As is to be expected, the models that come out from this application are not amazing. The program also has the nasty habit of merging the ground in the model, which makes it hard to take small objects as scans.
It could in some cases, but as of right now, most software alegorithms (including this program) use the ground plane, and/or other near by environment patterns to help localize position and change in movement. So the ideal is not to have a pure colour, but rather a sharply contrasting patern, or something with easily markable points. :)
Oh interesting, so you could go with a highly patterned background for an object that is a solid color possibly?
Well, not really, because it also needs to mach displacement of different locations on the object in order to calculate topology!
Oh so its multiple shots of the object? Could set up some sort of spidle, but really I guess it's a bunch of effort at that point that more professional software could be used.
And using a spindle means that some of the environment does not change(tracking it is no use), which makes it even harder for the software.
Yes, mobile technology is really sub par in that category, but there is no more real hardware reason for that. Also, proffetionall computer software's also require multiple images (usually 40 to 50 of them). How could the software know what the shapes are without seing all around the object! :)
Wow that is a lot of shots.
Guess my plans for simple setup is a no go, oh well. :D
Lol! You can get away with 15 to 25 photos if you don't mind errors in your model, or if its a simple shape. You might also be interested in some software that tracks the object, and makes the model as you move around the object. But that might cost a little...
Hmm yeah, that sounds a bit better to make as you go instead of using tons of shots.
I wonder if using a makeshift bluescreen would help when taking photos? That way the object stands out better.