Photography Modeling

Hello everyone
After 2 weeks research, Finally i found a simple,easy and cheap solution to make game models realistically like Start Wars battlefront and New Battlefield game. My goal is optimized models for low end mobile devices and especially for human modeling.
Autodesk Remake - Free trial version - standard quality (Not Ultra) - mid-range mobile camera - Offline modeling (You can use Cload services)
PC: Win 8.1 64 - 16GB Ram - Core I7 4790k

How we used Photogrammetry to Capture Every Last Detail for Star Wars™ Battlefront™



But, how would you create for example an interior of a spaceship? In that case using totally free Blender would be better. Sometime it’s faster to make a model than taking a lot of pictures.

Another example that’s created from unity editor:

Another usages:

Ok i can agree in some cases Photogrammetry is better choice, but not always it’s the best solutions and i prefer Agisoft Photoscan ($) or open-source Visual-SFM or Insight 3d

1 Like

I really love Visual SFM, I usually use it in combination with MeshLab, it gives me TOTAL control over the mesh, the density, the structure and sometimes even the topology. I tried other things like 123D Catch, but I never felt like I had control over the process.

Autodesk Remake
Based on these images: Link
Screenshots

1024*577 + 3x Sharpen effect in photoshop
Screenshots

Rocks

Do you know “The Vanishing of Ethan Carter”?

Nearly all the game’s assets were created through photogrammetry. Looks pretty incredible. They used Agisoft Photoscan.

“The Vanishing of Ethan Carter” Explains Photogrammetry
How the Vanishing of Ethan Carter Devs are using Photography to Mimic Realism
Visual Revolution of the Vanishing of Ethan Carter

1 Like

Yes, I saw some screenshot of game.
I’ve tested Agisoft and find his texture quality a lot better than Remake. But Autodesk Remake is a lot better in hard edge models like environments and rocks.

My goodness. Interesting software.:slight_smile:

Does it work well with faces?

If you mean for human face,I couldn’t get acceptable results. But it can!!!

That’s exactly what I mean, sorry for being unclear :slight_smile: If you ever get a system like this to work with faces, contact me immediately :smile: I can segment the results, determining what is the mouth, the eyes etc so it would be possible to automatically generate animated faces in that case…

Still, impressive none the less!

I did not test human scan , but you can find a lot of solutions for this purpose:

https://www.youtube.com/watch?v=F74AZ6e2m40

https://www.youtube.com/watch?v=geQCzan6Usc

http://polycount.com/discussion/140949/photogrammetry-thread-best-practice

1 Like

Images from here:Link
Screenshots

If you want to make static objects than ok you can make auto-retopology, but if you plan to make animated assets than you have to do retopology manually cause you will have a lot problems with rigging :wink:

The Fastest phone to take images is iPhone. Its can take several images in 1 second without blury problem. I’ve tested 4s camera:
Take more images, Result is better model:

Screenshot

Screenshots