"Once upon a time"-Billiard

Motivational Image

First of all, when I heard the theme of this year's final project I was already thinking about fairytale. My thirst thought was to try out something with an old castle and a dragon, but I realized that this might be too hard to achieve a good looking image at the end. So I decided to fit multiple "once upon a time ..." stories into one picture. My idea is to put them all together in a billiard game. So I do not want to simply render the picture from the billiard table as above, but plug in things from the stories and turn them into billiard balls. For example, I want to model one ball as an apple which is representing Snow-White. Another one will be the golden ball known from the Frog King. There will be of course more different balls all representing another story. I could not find a motivational picture which really shows my idea, but I think with the image and the description provided the link to the theme is clarified.

Features to implement:

Here is my proposed feature list I would like to implement. After the list I describe why I want to implement this feature and what for it is used for in my final image. I distinguish between the features which I want to be graded as well the features I am trying to do additionally, but which should not be graded.

Graded

Following features are summing up to 60 points.

• 5 Points
  • Modeling Meshes
• 10 Points
  • Motion Blur
• 15 Points
  • Environment Map Emitter
• 30 Points
  • Heterogeneous Volumetric Participating Media

Not graded - Additionally for the final image

• 5 Points
  • Depth of Field
  • Images as Textures
  • Bump Mapping
  • Simple Extra BSDF
  • Transmittance in Homogeneous Media (Beer's Law)
  • Simple Extra Emitters (spotlight, directional light, ...)
• 10 Points
  • -
• 15 Points
  • -
• 30 Points
  • -

The following pictures are all hyperlinks to the original picture found on the internet. It is one option to test the implemented features against this pictures. So the idea is to render maybe an equal scene (if the pictures are scenes) and compare the result with the original picture.

Modelling Meshes

For the fairytale "Beauty and the Beast" I want to model a rose which is in a glass ball. Additionally, for Snow White I will model an apple. Maybe there will be further models, as the queue if needed.

Validation: To validate the models, I will rely on real world objects. For this I will use pictures of the real world or even make my own pictures. Then, the models can be compared with those images.

Motion Blur

To make the picture not looking to static, I want have at least one ball which is moving. Therefore I want to implement this feature.

Validation: To validate this feature, I want to render the start and the end position in separate images. Additionally, one rendering will be with the motion blur with the same motion. In the motion-blur pictures, both positions (and of course in between) have to be represented. This can be checked with simple image overlays to see that those are contained. However, they will not be as strong present as on the single-images.

Environment Map Emitter

In order to make my scene look more realistic, I want to implement this emitter type. I do not know yet if the scene will take place in a billiard bar or maybe in a more fairytale surrounding.

Validation: I would like to discuss the possible test-methods with you closer in the exercise session. My first thought was about rendering the same scene with an existing renderer (as Mitsuba- or Indigo-renderer for example, of course not by copying code, but executing and render it). Then those images can be compared.

Heterogeneous Volumetric Participating Media

I would like to make a fortune-teller-ball, which can be associated with many different fairytales. To achieve this, I would like to render some kind of smoke in the ball which lights a little bit blueish (the color might change once I see the final rendering)

Validation: I would like to discuss the possible test-methods with you closer in the exercise session. My first thought was about rendering the same scene with an existing renderer (as Mitsuba- or Indigo-renderer for example, of course not by copying code, but executing and render it). Then those images can be compared.

Depth of Field

To make the scene more authentic, I want to implement also the depth of field. With this I want to achieve that the ball more far away should be blurred a little bit, and therefore give the scene some depth.

Validation: To validate this feature, I will rely on real photographs trying to achieve the same effects.

Images as Textures

To immediately let the scene look more realistic, the balls should have a real texture and not only a single color given by the albedo.

Validation: I think for this feature there is no huge verification needed, since I can clearly check with a sphere or any other primitive shape whether the texture is aligned and projected properly.

Bump Mapping

To give the balls some structure, at least for some special balls, the bump mapping is used.

Validation: To test this feature I will download already predefined bump-textures and will apply these to some objects (sphere, box). With the results provided on the website where I downloaded the images, I should be able to render the same effect within my renderer.

Simple Extra BSDF

If I can still find some time, I would like to implement some extra BSDF for some of the balls. Because I do not want this feature graded, I also did not think too much about what kind of BSDF I would need more to achieve better results.

Validation: To test this feature I think it depends on the extra BSDF I'm going to implement.

Transmittance in Homogeneous Media (Beer's Law)

I can imagine also to use the Beer's Law for one ball which contains some sort of poisoned wine. Again, since this is an additional feature which I do not want to be graded, I did not yet think deeply about this.

Validation: To test this feature, I will try out again the Mitsuba-renderer.

Simple Extra Emitters (spotlight, directional light, ...)

I have already implemented the spotlight during the first programming assignment. Maybe I'm going to implement also the directional light emitter.

Validation: I will try to generate a valid xml-test file as it already exists for the direct-light.

Additional

I wanted to ask if there is a chance to join the competition even though I can not be present at the 23.12.16 (oh the lovely Swiss Army....). Of course I need to have till then a valid picture which is of course the goal.It is not about winning or whatever, but I would like to know how my results are compared to the other projects as well I am thinking about the bonus points.