Decisions on what to be included in the final presentation

As exams is coming up and I need to set aside a lot of time for school projects, I may not able to finish everything I promised inside the previz. So the plan is I will focus on only three shots, which is 

1. POV from the robot. 

2. Robot walking on the ground 

3. A close up of the robot's eyes when he is blinking(or something else)

The good news is I have finished the 1st shot completely. For the 2nd shot I am in the rendering and comp state. and 3rd one should be a really really simple shot so I do not need to spend a lot of time working on this. 

That's today's update. =)

Shot0 in the making

Humm...Have not really cultivated the habit of recording all my process, so my post is a bit on and off. 

So there goes the latest update of what I am working on right now.

I mentioned that I am going to add a shot of Chuchu's looking out from his eyes. With the help from my BF, we came out with this mock up:

And for the last 2 weeks I was working on all the small little elements and animations. For now, all those little animated elements are done, and I need to comp them together to achieve an effect that is similar or even better than the above mock up. 

The scanning effect is done, and I am currently working on the 2 panels above the bird.  

I did the panels by matching move the camera and create a 3D scene. Those panels are cards inside the 3D scene. I added noise function to make them flowing in air. 

However the lines collecting the panel with the bird is not so straight forward. My current plan is to replace the graphics on the panel with something easy to track and use that track info for one side of my line, ends at the panel. The other end need to be manually animated. It seems no other way round. =/

 Here is my current comp graph for shot0. I believe comping shot1 will be much more simpler though.

HUD in the making

Was doing some sort of charts with Flash. And it is really time consuming to plan draw and animate. Sigh...

More and more HUD-scanning and some graphics elements

Example of some graphics element that may be used, I made it in Flash:

 Here is a simple scan effect. Basically, I modeled out the whole scene and rendered out a depth path.

The value of z is used in a expression to assign values to each pixels.

Here is the detail of the expression:

variables: 

head(user variable, used for feathering of the head), 

depth(user variable, use to control the scan depth), 

tail(user variable, used for feathering of the tail), 

r(red channel RGBA)

expression: r<=depth? lerp(0,1,(r-depth+tail)/tail):lerp(0,1,(depth+head-r)/head)

Chuchu HUD Test02

The problem for this test is there is little sense of volume, and also I animated the particles going around the model, it is not obvious when Chuchu is turning. (Especially when the rotation direction of the whole model is the same with the rotation direction of the particles)

Chuchu's Font

I designed the font on my way home today. I intend to give a squarish look to the font.

 Some of then can be easily guessed from the shape, such as "I". 

After I create the font, I uses the font to "encode" a paragraph on Wiki, and this is the result:

Houdini Modeling and Particle Effect

Since I uses "creep" node to achieve the particle effect, I need to provide NURBS model instead of polygon. Unlick Chuchu's body which I can use ray projection to project a spherical NURBS surface on to the model, I have to do a rough model of the leg in Houdini. 

chuchu's HUD test

(really short-17frames)

final comp:

Long-time-no-post and lighting reboot

a rendering test shows that the shadows of the scene is not accurate. All the shadows are generated by some rough modelling of the ceiling and pillars with a directional light. It does not account for shadow of the trees.

Therefore this time I try to generate a more accurate shadow by using the clip. With a lot of trial and errors (spend a whole day solving small problems) and finally get things done. Here are the basic ideas:

1. a texture is rendered out on the ground plane based on the back plate.

2. Extract the shadow inside NUKE

3. Map the B&W texture onto the ground

4. Render out the shadow using a orthographic camera. Need to match the direction of the camera with the directional light.

5. Map the above texture onto the opacity channel of the texture of a plane ( need to match the size of the plane with the camera size) which is put between the light and the scene.

6. more accurate shadows of the trees!

Texture and Test animation with texture

Chuchu animation playblast

Did the jumps with dogs in mind =)

Animatics

This is the animatics for the first part of the story. This part will be the so called "fall back plan". If time permits, will continue with the second part which is the collapse of the building.

Texture Test

Using color mixer and light surface to achieve a effect close to what I am looking for.

