I think the biggest hurdle right now is the software itself. More specifically how much RAM and processing power it can actually use, coupled with the net performance. I don't personally suffer from it, but plenty of people still have issues with lag and I've noticed longer loading times in recent years, despite very capable hardware and an SSD install.
If I'm reading you correctly and what you mean is either photoprojection, dynamic lighting and/or realistic matetials, I think it might prove hard to achieve within the limitations of the client itself (let alone mobile) and without making the size of your avi skyrocket to 5000kb's and above. I already struggle to keep it under 3k like I used to as texture sizes increase, or more novice developers come onto the scene, but don't have the skills to effectively compress.
I'm not a 3D modeller or dev of any substance myself, so apologies if I am misunderstanding things or incorrect in my knowledge. I certainly look forward to more advanced creations in VU, I know there's already been some cool strides taken with mesh heads and + materials. But I fear IMVU may have hit it's technical peak, especially when it lacks the user numbers and funding for upgrades and advancement. If only they allowed 3rd party clients as SL does with it's viewers, it would allow a little more creative breathing room.
