I try to take screens inside the houdini and it works for me. Because I can visualise the model collapsing before I render it. And I finally solved those problems I mentioned last week by re-tracking the camera and resetting the parameter smoke.
This allows for a more realistic reproduction of the smoke (debrissource)
Final collapse test results
gonzalo set us up with a shot dealing with London destroying the city of London, firstly I needed to download some footage of explosions and smoke to create the feeling of destroying the city.
I tried to match the explosion to the window and make some black burnt marks. But not knowing how to delay the appearance of black when the flame explodes, gonzalo taught me to use the node “erode fast” which can turn the material into pixelation to add blur and change the colour.
The basic logic is: first create a noise and add a roto, then merge it with “erode fast”, and finally add a “copy” and an “idistort” to create a distortion effect.
On top of that, I need to create the ambient light around the explosion. Initially I used “curvetool” to do this, but the effect was very small and fake looking, so I also asked gonzalo and he taught me to use cornerpin instead, which was a very useful solution!
First, use Blur and Multiple and Cornerpin to create the light and shadow of the flame on the ground and the light and shadow on the glass.
Second, we need to make the background deformation when the flame writhes.
This step is to use roto to separate the parts that need to be twisted
curvetool： This node helps us make the flame flicker throughout the scene
Click on CurveTool, copy its data, create a Grade, and paste it onto it.
It must be remembered that if you want to modify the brightness, you need to go to edit expression to a certain number.
Firstly, I need to do camera tracking inside nuke so that I can determine the location of the building as well as go to the scene to match it. A big thank you to minhan for using mocha to make a “character roto” which helps with the masks used for tracking.
I imported into houdini by exporting the nuke camera, and these locators below are the axes that used to be in nuke, which makes it tremendously easier for me to locate them.
Firstly, I need to group the models I downloaded online so that I can subsequently paste the materials. And then crumble it.
I ran into a huge problem when dealing with crushing, which took me a week to fix. Many models downloaded from the internet have no thickness but only flakes, however houdini crushes the flakes inside, initially I tried converting to vdb mode but it takes a lot of time and tends to crash if you use this mode, and in the end it doesn’t even render. After going through a lot of tutorials on the internet, I decided that it would be better to process the faces in maya and give them thickness, which is a very tedious step, so if you have the chance it would be better to model it yourself.
I managed to break the staircase in its entirety, next I needed to create the smoke produced during the staircase breaking process.
I used the node “debrissource”, which helps me to identify the particles in the middle of the breakup, through which the smoke is generated.
Still running simulations, I’ve tested at least 4 versions
After importing into nuke, I checked and found a few frames that were missing and had extreme problems with the camera tracking, I needed to fix them.
1.Modify scatter
2.Adjust the camera tracking
3.Redo the smoke spread
4.Re-render
Firstly, we discussed a number of versions of the story, for example identifying the theme as a lonely boy, but due to the difficulty of creating vfx for the story scenes and the lack of logic. So after a lot of discussion we decided to go for “The choice of the plastic bag” and create a vfx for the last part of the video.
So, after the theme was chosen, we explored the plot: the next step was to find a coffee shop where we could film and draw the subplot. The following subplot was drawn by Minhan Lv.
On Wednesday of this week we gathered together in a restaurant near King’s Cross railway station for a three-hour shoot. As I had injured my shoulder, I had to play the role of director to help them out, sakshi did the filming and minhan and mahadev did the roles.
Our plan was to create special effects in this environment, such as a city that starts out as a ruined and destroyed city, and gradually turns into an environmentally friendly green city.
Very enjoyable shoot, first experience of the rigours of shooting. You need to shoot a shot many times to get the effect you want.
Here’s some of our dialogue
Minghan :Hi bro how are you doing
Mahadev: Good ,Good
Is it a plastic bag?Minghan :yeahh (Pause)
You know these plastic bags are just so convenient. Who cares about reusable ones?Mahadev :
I doUsing reusable bags is so important for the environment.
Its about making small changes for a better futureMinghan
Oh come on. Whats the harm in using platic once in a whileMahadev : Imagine a world where everyone thought that way. Our cities would turn into chaotic messes, covered in plastic waste. Pollution would choke our air, and wildlife would suffer immensely.
Minghan : Thats so true, What have we done?
Mahadev: Small choices can make a big difference. Choosing reusable over plastic isn’t just about convenience; it’s about preserving the beauty of our planet for generations to come.
Mighan : You’re right. I never looked at it that way. Maybe it’s time for a change.
A few weeks ago, I have finished the flower bloom animation effect, now I need to synthesise the environment in which the flower grows, as I didn’t shoot this shot on location but found an image to use as a background so I need to change the environment of it, I chose to use ps to help me do this.
