| Title | : | Once a star |
| Name | : | Normand Brière |
| Country | : | Canada |
| : | ############# | |
| Webpage | : | www.noware.ca |
| Topic | : | Illuminated (November) |
| Copyright | : | Agreed - 2009-12-15 00:32:52 |
| JPG file | : | pw-1260837172-onceastar.jpg |
| Renderer Used | : | OpenGL |
| Tools used | : | Homemade Java software |
| Render Time | : | 3 minutes |
| Hardware Used | : | Macintosh Core 2 Duo, RadeonX1600 |
There she was, lying in the armchair, having thoughts about how famous she used to be or could have been. She should not have taken that spike heel off.
But through the door of her mind, she feels comfortable with this vision, a step forward into the light of dignity.
The Gods were with me on this one. I was looking for a completely different concept, and this idea popped out when I came across these two new models in the Archive3D.com website. Personally I don't see a difference between posing prefab models, or directly using preposed ones, as long as they fit. In this case, they did, they were perfect as is.
But once again I went into unexpected problems. Most free models are free because they have issues (no UVs, no normals, inverted triangles, bad groupings, etc). This time, the meshes were simply too dense. There was no way to work with them, especially for the ambient occlusion pass.
Indeed, vertex-based AO becomes prohibitive when the scene contains too many triangles. I needed to reduce the number of triangles/vertices. These models were defined using millions of triangles. I didn't have the choice but to improve my mesh simplifier. The previous version worked only on small meshes (< 10K), so I didn't use it for very big scenes because the simplification was slower than the AO itself in most cases.
Fortunately a new optimization can now eliminate 100 triangles per second for a 100K triangle mesh, and it goes faster as the mesh becomes smaller (maybe 1K triangles per second for 10K meshes, etc). As a result, I can easily divide by 10 the number of triangles for the biggest scenes without obvious visual artifacts, and one can imagine how fast the per-vertex AO now is.
Actually much more than 10 times faster, more about 100 times faster, because anything related to display (including visibility calculations) is also 10 times faster! This factor must be taken into account twice. For this scene, AO still took a few hours after the optimization, so I would probably still be waiting for it otherwise.
Most the time, when my idea is done composition-wise, I can simplify far-away objects on purpose for much faster display. Resulting in huge time savings for all interactions, etc.
_______________
Now more about the scene itself. It was really easy to duplicate the scene for handling two reflections, yielding 4 copies. My other entry ("Illuminatus") was so tricky in this respect. This time I made no mistake: the reflected scene below the floor first, then the floor, then the direct scene.
Recursively, the reflected scene in the mirror first, then the mirror and the walls, then the final "real" scene. This way, the copies are rendered bottom-left, bottom-right, top-left, top-right.
Some interesting stuff using my new adaptive transparency: the spot illumination made of two cones, and the ballerina-butterfly-girl both rendered with the normal-based opacity. A nice and versatile effect that I also used for the hair of the girl in the armchair.
The carpet is a textured rectangle with an alpha channel for the border. I simply used a copy of the carpet to add a very important shadow. The spotlight illumination on the floor worked successfully thanks to the 8-8-8 color clamping fix. This is one step (pun?) towards more HDR-like features. The floor was very important, and this perfect seamless tileable hardwood texture saved the whole thing. There is a pattern, but can one see it?
I may make some minor changes until the end, but as far as I am concerned, this picture is ready.
_________________________
EDIT: Just a quick point related to the "story" which is not meant to refer to any cliché whatsoever. I wish to make that clear. Someone wrote in the forum:
For the title, what about “Upon a star” with reference to “wish upon a star”?
My english is not deep enough to fully understand what it meant then, but I guessed it enough to explain my point. I know the picture is readable that way, but there is no reference to the collective unconscious, it goes the other way around. I answered:
"My intention is truly that the girl was famous in the past,
or at least she thought she was, or people told her she was.
Of course I don't mean a girl dreaming to become a star, only a girl hoping
she was actually a star, when she was one.
A girl only browsing through her memory to "validate" what she did,
hoping she did the best she could, when she had the occasion."
It is a vision of the past, not the future. The composition is intended for the dancer to be 10-20 years younger. Older, she's sitting pitiful, because having taken her shoe off brought back the total recall of her career and life. She thought maybe all this time, she has got it all wrong and failed.
Then she cried, just in case... it would release her, and it did.
_________________________
It is worth reading again the 2-line above story under this light!
| General statistics | ||
| No of ratings | : | 7 |
| Min. overall rating | : | 37 (15 / 10 / 12) |
| Max. overall rating | : | 49 (15 / 16 / 18) |
| Sum of rating | : | 296 / 420 |
| Date uploaded | : | 2009-12-01 09:12:07 |
|
Specific details
Note: The maximum value below is misleading as the voting system has changed.
|
|||||||||
| Rating type | : |
|
|||||||
| Artistic | : |
|
|||||||
| Concept | : |
|
|||||||
| Technical | : |
|
|||||||
| Overall | : |
|
|||||||
| Comments by members when rating this image | ||||||||||||
|
- Privacy Policy - Terms & Conditions - FAQ - Contact Us -
Copyright © Steve Paget & Colin Price with Stephen McAvoy & Thomas de Groot : 2007-2011 - All rights reserved