Tim Schweizer's Development Blog

This project has been both my most successful and my most difficult project to date. For the purpose of this reflection I will break it down into pre-production, maya production and finally unreal production.

For pre-production it was quite smooth. Setting up a mood board, collecting inspiration and getting a general idea of my piece sorted went by quickly and easily. Alongside that was creating a block out of the scene in maya which also went by smoothly with the exception of trying to figure out lighting. My consistent lighting issues in maya is what drove me to take up unreal and its robust lighting systems to create this work. Afterwards I set up a spreadsheet online to track both my assets and to work as a calendar to split up my time which I’m very glad I did as it helped me keep track of time and maintain a steady and maintainable workload throughout the project. I think of all my skills I’ve developed over the course of my time here time management and organization I am proud of the most. In such a difficult environment like VFX I think having the ability to set myself up to achieve my tasks without overloading myself during the process is incredibly important to have a long and healthy career.

My work in Maya, which was also divided with time in Substance Painter for texturing, was the biggest bulk of my project. Having to work on 20 assets over a period of a few weeks was daunting to me, never had I had to produce so much in so short a time especially with rigging and soft surface models being among the list. Thankfully with my preplanning I was able to produce all 20 assets by the point I needed to port them over into unreal. The toughest parts were the soft surface models such as the skull, ant and clothing of which the latter two both required rigging. With how much time I spent in substance painter, along with the feedback I got on my textures, I significantly improved my proficiency in texturing. Rigging also posed a problem as I had originally wanted to animate in Unreal but attempting to re-set up controls in unreal clashed with how maya had handled the rigs. As a result I had to abandon my original plan and animate in maya. It turned out to be much smoother to animate in maya and deliver those animations to unreal than I had originally thought and I was afraid that my animations wouldn’t work for what I wanted in unreal.

Speaking now then on unreal, I am glad that I had assigned myself so much spare time to experiment in unreal and work out any issues that I’d experience due to my inexperience in unreal. The lighting tools in unreal helped me to achieve the results that I wanted for my piece and to evoke the feeling of melancholy that I had wanted to evoke. Porting over my textures and models into unreal went a lot smoother than I had actually thought it would which left me with a lot of time to experiment with light, height fog and camera motion in unreal. Two issues I was faced with in unreal was having the camera smoothly rotate around the figure in the centre and for the particle system’s random dust causing issues with the animation ‘looping’. While I was able to figure out the camera work as detailed in my blog I wasn’t able to come up with a satisfying solution to my looping issue in unreal with the short timespan that I had left. Perhaps I’ll go back to it but as it is I am fine with it as there will be no chance for my work to loop when it comes to the presentation of it at the showcase.

Overall this project helped me to develop and consolidate all the skills I have learnt over the last year and 3 months in this course. It was tough and not every little thing I wanted was achieved but the overall view of this project to me is a success.

This is the final post of my FMP timeline. This week and last bit spent before submission has been adding the animations I created in Maya of the any walking and looking around into unreal alongside with tweaking light, colour, height fog and particles. What this all meant is that there was little in the way of ‘progress’ as really everything this week was about wrapping up.

The animations were not too difficult to do as they were quite simple and with all the rigging having been handled I didn’t need to worry about that.

To get my animations into unreal all I had to do was bake and export the animations and then import them into unreal. Then I just had to set up my ants in the sequencer and tell them to play this animation and to make them move I just changed their XYZ positions.

The final render with all of these animations and tweaks that I have made over the week can be found under my reels.

This week was dedicated to compiling all my assets into unreal, setting up the scene and animating the camera to get a first animation-less draft and at the time of writing I can say that I succeeded in doing just that!

While there are some issues with clipping and the jitter at the start those are all issues that I am proud to have as they are emblematic of me arriving closer and closer to a finished project.

Importing all my assets into unreal was actually quite simple and was at most a bit time consuming as I had to manually link up textures and assets

But since I organized all assets and textures into their own grouped folders it helped to make the set up go by that much quicker.

Past that it was about recreating my blockout from Maya in unreal and again while a bit time consuming it was overall quite simple thanks to all this being planned out all those months ago. However what was challenging was figuring out three things: the camera movement, dealing with wall clipping and the lighting.

In the blockout the camera swings around the pointer so that it moves in a perfect circle and always stayed locked onto the central figure. After some experimenting I was able to recreate this in unreal by pairing the camera to a cube that I placed in the body of the person and rotating that cube making the camera swing around in a perfect 360 degrees.

The wall clipping was solved by using the Visible in Scene Capture Only setting to make the walls invisible to the camera, while not changing any of the light conditions in the room that would occur if I just used ‘hide’, when the camera was inside them and then making them visible again before the camera would have them in frame again.

The lighting however was my greatest difficulty thus far as I am not a lighting expert. After looking online for brief descriptions about what all the lights do in unreal and playing around with them for a day I was able to get a lighting set up that I believe work in creating a beautiful but not necessarily uplifting lighting environment for the scene.

