My Compositing Journey
Hello, i started my Compositing Journey nearly a year ago at PIXLVISN and i wanted to share my lasted projects i did.
The Romans are more then we thought
Planning and References
For this project we had a theme guideline from the school which was Roman Empire and Greek Mythology. Because I wanted to work with real footage for this project it was quite hard to find a good plate. After some research I saw the Plate with the Astronaut on ActionVFX and knew I wanted to work with it. First, I was not sure how to connect an astronaut with the Roman Empire or Greek Mythology, but after some time I had the idea that the Romans could have traveled to the moon. So, the project "The Romans are more than we thought" was born.
Gathing references luckily was not the problem. There is some high-quality footage from the Apollo mission which really helped to build a To-Do-List for the scene. The time restriction of 1 week requires good planning and time management. Otherwise, you are not able to finish your project.
Schedule (7 Days)
- Human Matt-Painting (Day 1)
- Camera tracking (Day 1)
- Modeling Asset (Day 1-2)
- Enviroment building (Day 2-4)
- Character integration + compositing (Day 5-6)
- Cutting the video + some backup time (Day 7)
Start of the Project
In the beginning I analyzed my Plate. I quickly realized even though my character is in front of a green screen keying the character will be difficult. This is because of all the different green colors on the green screen. After some brainstorming, I remembered the tool CopyCat which should work well on this shot. The character is not moving a lot and the lighting does not change in the shot plus the background is static. The only question is how many frames I should take to train CopyCat. In the End I decided to take 11 frames. Rotoing the character was luckily not hard because of the helmet the character wore. After 2 hours I was done Rotoing and started to crop the image to make it more efficient. The preparation helped to reduce the time which CopyCat needs to learn, drastically to 25 minutes.
When I was certain I was able to use the plate, I tracked the camera and exported it to Maya to start blocking out the scene and get a better feel how big the scene would be.
The Next Steps
With my first Blockout I asked for feedback and how I can improve my scene overall and asked what future problems I may run into. The feedback which I got the most was regarding the lack of detail in the foreground, so I quickly added some details and break ups there.
Another crucial point was that the walls of the crater are too high, and you need more space, so it does not feel so squeezed together at the top. After I implemented that feedback, I started modeling my main asset in the scene of the ruin.
I started inside Maya to get the right shapes for the ruins. When I was happy with the general shapes, I exported the model to ZBrush to sculpt details on it. The details I later baked on my model with Substance Painter where I also textured my asset. Alone this process took 1 and a half days. I was already in the end of day 2 and I still needed to build my environment, render it out and later composite it together.
To maintain my schedule, I immediately began building my environment. The moon's surface is just a lot of small rocks and dust. So, I decided to use a texture with displacement to get the very small details and then later scatter some bigger stones to break up the surface even more. Luckly, I found an awesome gravel texture on AmbientCG, which I then color graded very settle to give the right look. Then I downloaded some stones from Quixel to scatter them with the Mash-Tool inside Maya. From the beginning, I planned to place some statues in front of the ruin to make it more interesting and to get a good feel of scale and depth inside the shot. On SketchFab I found some good ones, which I use for the scene. I found it a bit boring that only the character is moving in the scene. So, I added some astronauts in the midground, which are inspecting the statues. The astronaut model I found on Sketchfab and the animations are from Mixamo.
The lighting was clear from the start. The plate gave me a strict lighting direction and source. In the end I just added a single light source inside the ruin to make it a bit more interesting to look at.
Compositing
While the shot was being rendered, I started working on the compositing with a few pre rendered frames. I work on the lighting with the light groups I rendered out and improved the look. Integrated the character into the scene and started with the shadow. When my render was finally finished, I realized that the shadow from the plate did not work at all even when I animated it. In the end I decided to replace the shadow. I used the Astronaut which I already downloaded. Put a walking animation from Mixamo on it and imported it into my scene. The shadow pass which I rendered out was used to create better interactions between the shadow and the ground. It did not work in the end, and I animated a transition between the rendered shadow and the real shadow, which was not perfect but at first glance you do not notice it.
Until now I had nothing to see over the edge of the crater. I thought I would look quite nice if you could see the earth from the moon. The earth is just a sphere and some high-resolution textures from VisibleEarth. The atmosphere I added later in comp like the stars.
In the end I created a lens flare and complemented some other lens effects to make it a bit more realistic.
What software I was working with:
- Maya for modeling and building the scene
- Maya Redshift to render the scene
- ZBrush to sculpt the details on my model
- Substance Painter to texture my model
- Nuke + CopyCat for the compositing
- DaVinci Resolve for cutting my video
Resources i used for the project:
- Music: Thunder by 39222982, Bouncy Drill(Prod. by YoshyBeats) by Yohan Marchand
- Soundeffects: from Pixabay
- Statue and astronaut models from Sketchfab
- Astronaut animation from Mixamo
- Plate from ActionVFX
- Ground texture from AmbientCG
- Earth textures from VisibleEarth
Conclusion
I learned a lot in this project: How to work with real footage. What you need to look out for. How to replace a shadow. The full pipeline of a set extention. Implement Mixamo Animations. How to use CopyCat for Human Mattpainting.