Color test

Chuchu eye mockup

 LED eyes

Puss in Boots eyes

Obviously the LED eyes looks much better. It may be further animated to mimic the "scan line" and blink effect 

Lighting and material test

Problem: double shadowing.
Proposed solution: paint shadow mask for the environment and project into 3D space. use the matchmove information and export the shadow channel out. Use Chuchu's shadow channel minus the environment shadow channel.

HDR and Panorama 04

Used Stitcher instead of Photoshop. Also tried to map it on a sphere in Nuke and looks perfect. =)

HDR and Panorama 03

Shot another round of photos this morning. This is the final exr after touching up

HDR and Panorama02

Since we cannot afford those expensive equipment, a few DIY tools are made to improve accuracy. 

Such as this angle meter:

After stitching the exr files together:

Still needs a lot of touch up though. But it looks fairly decent.

HDR and Panoramas

I experimented with a view stitching program include Pos Panorama Pro and find it is very easy to stitching things in 1 direction (either horizontally for vertically) but not 2. There are some of my experiments with a free image stitching software:

It is a rather painful process. Firstly, I do not have a DSLR and needs to take more pictures to compensate that. Secondly, I do not have a tool to measure the rotation of the camera. To be save, more pictures are taken. The test above is by stitching 20 photos(!!!). And I also need to take 30 degree up & 60 degree up and 30 degree down & 60 degree down. This makes up to 100*4 exposure = 400 picures!!!!

That makes my end up buying a fisheye for my poor little camera. I get it from Sim Lim for 150 bucks. And this reduced the amount of works I need to do by 5-folds. =P

I dead another test about the HDR workflow tonight. There are some new finding.

1. Do stitching after merge HDR it is impossible to stitch those over/under exposed images together(no feature for the computer to match). 
2. You can unwrap the fisheye image in PS and make it a action for batch processing. You need to crop out the sphere, and ensure your image is square before applying a lens correction
3. There is no single-click lens correction for my fisheye in PS. I used a 8mm lens correction plus some custom adjustment.

after stitching:

Did not adjust the exposure but you get the idea. The image have irregular edges at top and bottom. That seems to be normal after reading this site:Creating HDR Panoramas. So I am not alone. All I need to do is some post processing work described by Farbspiel.

Incidentally, I found a very very very (xn) useful site: HDR tips from Professor Kirt Witte . It talks about how to setup and take HDR images so as to get the best results. 

When Chuchu open up his mouse...

The original thoughts is make chuchu's cover opens up as a whole when he starts eating. However, it looks pretty clique and ugly. (refer to the character design post below)

therefore it maybe better to cut his cover into pieces and move then individually...like this:

here are some thoughts on how the cover should be divided.

Character Design

here is the final character design I draw after many mocks (cannot find the mock files. might have been 

accidentally deleted.)

 still investigating how to structure his inner parts.

wanted to find inspiration from fractal pattern

here goes some of my reference images

current modeling stage 

Shotsheet

Production Schedule

As the actually time spent on each component may vary, this is only a guideline rather than a strict schedule.

Long long story board

And here goes the shorter version (Will do the long version only if I have time). The plan doing it one scene after another.  Something and a mix of waterfall and iterative pipeline.

PigSoul Trivia

Writing production blog reminds me of another graphics design module I took in NUS when we need to demonstrate "profound thinking process", and show how much blood and sweat we put in the final design.

Anyway PigSoul is inherited from the team name I used for other production. I always like to put a logo of some sort at the beginning (or end) of my video work.

This is the so called original logo I used for some other project. I made this very quickly in Flash.

However for this project I am going to use hexagons instead of square, so as to stay align with the style which will be used for screen interface that is going to appear in this project. I did a view mocks in Photoshop and finality came out with this one:

Then I made a procedure hexagon grade in Houdini. I am able to adjust the grid size and number of hexagon per row.

And after transferring the design into Houdini:

I plan to animate the appearance of the individual hexagon using their bounding box positions and I have yet designed how they are going to appear. Just a simple 90 degree rotation? Dropping down from above? (Someone: NOOOOO! You should be working on the actually project! No spending time on this stupid logo that no one is going to care!!)