Week_01_Notes & Essay

This week I finished the eye rig for the monster and build the scene for our final shot in maya.

I started to rig the eye after Ziyin finished the modeling of it. The set of eye rig is not conplicated, which has eyelid controls, IK&FK eye controls and global controls. I added an offset group beyond the eyelid skeleton so that it can control the offset of eyelid when eye direction changed. I also added several blendshapes to correct the wrong deformations in this rig. Since the time is limited, I only added about ten blendshapes. The eye rig can be more flexible and realistic if I add more blendshapes on it.

A test for the eye rig:

I also rigged the pupil so that it can be zoom in or out:

After finished the rig, I started to build a simple scene for the final shot.

I build a simple background (cave) by using several noise nodes in Houdini. After that I import it into maya and use quixel bridge assets to build the foreground and midscene. After that I import all the rigs with textures inside this scene to do some layout and lighting. I want the monster eye hiding inside the darkness and character’s flashlight shooting towards it. The atmosphere must have some fog/smoke so that it can show the direction and shape of the light (Tyndall effect)

Here is a rendering test for this shot:

This week I finished character rigs for project01.

Since the time is quite limited, I choose to use ADV rig plugin as the tool to rig these characters. The setting of the rig is quite simple, I just need to make sure that the rig has a good skin weight and can be exported to game engines.

The painting of skin weight took me some time to finish. Our character has too many props on his body so that some of the skin weights are hard to deal with, especially the armpit and neck areas. I finally solved problems and finished the body rig.

Here is a walking test in ADV. Only a little bit of mesh crossing problem can be seen in the test, which is just fine:

Another character is the main character’s partner. Both of these characters wore same clothes. Therefore, I asked Zhenzhong to just replace the second character’s head so that I can easily copy the skin weight from the main character to the second one.

It took me 2-3 days to finish these characters’ rig. The next part is to finish the monster’s eye rig. What’s more, I will export the environment to maya this week so that we can start to make our animations and cameras.

This week I finished the third scene in our trailer: the restricted area scene. This is the first version of it and will be adjusted in the next week.

The whole process of making the restricted scene is quite same as the other two scenes. Since we have already find the proper references and finished the storyboard, I can focus more on the layout and lighting of the scene.

I firstly find some references for this scene and try to feel the atmosphere inside of them:

Most of these scenes show the same two features: a little bit of artificial traces and foggy air.

I listed several compotents in this restricted area scene:

• Foreground: Tree stump, wreckage, barbed wire, grass, branches, rocks.
• Mid—scene: Abandoned watch tower, trees, rocks, branches.
• Background: Mountains, trees.

Based on this list, I gathered some assets and made part of props by myself (barbed wire, mountains, rocks). I then layout the whole scene in unreal engine, trying to provide a sense of desolation and stillness in it.

Here is the final result:

Next week I will optimize these three scenes, especially the forest scene. I’ll add more details to the foreground and adjust lights.

Scene Rendering

After adjusting the quality of the destruction and smoke sim, I rendered the scene and composited it in Nuke this week. Then I added some sound effects in it. Here is the final result:

Procedual Saliva HDA

After that I started to make another visual effect, which is a procedual saliva digital asset.

Saliva is an important element of realistic creatures. For example, we can find creepy saliva effects in film Alien (1979) and The Thing (1982), which brings a great sense of realisim and horror.

I firstly find some saliva reference:

As we can see, saliva can be separate into 3 types:

• Filamentous dropping saliva
• Punctual dropping saliva
• Filamentous dangling saliva

Filamentous dropping saliva and punctual dropping saliva can be easily simulated by using high viscosity fluid sim and surface tension. However, filamentous dangling saliva can not be easily achieved by using fluid sim. It is more like a fluid-like cloth simulation. What’s more, the shape of saliva has some regular patterns: both sides of saliva is thick and the middle of it is thin. Such kind of feature shows a sense of high viscosity.

I used a combination of vellum cloth and VDB to sim filamentous dangling saliva. The main idea of it is to build up lines using VEX, and then use vellum solver to sim these lines. After that I turned lines into particles and then turn them into VDB, and finally convert them back to polygons.

This is a procedual method of building saliva. Therefore, I can easily build saliva on different creature models by using the HDA. What’s more, I can change the shape and numbers of saliva by using customed attributes.

I did a simulation to test the result. Most part of it works well. However, a bending can be easily found on the saliva when the the mouth is closing. This can be a problem because it makes the saliva more like a cloth rather than a fluid. I think the problem comes from the vellum simulation. I need to add more elasticity to the sim to prevent this kind of results.

After that I optimized the volume of saliva, here is the result:

A rendering for the saliva:

A movement test for the saliva:

This week I finished the cave scene in project01.

I used Houdini to build up the frame of cave and background mountains first, which can easily control the shape of the cave. After that I sent it to unreal engine to build up other parts of the environment.

I used the same techniques that I have tested in the last scene to build the scene. This time I mostly focus on details and the layout of the scene. Based on the storyboard and refereneces, the cave should be deep enough to show a sense of mystery. What’s more, the scale of the cave should be large enough to show a better composition of the whole shot.

I added three layers of textures on the landscape to make sure there are enough details on the ground. Then I added some branches and small rocks to push the details to another level.

