Conception
It started the very moment I decided to put in practice what I learnt about 3D Studio Max software.

I was searching for an idea strong enough to lead me to the making of a short film equally rich with technical and artistic expressions. Having some experience in animation, I realized that the subject I should choose would have to be particularly strong and rich so to keep me going for months.

I had previously worked on short animations such as a character, inspired by the painter Matisse, that I made slide on water. I liked it and felt like using once again the idea of reflection by transposing it in a landscape made in a realistic way.

I then thought about a man living on a lake side.

My first idea about it, was to do an advertising for a weather forecast adviser.
A bearded man was drinking tea in his wooden house, with trees all around.
He was watching the rain pouring down when suddenly he saw a fairy flying over the nature changing the rainy day in a beautiful sunny one.

Later I gave up this idea realizing that probably they must have there own staff to realize there adverts.
The worst that could happen to me would be to have my idea stolen.
I then decided to extend my previous idea to the story of a man facing a quite grumpy mother nature.

It was quite exciting and a real technical challenge for me to recreate totally in 3D, nature, a character and a house seen for the inside as well as the outside.
My principal intention was to have a 3D result quite different from what usually shown.
Instead of geometry, I gave most of my attention to texture and did a kind of hand made job of it requiring more technics than plugins. Of course at that stage I didn't realize all the problems that were awaiting me.

I have located the story in Siberia, because it's an under populated wild land with harsh winters and very warm summers.

I have collected information in lending libraries in Paris searching for links with my story (Siberian landscapes, russian architecture, interior and exterior, clothes etc...)

Then I layout the first story board and outlined the character using NURBS and Metaballs (which both failed)

Creating the film sets

I have started modeling from the inside of the house then I took pictures of the furniture's textures (99% of the texture in my film come from my own pictures)

I then went walking through Paris looking for fine looking, wood, metal and other textures, then I made most of my holidays in the mountains to complete my textures collection.

At the time I had no numeric material at my disposal.
So I used a Pentax Reflex 35mm with a 35x70 focal then I printed my pictures on paper so to scanned them (by the way: my father is photographer)

Then I worked on the nature Surrounding the house, particularly on the lake, the boat, and the house seen from outside.

At the beginning I thought of creating trees and reflection in water using the pictures I had taken. Later I realized that for the last sequence of the film with an up lifted movement over the clouds, I needed trees in perspective that a simple photo couldn't furnish.
Also to have better effects on the lake, true reflection was necessary.

Music

The story I imagined took place at the beginning of the 20th century, therefore I searched for music parts composed in this same period.
I had a piece of luck listening to Sergueï Rachmaninov's symphony "Isle of the dead".
The first movement had exactly the dramatic and gloomy touch I was looking for.

The very same day, I discovered three orthodox hymns equally composed by the same great Russian composer, perfectly fitting my need.

Taking a new start from these musics, I draw a second story board (see it in the part comparing story board, animatic, final cut)
Dimitri

Until then I met with a lot of problems founding the right techniques to model the character.

I was looking for a professional technique that would allow me to model, animate, easily Dimitri.
Finding theses items in books or on the web is a difficult enterprise.

Then came Imagina 2000 (half in Monaco, half in Paris).
For the second time I visited this Festival while in Paris.
I questioned a few computer graphic designers who had just been awarded.
They all advised me to use the polygonal modeling.
After that, I put in practice these advices and finally decided upon the polygonal modeling with mesh simplified by extrusion of edges (create faces from edges)

Using this technique was so freeing. It made me gain time while modeling by using a simplified mesh; also it eased the animation with character studio because the mesh is only subdivided before the render.(with a meshsmooth)

Everybody uses this technique but of course at the time, I didn't know.

From the start, I wanted the character to be a success.
It's a main step and one easy to miss.
Most basically because one knows what a human being looks like and can easily tell when something is wrong in his virtual reproduction.

I wanted to give realism to the basic material of Dimitri's head. It took me a lot of time to catch and render the expressiveness of a human face.

As you can see, the hat has disappeared in the animation (I changed my mind)

In fact as it was attached to the skeleton I couldn't suppress it so I only had to hide it during the render.

I know! He has a crooked foot and a shoe missing.
A big problem still remained: The Modeling of the character's face.

I knew that modeling a face by extruding edges, would take weeks.
After searching through many files.
I found "morpher.max" a head example, furnished by Max.

I created my character's head, by distorting "morpher"
I used "Hair" plugin for hair and beard.

My father inspired me Dimitri.
I took pictures of him to obtain textures and scales so to model the body.
I didn't try to clone my father.
I only used him as a basis to create the character.

Animatic

Then, from the story board and the music, I realized the animatic.

(see comparative: story board, animatic, final cut and in video)

Animatic resolves most of the problems met while making a 3D film.

First of all, it helps making the 3D of the story board with cameras and sets.
Then the animatic, by creating a simplified representation of the elements, frees from the constraint linked to the massiveness of a scene using quite complex polygons and textures and then helps animating, editing and adding the sound track.
The final cut and the animatic being at the same scale, I was able to import cameras and use the animatic as a reference.

In the same time I edited my film in relation to the music and built up the whole story.
As I had built up the story board from the music, I was obliged to create very precise sequences so to respect the music pieces, therefore I had to add shots so the public couldn't be bored.

It took me a month to build the story.
Every evening, I launched the render of my work and next morning, I checked to see if it was OK.

Nature

Representing nature in 3D in a realistic way needs a lot of observation so to understand it well.

Personally I drew most of the observations and elements required for my story while staying in the mountains: lakes, larches, storms, wind...
There, in the mountains I learnt the main rule for realism: Diversity
And that rule is OK for many things
For instance:

-As far as texture is concerned, just avoid repetition. This needs a maximum of different textures per area.
-Ditto for the lighting: you need a lot of different spots.
-For the making of my film I always followed the same rules on the diversity of textures, forms, animations for each element of it: grass, clouds, trees, the lake.

