123D Catch experiments

As part of my dissertation I will be using the 123D Catch software to create 3D models using only pictures. While it is solely for my dissertation I can also use it to supplement scenes in my A3D project with it.
I am currently experimenting with the program using two models I have to hand, a Steamrunner class ship from Star Trek and Cortana from Halo, and I have had varying success.

Steamrunner

Cortana

However I have learnt a lot about the program. I have found out that the user has no modeling input whatsoever. The only changes you can make is by stitching together other images using reference points that you select or 2 different pictures. Each time this is done though the model has to be resent to be rebuilt using the new images as points of reference which takes a lot of time. Another thing I have discovered is that the minimum required images is quite high. I found that anything less than 40 pictures will get unreliable results, as will taking pictures using different quality cameras as my camera ran out of battery so I tried to take some extra pictures on my phone. This created a very poor quality picture if it could even be called that, it was fuzzy and there was no detail, just a mash up of colour.

The Streamrunner class was the most successful model although the underside of it was completely missing. This proved impossible to fix as the pictures of the underside wouldn't stitch with the others. The top half worked very well and would have needed very little work, mostly deleting and bridging gaps in the topology however the underside would have required extensive modeling. The pictures had no real structure to them, I simply took lots of pictures from various angles hoping it would work.

The model of Cortana however was a nightmare. As you can clearly see this model will need a massive amount of rework and even rebuilding. Her back is also missing and the rear of the base is completely the wrong shape. The face is completely lacking in detail, the only detail being her nose. The left hand is un-usable and would need to be completely re made. The only part of the model I am happy with is the display she is holding which appears to be clear and would only need the surrounding environment deleting and possibly extruding the display as it is a little 2D. I can't say why the model has turned out like this as the Streamrunner looked fine. It is possible that, as I used a high quality camera phone for Cortana rather than my actual camera, the photo quality has contributed to it. I will only know once I get my camera charger back. It is also possible that the material Cortana is made from contributed to the above image, again I can only know for sure once I get my camera back.

My next plan is to try this software out on buildings and interiors  along with different styles of lighting as one idea I have is to 3D model a network of caves so interiors and lighting will be a big thing I need to research and will be my main focus, once my camera is charged.