Below is the test shot for the scene. The LOD of the landscape hasn’t been removed successfully. I will figure it out next week.

This week I started to build our scenes. After discussing with Yaozhang, I choose to build the forest scene, cave scene and the restricted area scene.

There are about 6-7 scenes that we need to build for the whole trailer. The forest scene is one of the most important scenes and it is a little bit easier to be approached. Therefore, I made a pre-version of the forest to test the result.

The aspect ratio that we chose is 2.4:1, which can bring a great sense of cinematic to the shots. I gathered some textures and models to build up the scene and lightened the scene by myself. The landscape is made by unreal landscape tools and houdini, and shaded with a blended material. It is still the first version of the scene and the quality can be improved later.

I animated a camera to test the shot. The mp4 smoke sequence works not well at the beginning. I changed the media texture to exr files and solved the problem.

The testing result is fine, which doesn’t take much time to be built and rendered. However, I still think the lighting can be improved and the forest is still not dynamic enough. One of the biggest problem is the size of the file and scene optimization, which are some main issues that I need to fix in next week.

From this week I decide to use destruction and pyro techniques to make a more interesting and technicial shot. The whole shot should be finished in 1 to 2 weeks.

What I want to make is a scene in which a creepy monster/creature breaks through the side wall and running towards the camera. The shot will lasts about 7-10 seconds, which focus on the dynamic effects of buildings and creature and the atmosphere of the whole scene.

I decide to build up a corridor for this shot. I firstly did a research on horror corridors on the internet and gathered some lighting references. After that I started to build the frame of the scene. What I need to do is to build part of the corridor and duplicate it forward to make whole corridor. Each part of the corridor shares same UV and textures.

Since I don’t have enough to focusing on creature modeling and rigging, I searched on the internet and found an alien rigged model, which is pretty nice(I can use it to test some CFX later). The texture of it is not at a high quality (no SSS, and low res), maybe I’ll replace it later.

After that I baked normal textures in substance painter. Then I made other textures by combining quixel scaned textures and baked textures together in substance painter. The uv had some problem at the beginning and was solved then quickly.

I then placed the creature into the scene and render a shot with redshift. The whole atmosphere is quite close to what I want. What I need to do is to add a little bit more details on the ground and ceiling. The next part is to animate the creature (it is an alien model acturally) to make the shot more horrible and set up the wall for destructions. The animation is quite important because it drives the range of whole destruction. It should be powerful enough to make the VFX looks better.

After that I did a rough animation for the creature and set the position of the camera.

I import all the environment and animation to Houdini to make the RBD simulations and pyro sim.

Below is a test for the scene. The smoke sim is fine. However, I think the constraints of each fractured pieces still have some problem. I ‘ll add a hard constraints to test the result later.

Here is a rendered one testing the smoke and other textures:

This week we had the second meeting and finally selected some key shots for our film. Then Zhenzhong finds some music for the film and made a short cut for our shots as a reference for our future works. Most of these shots can be finished in UnrealEngine, and few of them (for example, close-ups) need to be done in Maya and Arnold.

After that we started to make assets. Me and Yaozhang will focusing on the environment of shots, while Zhenzhong and Ziyin will focusing on the character and props.

I decide to to some research on UnrealEngine and find out which techniques that we need in these shots. I will start to layout the environment after I fully prepared the UnrealEngine.

What may covered in our shot is forest layout, terrain layout, texturing, lighting and some simple effect like dust or smoke. I learned how to make fake smoke in Unreal Engine by using exr sequences and customed textures. With the combination of fake smoke and UE build-in height fog, we can easily create the atmosphere that we want.

The second sim is a gas explosion. I want the explosion to expand fast in first few frames and then keep rising to form a good shape. The emission should be pretty hard at first and disappear slowly.

I firstly made a rough sim to test parameters and the range of explosion. The first test has a lot of problems: bad shapes and bad emission, the speed of buoyancy is not satisfied enough.

After that I adjust the sim to get a better result. I also add a shockwave to make the ground sim more interesting and interactive.

I then build a scene and an animated camera to fit with the explosion. At first I tried to render it with arnold. However, it takes about 20min to finish one frame, which is way slower than I expected. Therefore, I turned to render it with redshift, which is a faster GPU render engine.

The volume shader in redshift is not as good as the one in arnold, which took me a lot of time to finish the shading. After rendering, I composited it in Nuke, adding vignette and grains on the shot.

Below is the compositing breakdown:

Zhenzhong told me that the color of the explosion may have some problem. I also think that the color of flame is too bright. I will test this short again to make it looks better.

In the first week our team focusing on the background research and storyboard.

We changed part of our script and make it shorter than the old version. Due to the limited time, we finally decide to make a movie trailer, which will last about 90 seconds. Therefore, we can pay more attention on our shot rather than rendering.

I firstly did a research on our characters and the environment on pinterest to gather some ideas. What I focused on is the suit of our character and the key color of environment.

After that I drew the storyboard based on the key shots in our script. I use a combination of drawing and photoshoping to create my storyboard, which can demonstrate the shot and the atmosphere that I want much clearer.

Our team will have a meeting next week to decide which part of the storyboard and concept fit our requirement. After that we will start to design and make our characters and environments.