There were some problems I also faced like the shadow of the character which I needed to replace for half of the shot. As well as some render problems I needed to overcome in time.
But in the end, I am happy with the result and the things I learned in the process.
Fire Operation in New York
Approaching the project
For this project I wanted to change some things. First, I wanted a longer shot, second, I wanted a shot where I needed to integrate some things. Plus, a third shot where I have no data at all, and I need to get it all from just the plate I have.
We also have some guidelines for this project. The timeframe in which we need to finish the project is one week.
At the beginning I looked at the footage and realized a few things. Some cars are driving through the space where I am planning to put my cars, so I need to roto paint them out. Also, the camera move is very stable so it should be no problem to track the footage. With that knowledge in mind, I started to prepare the shot and track the footage.
After importing the model into Maya I started to place some firetrucks and police cars inside my scene. Also, some firefighters, who were moving around. Then I added some smoke, which I created with EmberGen and imported the fire then into Maya as a VDB.
The most important part for the CG-Integration is lighting. For that I looked for a dim HDRI for the fill light. I placed an orange light at the place where the fire should be and placed one light for every siren. Instead of animating it inside Maya I rendered all lights as a different light group so that I could changes everything later in comp. Also, I made a few different render layers:
- The lighting on the buildings
- The cars and people
- The shadow from the cars and people
- The volume
- The shadow of the volume
Later in comp I did the full paint out of the cars as well as adding some more effects on the sirens, so they are easier to spot. In the end I did some final touches to improve the look of the shot.
What software I was working with:
- Maya for modeling and building the scene
- Maya Redshift to render the scene
- EmberGem for creating the Volume
- Houdini Karma to render the Volume
- Nuke for the compositing
- DaVinci Resolve for cutting my video
Resources I used for the project:
- Music: Leonell Cassio - The Blackest Bouquet
- Soundeffects: Pixabay
- Firefighter trucks, police cars, ambulences and firefighter from Sketchfab
- Firefighter animation from Mixamo
- Plate from Dronalist on YT
- HDRI from HDRI Haven
Conclusion
The project was a challenge. I had a lot of problems with rendering, which I was able to fix after some time, but this time I was missing them in the end.
But in the process i learned a lot. The most in rendering with Redshift and roto painting out objects with a changing perspective.
But in the end, I am happy with the result and the things I learned in the process.
The Temple Of Time
Planning the project
We were given the theme of "Nintendo" for this project. I was searching for a recognizable building in the Nintendo universe and found the temple of time, as well as some nice artwork with it. I had the idea to make the camera come out of a cave. For this project I wanted to do a matte painting for the background, so I chose a mountain which fit the scene.
Another thing which you need to consider is the time for the project. We just had one week for the whole project, so the schedule is quite full if you want to model, build an environment and do a matte painting as a background.
Starting the Project
The first thing I did was create a blockout of the scene to get a good feel for it. This helped me answer the question of how big was it? What should I be careful with? Another very important part is setting up your camera. This is the first thing you want to be done with before you go further into details.
The next steps were to model the main asset the temple of time. I calculated with one and half days to finish the asset. While working on the asset I had no problems and stayed inside my schedule. After I had the asset finished, I implemented it into my scene and started working further on my environment. The cave was the first thing I worked on. I added some mushrooms and some vines Infront of the cave to make it a bit more alive.
The next day I started working on the environment around the temple. Scattering stones, trees, grass, and rocks really help to create a good-looking environment. Also, I added a bridge which connects the cave and the temple. In the cave I put some particles around the mushrooms as well as some caustics on the walls. The small details are there to give the whole scene a bit of movement.
The last thing I worked on was matte painting. For this version I just slapped all mountains together with the sky and put it behind the render.
What software I was working with:
- Maya for modeling and building the scene
- Maya Redshift to render
- Substance Painter to texture my model
- Nuke for the compositing
- DaVinci Resolve for cutting my video
Resources I used for the project:
- Music: The Light of the World by Jeremusic70
- Soundeffects: from Pixabay
- Toad, Rope Bridge, Oak Trees and Cave model from Sketchfab
- Fooliage from Quixel Bridge
- Textures from AmbientCG
- HDRI from HDRI Haven
- Birds Alpha from Asma Begum
Conclusion
I learned a lot in this project: How to use the 3D-Space in Nuke in different ways. How to do a matte-painting. How to do compositing with a full CG-Shot.
Luckly, I didn't have a lot of problems in this project, but the fireflies in the render were really taking a lot of time to deal with, but I was able to find a solution for it.
In the end, I am happy with the result and the things I learned in the process.
Comments (2)