|Page (1) of 2 - 10/21/04||email article||print page|
RealViz ImageModeler 4.03D models from 2D photos
Creating realistic 3D models for any level of production can be a challenge. Building a high quality model takes time. Using reference photos to build your object and place the model in a photorealistic space can take a lot of pre-planning as well. Wouldnt it be easier to take those reference photos and have an application generate the model for you in as few steps as possible? Enter the latest update to RealVizs ImageModeler.
ImageModeler 4.0 allows you to easily create 3D objects, for film or video all the way down to virtual products for websites. The models can be as simple or as complex as you like them to be, and it all starts from a series of photographs.
Still Image Capture
ImageModeler 4 extracts 3D information from your still images to construct a virtual 3D environment from which you can extract and build your object. In order for ImageModeler to extract the data correctly, it is best to obtain multiple images from various angles around the subject.
Because ImageModeler 4.0 has a very powerful texturing feature, making sure you take high quality images to begin with is key. ImageModeler 4.0 supports Cineon, Maya image files, JPG, PNG, Portable Pixelmap, SGI files, TIFF, Targa files, and BMP. For future texturing and to ensure ease of marker placement, I would recommend that you use the largest image size your camera supports.
For the best texturing of your model you should shoot the object(s) in a very diffuse light to eliminate as many shadows as possible. While this may not always be possible when you are shooting building and other subjects lit by the sun, a diffuse light will allow you to pull a better texture later in the modeling process. There are new texture features in ImageModeler 4.0 that can help if you are in one of this situations, Ill cover that in a moment.
With the images captured, they can be imported into ImageModeler for the first step in building a 3D object.
Calibration and Modeling Markers
In many instances the camera data that you import will include information regarding the focal length of the lens, exposure, etc. In order for ImageModeler 4.0 to take the 2D data create a 3D environment, requires you calibrate the images by dropping calibration points throughout the images. Placing calibration points is not a terribly difficult task, but is one that requires you to pay close attention to where specific features are in each photo.
Even though you may be working with large images, your screen resolution may limit your ability to see as much detail as you would like. Because of this, when you place a calibration or modeling marker, ImageModeler has a pop-up zoom that gets you in very close. Combining this handy feature with multiple views and quick navigation controls creates a splendid environment for quick marker placement.
When I was familiarizing myself with the application, I was concerned that the sample projects had a large number of calibration points in them. While it is important for the application to be accurate, laying down two dozen points for a simple building, I believed to be over kill.
One of the key improvements in ImageModeler 4.0 is a new calibration engine. After placing 10 locators, ImageModeler informed me that enough points had been dropped to do the calibration, but if I wished to fine tune the settings, I was welcome to lay down more. I decided to place a few more, but based on the results returned, was satisfied that ImageModeler 4.0 was doing a great job.
Following the placement of calibration markers, you then have the option of creating a cloud of modeling markers to serve as anchors for primitives or polygons. Depending on where you place your calibration makers, you may be able to use them as modeling markers too. I found using a combination of these work best.
There are a couple of ways you can lay down modeling points. The first is to place markers in places that define key points on the object. For example, if you were modeling a building, placing points at the corners of the building would be enough.
The other way would be to place many makers along the contour of the object. This method would work best for modeling complex organic shapes.
With multiple camera views and fast navigation tools, placing points in the viewer happens very quickly. The first few times you work with ImageModeler 4.0 you may have difficulty placing your points in the exact place you wish, but with practice, laying points is a breeze.
It is probably most helpful if you have worked with other 3D applications that allow you to build objects via point placement and polygon creation. If you are an old school modeler and are familiar with creating reference points on a physical model and then using a point scanner to enter the data into a 3D application then you will understand how ImageModeler likes to work. This will have a huge impact in the time spent on modeling.
Related Keywords:RealViz, imagemodeler, 3d, 2d, photo, animation, modeling, schleicher