The ground in this scene was too smooth and didn’t look like a soil environment for flowers to grow in, so I added brown soil with stones to this image
When I finished the background for the first shot, I had a discussion with manos. One of the ideas he offered me was to add an additional top-to-bottom shot in front of my shot, which would better illustrate the story.
So I’m still using ps and have filled this image so as to give the camera plenty of room to move!
In order to make the plants grow more realistically into the real environment, I will use the roto tool in nuke to add shadows to the soil and ground.
Additional Narrative Footage
I’ve been thinking about how to represent the fact that the world has been invaded by zombies, and in addition to a lot of ruins, I think I need to add some dated blood and handprints or patterns or something. Therefore, I downloaded the material from the internet for learning and imported it into nuke for compositing, and added the element of grass, which also symbolises the blossoming of life in no man’s land.
Batch Size: Leave it automatic.
Crop size: Smaller crop Faster to train, is only aware of smaller details
Use or 256 or 512 for better quality.
Log2Lin4：typically used to convert an image from a logarithmic color space to a linear color space. This is a common operation in the compositing process, as the logarithmic color space is better suited for images with a wide dynamic range, while the linear color space is better suited for compositing operations and adjustments.
Check point. You can link other data set with similar content in your training.
Change the RGB channel in the shuffle and connect the mask to the grade, so that the roto of the eye is reflected.
Particle in nuke:
ParticleTurbulence: change the particle direction and add some blur.
Test of head:
Edgedetect：This node helps me extract the edges of an object, but only if the object is alpha-able.
Connect this node with “grade” to adjust the value of gain to change the brightness and color of the edges, which is very suitable for some cool video effects.
IDistort：This node regulates the shape of the border that has been processed, deforming it and giving it a dynamic feel.
Glint：This node is used to help the overall glow and to change the direction of the light source.
First it was done for the character roles in the foreground and then the background. And next the form of processing in different ways is described (about Gizom)
These are two very important nodes.
However, the gizom node can replace these nodes above (edgedetect and grade and so on)
First, I categorized my flower models into flowers, roots, and also leaves. I started with the root part of the flower, I needed to simulate the growth process and curve of the root, so I first used “add” to add its line, and then I added “for each” in order to create the effect for each root.
The node “pathdeform” is very useful for me, the principle is that I first create the point of the rhizome, and then combine the “tube” with the path of the point, so that I can simulate the rhizome of the flower.
The final “foreach” combines to the root of each plant, including when I make leaves, the process is the same.
This was the most challenging step in making the leaves, I wanted them to look more realistic. So I tried to add vellum to make the leaves wiggle randomly, but vellum takes a long time to calculate and it can’t make leaves growing on one side wiggle, so it was a failed method. But this reminds me that I can use the “noise” node.
On top of that, I need to make a fertile soil that incorporates the growth of the flowers and creates some collisions. At the most basic level, the first thing that needs to be done is to create a sphere and change its shape in order to fill this irregular sphere with particles. And set up the colliders (flowers and leaves) so that the leaves get a more realistic feedback when they grow and touch the soil.
Successfully, I did it. It was very not an easy challenge for me, I never learned to code so it was very hard for me and I had to refer to other people’s videos for the codes. But, I would go and try to figure out what the codes meant and which ones the controls were, which was very helpful. Next, I’ll be going for live action videos and 3D tracking in nuke.
At the very beginning, I looked for a background reference image, which is a panoramic view of London
I tried importing the models into maya to adjust their perspective as well as to adjust the camera angle.
This step is easier for me because I downloaded these models from the internet as I wanted to portray a surrealistic future world, so this saves me a lot of time.
Some of the models had problems with the mapping, so I had to import the models into Substance Painter to recreate the UVs and maps, and I was able to deeply feel that I was more proficient with these software compared to the first semester, which was a memorable process for me.
For this step I’m going to use the Arnold sky lights from maya, which will be more realistic as opposed to area lights and spot lights.
I rendered using the tiff format because I found that if I exported to the png format I would have some color formatting issues that were wrong, giving me a serious image.
I then use ps to adjust the tones to make them look more coordinated.
In this step I want to add a hazy mask (roto) behind the entire environment, this will make my scene look more spatial and make the front and back model’s look primary and secondary.
Add clouds – this is a very critical step. It will make the whole scene look more atmospheric!
I used project3d and the card node in conjunction with camera movement to help create a nudge effect in the scene.
Overall, I don’t think matte painting is very difficult, although this is my first go at this method. Probably because I would have used ps to adjust the image before, as well as adding nodes inside of nuke all from what I learned in the first semester. This experience was very useful for me, matte painting allows me to create wide shots that are realistic and difficult to photograph.