So to obtain that diversity, one has to battle with tools based on mathematics. When used plainly, they repeat exactly the same action.
I therefore tried to break this repetition system by varying a max.
Alas! The computer tool has his limits...

As I couldn't find a Forest plugin, I decided to create my own system so to design trees.
I started for "Clay Studio" plugin to draw the trunk and the branches.
As foliage was concerned, I decided to create groups of thorn textures in 3D which, after render, served as mapping to simple faces, placed on 3D trunks.

I drew 7 different 3D larches, put them on automatic rotation, changed there colours and obtained 336 tree textures while doing so.

I only had, then, to apply these textures on hundred's of faces to create a forest, each face containing a tree texture (trees having referenced and instances modifier for sizes and different movements)

 

To realize the clouds, i did a few tests, without success. (here also I had a hell of a time finding the right techniques)
I finally had luck (and needed a lot) to find (again) a helpful plugin! "Afterburn".
It took me a lot of time to handle it but it was worth it.

The grass needed an other plugin...
I could have done a few tufts but couldn't have animated thousand of them. As for creating the different sizes and colours of the blades so to be very close to the natural aspect, I had to realize a few dozens of implantation areas, each of them containing several different species of blades.

To create the lake, I realized a mesh surface, using the modifier "Noise" for animation and a "displacement" from the picture of a lake so to catch the shape of the waves.
The reflections are calculated by raytracing and this, believe me, needed patience.

Sound Capture

As for textures, I started by searching for material on the Web and in CD's.
It was a complete wash up as, ready made sound samples are hopeless.
So, recording being one of my hobbies, I decided to go and capture my own sounds.

Beside, I was thrilled with the idea of walking through mount and hills catching, here the rustle of the wind in the larches, there the clash following a lightning when the storm is roaring through the mountains (by the way, all the storm sounds in Dimitri are real. I have captured them during a one storm on summer)
By August 2001, I had collected most of the sounds I needed.
I created the very few things left with a software sound processing.
The equipment I used for the sound capture, was a MZR30 Sony minidisc recording and an ECM-MS907 Sony microphone.

 

Production

After displaying the set and starting the animation of Dimitri rowing, I paused for I had a great problem to resolve:
Should I or should I not launch the final render?
Indeed, each picture needing an hour and a half calculation, I had to be sure to take the right decision.

Once I had decided that my work was finished ant that I couldn't do better, the making of the film gained in speed.

First, I linked up the animation and render of the sequences in a chronological way; then, I tried to finish the animation of the sequences requiring the longest calculating (such as the lake sequences) do to be sure I could furnish my 3 PC's with continuous work, up to the end of the project.

One of the difficulties was the character's animation.
I didn't know a thing about it.
I did my best but the tool very often lacked flexibility and preciseness.
This difficulty was counterbalanced by the quickness of the animation tests (thanks to the simplified mesh!)

What gave me a lot of problems was the living together of so many plugins. (rendering phase)
When it came to post production, I had to correct many bugs and mistakes.
A few sequences such as the grass sequences bugged during the night, in most uncertain ways.
I had to resign myself and calculate these during the day, so to keep an eye on them.
Happily, I worked in layers, which stopped me from having to start again. (see in post production)

Despite of luck, technique, the key to succeed in a project that complex (for a single person) is organization.
The last 6 months, I prepared a very precise planning for each of the 3 PC's I was using, each of them having his daily and nightly tasks, well prepared.
Constantly I reorganized the planning taking into account the excesses or the advances.
Thanks to this planning, I was able to organize the work on the 3 computers, so they would work continuously and give me the layers I immediately integrate into "Combustion" and "Premiere" .

I managed to work on a computer while making render tests on the others, and so, loosing as less time as possible waiting for the end of a render.

Post production

While visiting the "SATIS" 2000 exhibition in Paris, I discovered the software "Combustion" and realized that, using it, I would be able to work with layers keeping 3D's informations (such as depth, opacity, textures etc... of objects) on 2D pictures.

It encouraged me to realize a short film with quite heavy geometry and texture in it.
From that time on, I could render separately different sequences, assemble them with "Combustion" and still I could add special effects (depth of field, lens flare, fog, colour grading) by selecting, by layers, or by object.
In a word, "Combustion" gave me that plus of ease that counterbalances 3D tools heaviness.

Thanks to this marvellous software, I was able to work on again a great piece of the aesthetic part aspect of the film.

After compiling layers and adding effects, I transferred all of it on "Premiere" then I edited the sequences and added the soundtrack (music, plus, sound samples)

I did all the sound track editing with Premiere and did appreciate it's speed and simplicity.

Conclusion

This film couldn't have been made two years earlier: Techniques for sofware such as 3D studio max weren't avaible then.

When I mention techniques, I sometimes mean software techniques such as plugins, other times I'm talking
about the equipment (hard drive, memory, processor...)

That's all for now folks! I've nothing left to tell you about Dimitri.

Hope I didn't bore you with my making of....

Figures:

My computers have progressed from a simple pentium 233 with 128 Mo of RAM to 3 AMD (2 of 1,3 ghz and 1 of 1,6 ghz) with each 1,5 Go of RAM.

The rendering has progressed from November 2001 to October 2002 and has generated 58 Gigas of RPF files contained in 39.424 used files.

The film totalizes more than 15500 hours of render (I added up the timings of three computers to do the estimation)

From October 2001 to April 2002 with 2 PC (more than 6500 hours)
From May to October 2002 with 3 PC (more than 9000 hours)

 

z