Gantt chart & Specification

This week I handed in the official specification for my project (below). As part of the specification I also re-made the Gantt chart I will be using the schedule the project, seeing as my project has changed quite significantly since my initial plan. Most of the major project work in my schedule is pushed back to start after February, after my Masterclass and Innovations projects are handed in. Once they are out of the way I will be able to focus almost all of my time onto this project and power through March and April, in order to hopefully start rendering in May. The lighting and rendering section of the chart runs through the project as I intend to constantly update my lighting with my assets, instead of leaving it to the end of production which could uncover some nasty surprises.

BA (Hons) Computer Visualisation & Animation: Level H

Title: Sleeping Forest

Author: Katharine Hill

1. Introduction

This project was inspired by the research that went into Disney's Tangled. The artists working on the feature aimed to create full CG environments that retained the charm and appeal of traditional hand-painted backgrounds, with the goal that a single frame would look like a painting. With that in mind, this project explores how CG can achieve the same appeal as traditionally painted backgrounds whilst utilizing all the advantages that CG has to offer, in order to create a kind of living illustration. Another part of this project is the exploration between space and narrative. This project explores how a presence or tone can be created and changed to help tell a story through the environment.

2. Aims & Objectives

In order to explore the theme of 'living illustration', study the paintings by Eyvind Earle for Disney's Sleeping Beauty and use his style to create a fully CG environment. The background plates painted by Eyvind Earle were chosen as they have a very unique painterly style which was inspired by pre-Renaissance European art, tapestries, and illuminated manuscripts. These inspirations resulted in a flattened, highly detailed stylization which will provide an interesting challenge to bring into 3D.

• Build a CG environment heavily inspired by Eyvind Earle's paintings.
• Use various dynamic effects to create movement and life.
• Create a short animation in which the camera will traverse the environment.

3. Implementation

THROUGHOUT: Keep lighting and rendering up to date throughout project

1. Research the style of Eyvind Earle and identify the elements to be taken into this project.
2. Create proof of concept to test the pipeline.
3. Plan the composition and progression of shots by creating a storyboard, block test, etc.
4. Model and texture a library of assets of varying LoD.
5. Model and texture a base model. Set dress using the asset library.
6. Implement dynamics to create subtle animation in the grass, leaves, and plants.

-STEPS 1-6 MINIMUM DELIVERABLES-

7. Implement growing thorns animation.
8. Add fluid simulation to create rivers/ponds.

4. Proposed Schedule

Completed tasks:

• Research on Eyvind Earle, his artistic style & influences.
• Identified key stylistic elements to be used.
• Finalised Storyboard.
• Draft block test.
• Preliminary research & tests on growing thorns.
• Modelling & texturing test.
• Created Gant chart for scheduling year.

December

• Block test – used to decide composition & timing.
• Proof of concept – take an asset through the entire pipeline.
• Lighting & rendering tests.

January

• Start modelling asset library (trees, plants, rocks etc.)
• Base/ground modelling & texturing.
• Check lighting & rendering with updated assets.

February

• Finish all modelling.
• Start texturing.
• Check lighting & rendering with updated assets.

March

• Finish all texturing.
• Set dress base model using asset library.
• Implement grass & leaf dynamics.
• Start implementing growing thorns.
• Check lighting & rendering with updated assets.

April

• Finish thorns.
• Implement water.
• Matte paintings.
• Finalise lighting.

May

• Rendering.
• Compositing.
• Report writing.

June

• Finalise report.
• Breakdown reel.

Proof of concept

I've started working on a proof of concept asset, which I will take through the entire pipeline in order to refine style choices and test the pipeline. I chose the cliff/rocky outcrop seen in the background of this frame as it's relatively simple but still provides a good variety of shapes and textures.

So far, I have modeled and textured the asset, although this isn't the final version as it still needs several tweaks. The modelling flagged up several issues that I had not previously thought of. To replicate the strong geometric shapes in the original image I extruded lots of panels from the model which creates lots of sharp edges. At first, I tried to make this work in smoothed mode (assuming this would look better) by adding lots of extra edge loops, however this didn't work very well and ended up making the mesh pretty messy. This render is actually in unsmoothed mode; I think for this aesthetic unsmoothed works well although I might go back and do some tests with bevelled edges to make a slightly softer line. The other main issue in modelling is the border between the cliff and the grass, which looks very harsh and unnatural. To solve this I think I need to make the surface of the grass more curved so that it hits the side of the cliff at a similar angle, not the harsh 90 degree intersection as it is now.

Texturing was a relatively simple process which combined fractal layers, photo reference, and some hand painting. I experimented with painting in cuts and lines in the front section of the cliff and using that as a bump map, which could work nicely. I need to spend more time looking into the rendering of bump and displacement maps to really make the most of these features.

Next, I need to make adjustments to the model and textures then move on the next stages. I want to try covering the ground in a grass bed using pfx which I think will make it looks much more realistic, as right now it looks like a low-res game asset. After that I will move on the lighting and rendering which is my least experienced area, so this will be a good way to practice before tackling the whole scene.

Storyboards

My storyboard has gone through several iterations since my original black test. The most dramatic change is that I am no longer building the indoor scene. Instead, I'm focusing on the change of tone in the scene as the thorns grow. 

I drew lots of thumbnails to sketch out ideas then put them together to form the story progression.

version 1: The camera follows the progression of the thorns as they overtake the forest.

Version 2: The camera follows a leaf falling to introduce the scene first, and then the thorns enter and start to grow. 

Version 3: Includes more back-and-forth shots as the thorns are introduced to create suspense. Also more focus on close-up shots mixed with long shots to create variety.

Thorn tests

One of the main challenges for my project is the generation of growing thorns. My initial plan splits the problem into 3 sections; background, midground, and foreground. Each level of detail presents it's own set of problems.

Background:

The thorns shown in the background require a low level of detail and won't need to be animated. A simple solution to this is the use of alpha planes and matte paintings which are quick and easy, allowing me to focus my attention on the higher detail assets. I put together a very quick and messy example of what the use of alpha planes could look like. By rotating and scaling this single image (below) I filled a large area with thorns without much obvious duplication

Midground:

This will be where the bulk of the thorn generation takes place. These thorns require animation but still a relatively low level of detail, and should work fine with a simple shader as opposed to painted textured. My first test was using Maya Paint Effects, in which the user defines curves which the vines climb up and a series of parameters which control the final geometry. This technique gives me a pretty high level of control and decent results without much time being invested, once the original parameters are set up. However, there are some issues with geometry intersection and it can be difficult to get the right aesthetic. Also, as I have to define each curve separately and define all the keyframes for animation it would be a very time consuming option for covering large areas.

The other technique I intend to implement is a script which will procedurally generate the thorns using l-systems combined with vector tropisms to define the growth patterns. This will allow me to simply select a growth point from which the thorns will grow, attracted to nearby geometry. This will automatically cover large surfaces without my having to define curves or animation keys. The user will be able to define growth parameters and the general direction of growth but the results will be largely controlled by the system. I will be working on this for my masterclass project so all of the research towards it will be found there.

In contrast to the pfx option, the script offers a quick solution for covering large areas, but without the high level of control. The best solution may be a combination of the two, using the script wherever possible but resorting to pfx whenever a higher level of control is needed.

Foreground:

There will be a couple of very close up shots in my animation which will require highly detailed and controlled sections of thorns. For these I intend to use the traditional method of modelling, texturing, rigging, and key frame animating pieces of geometry. To start, I will use the thorn wall from Maleficant as a reference for a design and animation style of these pieces.