The fellow who took on the design of the Align View tool had his hands full:
- a competing product had this feature. So Archicad needed it even tho sites rarely provide the geometric accuracy to place orienting nodes - certainly interiors are easier.
- photographs are distorted, meaning that aligning perfectly with archicad's perfect world CANNOT HAPPEN in a photo.
- few users will understand to zoom in to the exact pixel level when placing the orienting nodes for absolute accuracy because nobody except me (Illustration in Archicad) warned them about that.
BTW: As frustrating as it is, the Align Tool DOES WORK. The proof of this is inserting a piece of furniture into an Archicad rendered image. This approach eliminates analaog irregularities and proves tool operation.
This is a delicate subject for me because since we have the live OpenGL with photo background capability, it was always faster when using photos where the viewpoint was known to enter the camera coordinates and use navigation tools to "walk" the camera. How hard can this be to understand? Once you know the picture target (by criss-crossing the image to find its center) make the camera target that spot. Drift the camera view left and right, up and down, and experiment with camera view angles unless you know that from the EXIF data in the image file.
Perhaps you are offended because this looks like "fudging." It is. I regularly get files from people who ask me to align their camera view to a photo. In those cases, however, the photo has the survey building volume described by site sticks, so making a roof peak line up between two sticks at the ridge height is fairly easy to do navigating in OpenGL by trial and error.
This brings me back to the main point - neither system works without clean visual references within the background image.
Dwight Atkinson
Noemi Balogh
