Make Solid in meshmixer reduces resolution of model. What fix can I apply?

[awaisrauf95]

I want to 3D print a model of a human being. In order to easily clean up the model in mesh mixer, I use the make solid command. Unfortunately, the face loses it's clarity (I don't care about the rest of the body). I have another 3d scan model of the face close up.

So, how do I add this face model or it's texture to my full body model. Which software and what instructions should I follow?left side: original; right side: after make solid command

As you can see from the picture, the model on the right is ready for 3d printing (fill holes, make solid, hollow, etc.). But before I 3D print it I want to improve the resolution of the face. 

Thanks in advance.

[Anonymous]

Guess this scan is a result of a PrimeSense sensor?

There are two ways to apply color to a mesh:

1. Texture: There's a geometry of triangles. On each triangle a certain snippet of an image is glued to. 

2. PerVertexColor: Each vertex of the mesh owns some simple RGB information ( a certain color).

MakeSolid does PerVertexColor and tries to convert texture (if exists) information . So it depends much on the resolution one sets in MakeSolid's tool options. The bigger SolidAccuracy and MeshDensity (up to the source density) is set the better your result will look  (at costs of processing time and file size).

---

If you've a scan of the total body and a high res scan of a the face AND both objects were scanned using the same scale you might try to set the body as a target (magnet icon in ObjectBrowser) and  SelectAll (Ctrl + A) of the face. Transform the face to fit to the target as close as possible (using Deformation/Transform).  Now run Edit/AlignToTarget. After some ImproveSolution hits it should fit to the body .  

Release the body from being a target (> OjectBrowser). Select the bodie's face and discard it (hitting X).

Leave SELECT and activate the body + the face (press Shift while clicking a second object to activate severals). Run EDIT/Combine .

Now do EDIT/MakeSolid whit a high res setting. 

[awaisrauf95]

Thanks @Anonymous. I'll try it out and let you know.

I don't know if it has been scanned using PrimeSense sensor as I've been given this model by someone else to fix it up.

Best,
Awais

[awaisrauf95]

Dear @Anonymous,

The scan of the face is only of the front part and not the head as a whole.

So discarding the head (hitting x) will remove the complete head from the body. I need a way of merging the face texture onto the face of the model.

Thanks in advance,

Awais

[Anonymous · Accepted solution]

If you have a Body (a watertight object - yellow) and a single sided Mask(red):

Align the Mask to the Body:

SelectAll of the Mask and do Edit/Extrude:

Make the Body the active object and use SCULPT to push its surface to the rear so it doesn't intersect the Mask's front any more:

Activate both objects and run BooleanUnion:

Accept and use Sculpt to smooth out:

[awaisrauf95]

Can I hug you as a brother in humanity?

Thanks @Anonymous. It's done. Although do share your tips, if you could, on how this could be done if I had a face image that I would like to attach to the model instead of a model to model attachment.

Nonetheless, Thanks again.

Best,

Awais 😄

[Anonymous]

Sharing knowledge isn't that much a humanity thing. It's the key to improve common and also my personal knowledge. You can't imagine the amount the requests in this and other forums pushed my knowledge. Each question (might be basic or advanced) makes me think about tasks never coming up in my personal mind. Ones personal ways to do something is a scissor in ones head. Different points of view open ones personal mind. Well, one needs to invest time but hiding knowledge (doesn't matter which kind of knowledge) is dropping the common sense to a medieval way when an alchemist was cooking his personal soup.

(ok: somehow off topic)

---

now:

Think about a texture as a wallpaper glued to a wall (the geometry). You need to cut the paper to fit to a certain wall and glue it onto it. The walls of a room are pretty simple: a standard room owns 4 rectangular walls . Now on a mesh a "wall" is equal to a triangle: You need to cut the paper to (maybe thousands or millions of) triangles to glue it on the "walls". Ok - if you find the time that's no big issue: Simply measure the room of thousand walls and cut it from the paper so the edge of snippetX  glued to wallX fits to snippetY glued to wallY. That's the magic texturing does.

Now you may pull off the paper and glue a different image onto the geometry using the same cutting pattern. You'll notice soon that the relation geometry/image may be different. A different face image might have a different distance of the eyes. A different expression of the same face might draw the mouth differently:

If you glue another paper to the same walls using the same pattern of cuts will never give a "right" result.

So you need to calculate a different cutting pattern for the new image ( a new texture atlas).... this is not at all trivial image processing as one needs to find fitting shapes in the images (e.g. find the corresponding pixels of wrinkles) . MM doesn't do this and I don't know any tool doing something like this straight forward (well, there are tools where you can project a 2D image onto a mesh by hand, other attempts convert different geometries to a common tessellation (morph targets)... one can do awesome stuff using those tools, but it's not that simple).

Sorry for being that verbose. Think, state of contemporary art is to weld textured meshes and so keeping their individual textures precisely . MM does it for free - and this isn't, not at all, a low budget feature.

[awaisrauf95]

Solid reply mate. Thanks 🙂

[kshepherd2005]

I was able to fix everything by opening the model in Photoshop and generating UV's.

[nathanyoast]

Once you're in the make solid menu choose solid type accurate. Turn solid accuracy and mesh density up all the way. you can then click on their respective numbers and increase those beyond the max that it shows. I would suggest doubling those numbers as a start to see how it goes. It will take an increasingly longer time for it to generate the model and you may hafta grab the screen and move it with the mouse to get the model to show up after it has finished processing but it does work. I believe this is the fix you are actually looking for. I will say that, if you go too high with your numbers, the program will fail out so try to experiment with just how hi you can go. Hope this actually gets to you and helps you with future endeavours.