Conclusion

PhotoModeler seems to be a very interesting and exciting piece of software with great potential in the architectural field. The software could be used during the site analysis stage of the design process. 3d digital models of existing buildings could potentially be created from photographs giving context to a project. These models could also be very helpful when renovating or adding on to an existing buildings. The software also has the potential to be used in the sketch design process, photos of handmade models could be used to create 3d digital models which could then be exported into other programs such as Cad, Rhino, Vis and ArchiCad for design development. Another possible use of this software is the documentation and cataloguing of historic buildings.
I attempted to use the software to create a 3d model of the RAIA offices on Broadway by Julius Elischer. Unfortunately the software proved more difficult to use than first predicted. After 3 failed attempts at recreating this building in the digital world I moved to a simpler object. The jewellery box I used proved to be much easier and I quickly created a 3d view that well represented the original photos. This proves that the software will work with personal photographs.
For my model I think I am going to have to go to plan B. I propose building a simple handmade model photographing it and attempting to use PhotoModeler to recreate it. I will then export the file to Cad adjust it as necessary and print to the 3D printer to see how accurately the program can recreate a subject from a photograph.
I also looked at a program called Photosynth. Unlike PhotoModeler this software stitches photo’s together, similar to Autostitch to create a curved panoramic view that gives the allusion of being 3D. It is an interesting piece of software that could be used in the site analysis stage of the design process. The synths created have the potential to allow international/interstate architectural firms to view the proposed site in a 3d representation. The main problem I found with this software is that the resulting synths can only be viewed online by the synth viewer, it does not allow for exportation of the project into other software. If it was possible to export the resulting synth into programs such as Photoshop, Indesign and Powerpoint it could become a very powerful presentation tool.
The synth I created of the RAIA offices on Broadway was successful but as it cannot be exported for use in other programs it is not going to be of any use in creating a model. I could create a synth of the final model.
In conclusion it was very disappointing that my exploration into PhotoModeler was not more successful. I still believe that the program has potential in the architectural world but feel that a lot of training and practice would be necessary to create accurate complex representations of buildings.

Trial Four - PhotoModeler

As my trials in PhotoModeler had been so disappointing I decided to run a trial on a much simpler object to see if I could get better results to prove the software could work with my own images. I used the following images of a jewelry box.

I went through the same processes as I had in the previous trial. The resulting 3d image is shown below with quality textures mapped on to it.

It is a lot more impressive than previous results and the final image is instantly recognisable as a representation of the jewelry box. This proves that the software does work with my own photo's. This trial was a success and suggest that the software does have potential. This potential needs to be further explored.

Trial thee - PhotoModeler

I went back to the building again with the intention of only modeling the front left hand corner of the building and took several photo's insuring that every point was visible in at least two photographs. These are the images I used for my final PhotoModeler Trial.

I uploaded all of these images into PhotoModeler. Opening photo one I marked the outline of the reveals. I then went through all the other photo's and referenced any corresponding points

Each time I referenced the corresponding points in a new image I hit the process button to ensure that I had entered enough information to orientate the photo's and solve all camera stations. A small image of a camera appeared over the photos on the left hand side of the screen to show they were orientated.

The resulting 3d image showed the outline of each window reveal had been solved.

I then marked all the points of the window reveals visible in photo one...

and referenced them into every other image they appeared in.

The resulting 3d was missing all lines to points that did not appear in the first image as they had yet to be marked and referenced.

I opened photo two and marked any missing points which were visible on this image, then referenced them into every photo which showed a corresponding point. Moving onto photo 3, 4, 5, etc I continued marking and referencing corresponding points on the images. Every now and then I would hit the process button to update the 3d image on the right and see the progress of the project.

When all the points were marked and referenced I hit the process button again. Unfortunately the 3d model, despite showing all the points was not accurate. It was warped and twisted and some points were severely miss calculated. I went through every image, zoomed in on every point and adjusted it to more accurately match the photo's but the resulting 3d was still inaccurate. I then spent hours going through all the images and re-marking and referencing every point. I worked methodically, doing one window at a time and adjusting every corresponding photo.

The resulting 3d image improved slowly as I worked through all the images but some points just refused to solve even though they were marked and referenced in several photos and other points remained warped. I think the program was confused by the amount of data I was trying to enter.

I went through the photo's again and adjusted the points not to match the photo but to create a more accurate 3d model but still couldn't get the bottom middle window complete. I added surfaces to all the other windows using the surface command. The window reveals appeared to float ion space as the wall itself had not been modeled due to an inability to take photos which showed enough information to model it.

I mapped quality textures onto the surfaces. It showed the reflections of the windows in the photo's fairly accurately but without a surface on the wall it was not a good representation of the building.

I would have to say that this trial was again unsuccessful. Perhaps professional training in the software and a lot more practice would reveal the true potential of the PhotoModeler. The results of the tutorials were quite convincing, but on a more complex object it was not so easy. This was a disappointing result.

Trial one - photosynth

I took several photo's of the environment around the building to explore the potential of PhotoSynth. The following are the images I used.

Link to result: http://photosynth.net/view.aspx?cid=9a3d069f-3481-49dc-82b3-2cbab729d64f

The Results:

The synth was a success 100% match. The resulting synth can only be viewed online at the above link. It can not be exported for use in other programs which is disappointing. The following are screen shots of my synth.

This software could be useful during the site analysis phase of the design process. The 3d panoramic images can be created and saved online so that interstate/international firms working on a project can view them. It would also be a good tool for creating visual aids for potential clients to help them better understand the project.

I believe that the software would be more useful if the resulting images could be exported into other programs or saved as different file types. This way they could be brought into programs such as Photoshop and in design or Powerpoint and incorporated into a presentation.

Shown below are print screens of the resulting synth.