Cineversity

Hi:

I’m supposed to create and rig a model of a backpack that can open in a few places to have a laptop and some accessories fly into it.

For the first time I played around with a photogrammetry approach to create a not-horrible model of the backpack as seen in the scene file I’ve linked to.  Of course this is too heavy to rig and animate, so I want to retopologize it.

I’m wondering what a good workflow would be to do that. 

After watching some Polygon Pen tutorials I started tracing over the generated model.

One question I had is how is Reproject Result in the Poly Pen options different than turning on Polygon snapping for the whole scene?

Once I had a coarse mesh built I was going to subdivide it so I can conform it to some of the dents and folds in the bag.  From that I’d rig the model and create UV maps.

Is there any simple way to magically transfer the texture maps and mapping from the photogrammed model and apply those to my lighter-weight, rigged model?  Some kind of reprojection from one model to the other?  Probably not.

I haven’t really tried to retopologize something before, so any suggestions on the most efficient way to do that would be great.  I know that the shrink wrap deformer is an option, too.  I don’t know if that would be a better approach.

The scene file is too big to attach, so I’ve posted it here:

http://marshall-arts.net/Support/RetopoTest.zip

Thank you.

Shawn Marshall
Marshall Arts Motion Graphics

Hi Shawn,

Thanks for the file. Polygon Pen is the cleanest way, as no code can do that for you. What I mean by that, to have a low-res mesh where no deformation will be and a denser where it folds softly while having the edge following the fold direction.

Several ideas, but not a single “one size fits all” suggestion.

The animation of open the backpack would require new textures as well, just to get this out of the way.

Textures, and with that the UV, could work with VAMP, but I haven’t gotten any good UV transfers with this kind of projections. Of course, I have run a test with the file, not even remotely anything useable.
https://help.maxon.net/us/#11344
Why isn’t it working? The UV(W) polygons are tiny, and can’t be translated into a rough mesh. Since the original UV(W) polygons could be all over the place regarding the texture, there is no option to find the nearest point. (I guess there are too many nearest points… to say it in a nutshell)

To create a new mesh, perhaps Instant Mesh can do something for you. But I did not get something ready to use mesh from it, a little bit of care was mostly needly.
https://www.cineversity.com/vidplaytut/turn_any_object_into_hexagons_with_instant_meshes

Next, and how I would do it, the PolyPen is a great start, you could go even relatively rough to work (mesh size) and cut later for details. If that is a given, set up a camera that captures the best view, and certainly a second one to cover the “blind spots” of the first camera. Then render the two cameras from the high-density mesh, and camera project it to the Polygon Pen results. You can create a UVW Coordinates from each, and the cameras are then backed into the UVW of each material. Works around 95% in my experience. This gives you at least a quick result. The quality of the initial image will suffer, there is “no free lunch”, so take the source images as hi-res as you can.
Here is an example, but I missed a tiny polygon, I left it there, as it illustrates the process. Either Polygon selection or (preferred) alpha masking for several of those (Projection Man comes to mind)

Besides that, keep the hi-density mesh and create a low res cage, animate that cage with the Mesh Deformer. The Volume Mesher/Builder could give you quickly a descent Mesh (1cm in the builder, while the Mesher has the adaptive settings to a few %)

Another direct way for animation would be the PoseMorph or for very simple point cloud animations with the Cluster object. (I like the Pose Morph better, but the Cluster with its simplicity has something to offer as well.)

I know that this is just a test-run, but I would always try to retrieve textures and rebuild if possible, some tips to gain even from complex surfaces textures are in my link below the signature.

Last but not least, Polygon reduction, but perhaps more last than least after all.

Let me know if there are any further questions, I’m happy to look into it.

Enjoy your weekend

Again, thanks for the quick reply.

I’ll play around with some of the techniques you’ve described and let you know if I get stuck.

It looks like 3DF Zephyr has a workflow that allows you to create more cohesive UV maps than the mess I’m currently using.

Best.

Shawn

You’re very welcome, Shawn.

The application 3DF Zephyr is limited to one OS, and so I can’t tell you anything about it. I hope you find what you are looking for.

I’m confident that we will see a lot of advances in this area in the future, with Artificial Intelligence as the core of it. Deciding what an edge is and what just a painted pattern or even a shadow line will be an indicator of how far this has gone. I’m somewhat uncertain if it will ever match the quality we can do manually, in both, image quality and the aesthetic of a mesh, in terms of usability (rigging and any form of animation).

Even if I sound old-school here, a series of high-resolution images taken, perhaps with a camera position, yet taken from the meshing-software, and without any changes (ever pixel move is lowering the quality), camera mapped on it, sounds to me a good idea.

My best wishes