This week was dedicated to finalizing my last few models and textures before final implementation into unreal for scene construction and with a lot of effort, time and dedication I was successful in achieving that goal. Specifically it was the construction, rigging and texturing of the character’s clothing and the creation of the face texture and creation of a human skull.

Going over the face texture first I created by taking pictures of my fellow students and mixing and matching parts of them in a face texture that I had created. The base texture was one I made from one of my friends

I used an online tutorial on how to set up the texture and what pictures to take. I made this image from 3 photos, front view, side profile and a middle between the two.

And finally mixing and matching resulted in this.

While it is a strange visual it is exactly what I was looking for. A homunculus composed of many different peoples to act as a visual stand in for as many people as I can.

Now the most difficult part of this process for me was getting the UVs for my face to fit comfortably in the 3d model I constructed around my skull. It took a lot of tricks and help from my friends but I was finally able to get my UVs to fit. Out of all the challenges I faced this week this was easily the most challenging as thus far I’d created my own textures in substance so I had no need to worry about conforming UVs to a pre-existing texture.

The skull underneath it was also textured

The skull was actually quite easy to construct. With all the refence images I had blocking it out wasn’t that bad. However what was difficult was making it look more organic, I am still not totally comfortable with the sculpting tools but I think the final outcome was really good!

Lastly the clothing, this was the refence image I was using to construct the outfit.

The technique I used to construct the outfit was to take a base model available in maya and cut out sections of its topology to be used as a basis for the clothing. This was a quick way of constructing accurate clothing and proved to simplify the process. What I wanted to do with this texture was to, instead of having each item have its own texture like I had always done, I have the entire outfit as a single texture with their UVs properly organized. I wanted to do this as to optimize my textures and not have the issue I tend to have which is one item having 3 or 4 textures which results in 20+ images that I later need to hook up to produce the texture in maya/unreal.

In substance to create the individual clothes I just had a series of white masks to, in substance, separate the clothing into individual parts. While this obviously would have an impact on fidelity the final result doesn’t really show that loss of quality. As a result this model had a complete outfit as a single texture which tremendously streamlined my workflow for setting up the texture in unreal.

I also rigged the outfit using Human IK so I can later pose it in unreal.

This was the first time I had to rig fingers which wasn’t to difficult thanks to an online tutorial.

This week was again getting the tree and ant models completed, with the addition of the ant model being rigged! I started the week off working on the tree.

Using this image as reference for the tree

I started with a blockout and adding details via sub-divisions, loops and circulations

The tree itself is thicker than the original references as this tree will need to ‘absorb’ the leather armchair that I have made and I’d use the modelling tools in maya to imitate that ‘absorption’ trees do as they grow around an object.

This was a new experience to me a I very rarely use maya’s more complicated sculpting tools to grab and shift topology like I was modelling an actual piece of clay. For the foliage I decided to use maya’s brush generators to create the leaves.

This was a lot simpler than I had originally thought it would be and came out with a decent looking effect. It might be that I end up replacing these with foliage in unreal.

To help alleviate the more depressing aspects of my piece and further the idea of it being meant to be taken more melancholically I decided to make the ant model based on a cuter more stylized reference. This meant more simpler design and rounder shapes. The texture however is more realistic to not make it stand out too much in reference to the other realistic models.

I also rigged this ant and it is one of the best rigs I’ve managed thus far. One of its major improvements is my implementation of IK handles as previously, due to my inexperience, I hadn’t included them and when animating this would result in weaker results with longer times to animate.

This week I was focused on creating the walls, ceiling and floor of my scene. Using my more simple chair model I scaled it up in my block out scene and used it to ‘measure’ the width, length and height of my block out scene so I could recreate that size exactly when constructing the proper walls again.

Using that I constructed the scene like this making instance of simple cubes. With it set up like this I would quickly be able to construct the room in Unreal as all the parts of it have been made to fit together in unreal. As I am still inexperienced with unreal I do not want to make it one object in order to leave myself room in Unreal to change them individually should the need arise.

When constructing the window I originally didn’t have a border around it and it blended into the wall. A fellow classmate pointed out how that would end up making the windows look strange as that border is necessary in reality for how windows are constructed and would make people look at my wall and feel that something was off.

I then took all these and individually textured them in Substance painter:

The texture work here I feel is better than my previous ones as I had received a lot of useful feedback from my professor and from other students with tips and hints on how to improve them. As a result I also put time aside this week to go back to my previous textures and rework them in order to keep a consistent visual language.

I think these two have a marked improvement over the previous textures conveying the “old and left to rot” feeling far more.

The hardest thing this week would be making the floor, ceiling and walls as I had very much underestimated what I would need to do for them. In my own mind they were far simpler than what was actually needed by the end. All that being said my biggest fear going forward is that I am now moving away from all the models that I had confidence in producing. The next model I have planned is for a tree which is more soft surface. As each week goes by the deadline feels realer and realer and it takes some strength to not force myself to rush through things to ‘give myself more time’. I know though that that would only hinder my final project and I am still well within my projected timeline.