Gimpel3D 2D/3D Stereo Conversion Software is now opensource! The code has been released under the GPL and is available at:

http://sourceforge.net/projects/gimpel3d

An alternate link for the codebase can be found here:

https://dl.dropboxusercontent.com/u/5171737/G3D/G3DCodeBaseFULL_11_17_2013.zip

Developed by René Gimpel (an independent stereo researcher) between 2008 and 2011, Gimpel3D is recognized for its advanced modeling features not seen in similar programs (including commercial packages at the time of release).

Since the original freeware release, many of the ideas presented in G3D have become standard features in commercial software. As a result, there is no significant advantage to developing G3D as a stand-alone proprietary solution, although the codebase makes a nice platform for future research and experimentation.

René Gimpel has released all source code for G3D under the GPL so that other stereo researchers, hobbyists, and/or professionals can build upon his existing work, and integrate/test new capabilities using the projective modeling environment.

A Brief Message From René: History of Gimpel3D - The Little Drawing Program That Could:

Gimpel3D 2.0 - 2D/3D Stereo Conversion Editor:

Gimpel3D converts single images or frame sequences into stereoscopic 3D using a combination of traditional approaches and a proprietary projective modeling system.

The user works in true proportional space where the scene can be viewed from any location. The scene is edited geometrically in space using tools specifically designed to work with the perspective projection of the image.

This creates a virtual workspace that is intuitive for the user and proportionally accurate.

* - Wide selection of tools allow for any level of detail.
* - Easy construction of regular geometry using Auto-Alignment.
* - Complex scenes can be modeled once and re-used.
* - Surfaces can have physical contour and/or relief texturing.
* - Custom models and/or depth maps can be applied selectively.
* - Realtime anaglyph stereo view provides immediate visual feedback.
* - Objects and stereo settings can be animated and keyframed.
* - Multiple options for gap-fill and post-processing.
* - Transfer edits between varying resolutions.
* - Can import data to leverage existing pipelines and/or content.
* - Can also be used as a start-to-finish solution.
* - Extensive help documentation and examples.

The fully functional FREE version of Gimpel3D is available here: Gimpel3DSetup.exe
(alternate link) Gimpel3DSetup.exe

Credits:
Core Algorithms and Concept: - Filippo Brunelleschi
Code Implementation: - An original work by René Gimpel

-----------------------------------

I developed Gimpel3D as a pet project while studying traditional illustration techniques. Many of the concepts used in the underlying process can be directly traced to my original interest in drawing (this is the main reason I was attracted to computer graphics in the 90s).

I maintain exclusive ownership of the code, and there are no other contributors aside from a small group of volunteer testers. I also wrote the complete help documentation and generated the examples.

-----------------------------------

I have developed a plugin system and SDK which allows me to continue developing Gimpel3D into the future. I believe there may be interest in licensing or purchasing this SDK (or the entire codebase), and/or employment opportunities doing custom integration and support.

If you believe that my skillset may be useful to your company or project, please feel free to drop me a line and discuss it, I am always willing to listen to any proposal.

-----------------------------------

Some Test Images Used For Development:

The anaglyph images below are a short compilation of various test images used throughout the course of development, each were chosen for different purposes as test cases for specific editing features.

NOTE: Due to licensing and copyright restrictions, it is not possible to include any of the live footage converted during development testing.

Click the images for a fullsized view.

The following videos briefly and informally demonstrate some of the most basic features in Gimpel3D:

This video shows a basic example of using the perspective projection of the image to establish the proportional 3D space and original camera view. This is usually the first step in converting any image or sequence, and provides a solid framework for more advanced editing (texture, contour, characters, moving objects, etc.).

This proportional framework can be re-used for subsequent shots or images that show the same geometry from a different viewpoint (ie. moving camera).

This video briefly demonstrates the 4 most basic editing operations for individual pieces in Gimpel3D:

- Depth
- Orientation
- Contour
- Relief

These are the building blocks of scene construction in Gimpel3D. Extended editing tools (such as Depth Painting, Anchor Tools, Projection Geometry, etc.) are not covered in this video.

This video shows a basic example of using a generic head model to create facial features. The projection alignment is reached by associating feature points in the frame with points on the geometry model.

Once the initial position of the model is established, it can be animated and keyframed to follow the face throughout the clip.

This video demonstrates the 3 basic stereo view settings:

- Eye Separation
- Focal Depth
- Linear Scale

The stereo camera is visually represented in the 3D scene, and the effect can be previewed in anaglyph mode (as with all other editing).

Stereo settings can be animated and keyframed throughout the clip.

This video demonstrates application of a procedural relief texture to a surface.

There are 3 types of adjustable relief texture available:

- Random Noise
- Grayscale Image
- Procedural

This is used along with other editing options (position, orientation, contour, projection geometry, depth painting, etc.) to create the final surface.

Relief texturing is useful for plants, trees, rocks, or other rough textured surfaces that would be difficult or impractical to model by hand.

This video demonstrates the use of Anchor Points to "anchor" a portion of a piece to a specific location in 3D space.

The piece is treated as a flexible membrane that is stretched to fit. Multiple anchor points can be layered on a single piece. The influence range is adjustable for each anchor point, and they can be animated and keyframed independently.

Anchor points can also be used as a general sculpting tool for scene geometry.

This video demonstrates the Depth Paint tools used for final touch-ups after the primary 3D editing is complete.

Depth painting is generally reserved as the last step in 3D editing, though it is possible to re-edit the underlying pieces.

The 4 Depth Paint options are:

- Push
- Pull
- Blend
- Flatten

This provides fine control over small details as well as broad control over large areas.

This video demonstrates duplicating the scene geometry for all frames in the video clip. In this example, the camera position changes at each frame.

After the initial scene geometry is created, the camera path is generated from tracked feature points in screen space. Knowing the virtual camera path allows the software to re-apply the reference geometry to the entire clip.

NOTE: In Gimpel3D2, the reference geometry for a particular scene can be saved to a "Virtual Set" database and used for future video clips that show the same area.