For this weeks task, I attempted to use several different methods in order to remove the markers correctly. The rotopaint and cloning technique was not working very well when the markers changed position. I ended up using a roto node combined with a blur node to remove the markers. This was surely the most effective method, as the others all created different issues.
For this weeks task, we were asked to remove the green screen and despill from a clean plate.
This was my second attempt at the project, and I wanted to develop the same concept into the scene, with a small bridge leading over a gap in the floor. This time however, I created the models and applied the textures in Maya and Substance Painter, and rendered out the sequence in Arnold. I also recreated the rotoscope in Nuke and correctly removed the light in the left side of the plate using RotoPaint and Roto nodes. Starting from this point was also proving much more beneficial, as I was aware of the mistakes I had made previously, and I did not want to repeat them.
For this project, we were tasked with implementing objects into a short sequence realistically, that is to correctly follow the camera movement and perspective according to the track. FIrstly, I began creating a 3D scene to attempt a recreation of the garage. As we had covered 3D lighting in Nuke a short while ago, I found this area very interesting and quite powerful. To implement this scene, I created several planes and extruded them in Nukes 3D viewer, and created 3 different lights. Two lights in the first area of the garage, and one glowing red light into the machine room. These were all set to vary in intensity to capture the lights into the 3D scene.
I also added the mouse created in my collaborative project, and animated it to a basic degree so that it winds and turns to look at the camera as it approaches. I wanted to also have the mouse jump over the bricks into the machine room, however this did not quite turn out as I had hoped. Instead of jumping over a 3D wall asset, I instead faked the disappearance of the mouse with a Multiply node keyed to make the mouse disappear as it jumps upward towards the broken wall.
Shortly after creating a basic shape of the 3D walls however, I started to have serious issues with the way the geometry followed the perspective of the plate, and creating geometry to follow the shape of the broken wall was turning out to be extremely difficult as the sequence played on. as well as Nukes modelling toolset to be a bit less usable when compared to specific 3D software. All of this was generally an attempt to capture 3D lighting, but this was not proving accurate as the track played through. The largest issue however, was the strange results I was getting when the camera moved closer to the broken wall, as the lighting would bounce off seemingly thin air due to the perspective change.
Another major hurdle I encountered was texturing, as moving from 3D modelling and texturing software I made the mistake of thinking the same process could be done in Nuke. As I was having some success exporting geometry from Maya to Nuke in Alembic format, I experimented at first with exporting the geometry with correct UV tiles, and tried to texture using the ‘stmap’ node and exported texture maps. This was largely unsuccessful, as the textures stayed in the same position whilst the camera moved, creating a dizzying effect and not at all what I had envisioned. In an attempt to fix this issue, I tried to transform the textures according to the camera movement, but this again was proving not to be a solution. The final alternative was to use flat colour swatches without any texture to alleviate the dizzying effect. This is not how I wanted the final models to look however, hence why I began again with fully textured sequences.
I enquired with several of my classmates as to the issues I was experiencing, who tried to help me as much as they could, but through my process I soon realized that I had approached this project from almost completely the wrong angle. As my specialty has ultimately been working in 3D, understanding composition when working with 2D plates is another aspect which I have to consider more carefully, and furthermore realize that the rules which apply to 3D software do not necessarily translate into developing compositions. The reality was becoming clear, to begin the project again, recreate the rotoscope and eliminate the 3D scene. Also, to replace the objects and textures with sequences taken from Maya with already textured and lit assets. As well as this, I wanted to develop a much more cohesive script, which does not create a large amount of issues as created by my first attempt.
Despite all of these problems I encountered during my first attempt, I still feel that I learned a great deal about working with Nuke, and surely what not to do when developing a balanced and understandable scene.
For this task we were asked to remove the green screen from the actor plate, as well as any film noise or green despill. We were shown several ways to accomplish this in Nuke, including KeyMix, Primatte, Ultimatte, and by using gizmo nodes.
