Really interesting workflow. But honestly the 3d model quality looks terrible. Wonder if that can be massively improved with the current software that’s available?
a few points: 1. It's pronounced "gow-ss-ee-an" after Carl Friedrich Gauss - google for more information. (normal distribution etc.) 2. The DSLR you used is not the reason you have "bokeh blur". that's the lens/f stop. - this means that you don't need to use your phone, and in fact its usually better to use a higher quality sensor. just control your depth of field. 3. A crop box... is a crop box. it doesn't "help the object come into focus". it crops the splats inside the box. 4. Would be great if you could spend a little more time on your script instead of jumping head-first into a video with such limited research.
So once the splat is generated, can we edit/refine the model in 3d programmes so it doesnt look so terrible? Or are you always stuck with the splat that gets generated with no options to customise it
Higher quality can be done with better lighting and more photos (probably better than the quick ones I did) Also I didn't test that, but if you do, let me know! I know you can generate an actual 3D model too, so I'd assume so!