I’ve updated the PhotoSynthToolkit, it now integrates the CameraExport 3DS Max script created by Josh Harle. I’ve converted his code from C# to C++ with his approval, so the 3DS Max script is now generated directly when you download a Synth with my PhotoSynthDownloader. So now you’ll find a “cameras.ms” file in the “bin” folder. In the video just below you’ll see the difference between the rendering using only vertex color and the one using texture projection:
For 3DS Max beginner like me, this is how I have setup the rendering:
- add path to your synth: Customize -> Configure User Paths -> External Files -> add Path to your pmvs\visualize folder
- when importing the obj file be sure to uncheck Flip ZY-axis
- change environment global lighting: Rendering -> Environment -> Global Lighting: level 0.0, Ambient white
As usual the new toolkit his available on his dedicated page.
The video is very impressive. I found an alternative way to create dense mesh: I simply load my PhotoSynth photos to http://www.my3dscanner.com to get free dense color point cloud. Than I use .ply in MeshLab. Sometimes it works very well, sometimes it doesn’t
Superior quality!.
. So, by transferring vertex color did not generate nicely mapped models. This way, we can generate medium-low res geometry with HD textures, perfect for realtime visualization. What I believe will happen and dislike is that unlike with the color from vertex, when projecting, objets in front of our objet such as trees, cars or so will also projected… lets see what happens.
I was stuck with this because in meshlab I cant generate high res meshes because it closes suddendly
Wow, that’s pretty impressive! It’s amazing the difference the textures make.
@Pablo;
You can get around problems of projecting objects onto your subject by masking the trees, cars, etc. out of your projection images;
if PMVS2 can reconstruct the vertex, then you have at least 3 images of that vertex to use for projection mapping.
I have a video in the post linked to above that goes over some options for controlling the mixing of projected images.
Hope that helps.
Integrate J. Harle code was a great idea. Unfortunately 3DS is a commercial (and expensive) program. Is there a way to do the same using Blender?