For this weeks Nuke project, we were tasked with compositing a rendered car plate onto our choice of three environments. We also had to colour match several different channels to incorporate the car further, trying to match the tone and colour scheme set by the surroundings.
This task was interesting as lighting and colour grading are some of the processes which I most enjoy working with in composition, so the tools in Nuke were quite impactful when editing a scene with ability to match white and black points, grading each individual channel, and balancing shadows, as well as much more.
For this weeks Nuke task, we were given a short video plate with tracker points, and our job was to track the scene using 3D Camera Tracking using Nuke’s 3D toolset. We learned how to import geometry and work in Nuke’s 3D editor using Cards, Axis’, and setting origins by selecting Camera Track Points to identify ground planes and setting grids. We also used the LensDistort node to correct distortion between the camera and plate sequence. I also wanted to try importing 3D geometry into the scene, so I exported a base model from Zbrush and imported the model into the script.
For the next part of this task, we were asked to remove the trackers in the clip with the use of Rotopaint, Roto, and 3D Camera Tracking.
For this weeks task, we were asked to remove elements and posters from the scene using rotopainting tools in Nuke. We learned to pay attention to the grain of a film and to match it using adjustment nodes, and also to match tracking information gathered from objects in the scene.
For this assignment the task was to replace the posters on the wall with different designs. This task took me a fair amount of time to understand the nodes and how they worked together. One particular issue I had was scaling the picture down to the correct size but after some research I found the solution: using corner pin trackers and transform tools.
In this revised scene, I wanted to have the balloon move more realistically, whilst still maintaining a sci-fi aesthetic and style. I changed the background of the scene to a night sky which slightly rotates. I used colour grading throughout to try and make different elements either stand out or blend in. I used an exr file to create the ray which expands from the ship, and I used rotoscoping to change the opacity of the balloon as it gets “absorbed” by the UFO. I also rendered a separate pass from the balloon model with an ‘X-Ray’ material applied as the balloon dissipates. I also added some sound effects from YouTube and slight dynamic glow to the x-ray plate.
This was inspired by several science-fiction films, namely Star Trek, Star Wars, and Independence Day. The contrast between a sci-fi scene and a 1900’s hot air balloon may have been slightly difficult to achieve, I attempted to balance the colour grading by keeping the UFO and its movements to the top half of the frame, and the hot air balloons to the lower, mountainous region with a dulled hue.
I generally wanted this project to be comprised of as many original assets as possible. To this effect, I did use a video clip from YouTube for the rotational stars, but other than the original plate and what was given to us to create this sequece I developed several key elements based on the genre of science-fiction for the final render. The portal and firing particles, the UFO, and of course the balloon.
I wanted to change the scene to make it more interesting and realistic. I created a portal effect in Maya using particles and 3D fluid containers, to simulate an alien beam ray which absorbs the balloon.
For this (unfinished) short sequence, I used several different nodes such as colour correction & grading, rotoscoping, keyframe transforms, OFlow time adjustment and other time adjustment nodes. I also used Maya and Substance Painter to model and texture a quick UFO, and I also used Maya’s fluid and particles system to render a fire simulation.
Initially, I wanted to create a fast-moving animation so I could utilize some of the nodes I had learned to use. As well as the tools, I also wanted to incorporate some of what I had studied in 3D Animation just with general transforms and keyframing. Despite this version being unrealistic due to the speed of the balloon, this short sequence taught me a fair amount about how to use merge nodes, the general layer-stacking in Nuke, how to animate and sequence different plates, and further knowledge of rotoscoping.
For this weeks task, we were asked to replace the screen’s image and rotoscope the fingers moving over the screen. While the rotoscoping was understandable, I found the tracking of the corners of the screen to be quite difficult to understand. For the image, I chose a short playlist from Spotify which was an appropriate size, and tried to match the movement of the hand to the screen scrolling. From an animation standpoint this process was recognizable, but otherwise I think there is a lot of understanding to be had with Nuke and how everything works in conjuction with each node and their placements.