'd model later could be tracked onto it.
Above is the sample of footage we chose to be used in our VFX composition.
Outside of this real time recorded footage Jess had recorded inside of her Grandmother's house, footage which would serve as a large bulk of the final composite once Eduard put his UI designs over it in post to create the immersion of a VR experience being played by someone, something that looked very like our initial plans in the final composite. Here are her best takes.
In order to make sure we were safe in many aspects of the filming process, Jess kindly put together a risk assessment largely because she'd be the only member of our team present at all filming locations, meaning she'd have the best knowledge of how to critically assess for risks and hazards in certain areas.
Keying Doll Shot
Whilst other members of the group were busy with their own part of the production process, creating the sound, I was keying the footage of the doll we had collected to allow it to work inside the post-production process.
I took the footage and imported it into After Effects where I used the Keylight tool to remove both shades of green in the initial video footage. Once I had a doll and a black screen I had to change the levels of the 'Clip Black' and 'Clip White' sliders within the Keylight UI to create transparency within the image. Once I had the transparent outer area it was easy to move it over the cradle shot Louis gave to me from within the Unreal Engine scene to get this part of the footage.
House Modelling and Tracking
Using the footage we as a group went to collect down at Epsom Common I modeled a house which would be textured so that a front facing perspective view of it could be tracked to the clearing. The modelling process was pretty simple as the model itself was a simple thing to construct.
Here's the house model with it's texture groups, normal maps and textures. The final front-facing perspective of the model would be taken and tracked inside After Effects in post.
Stage 3 - Post-production
The entirety of post-production basically consisted of compiling everything we had into one full video whether that have been UI implementation or tracking assets and effects onto video footage. Due to imbalanced skill sets from peers and I we decided to make a group composite to make the last stages of development as comfortable and stress free for everyone as possible. Whilst Louis was finishing with Unreal Jess created some UI designs which Eduard would be able to reference from when it came to making those.
This would now finish the staircase shot. I tracked the house onto the forest footage and put black eyes and effects onto the baby doll footage at the end. Louis finished with Unreal and implemented the sound alongside some fade in and fade out effects. It was done once it finished rendering in Adobe Premier Pro.
Here is the final video. I found the whole project was difficult but mainly so due to the fact improvements could have been made in the post and organisation department, as an entry into what we should expect from the main final group project however this was
understandable. I feel with more time we could have made this feel more immersive but as a whole it's ok as is.
Comments