ontop option in Image?

[baba-yaga]

Happy New Year, everyone!
I discovered that there is no longer the option ontop=True in Image() function which was very useful to me in Image2D() in the previous versions of vedo. How can I keep a picture on top of everything else in a renderer now?
Thank you!

[marcomusy]

Happy new Year Sergei!
Yes try e.g.:

from vedo import *
s = Sphere().lw(1)
img = Image(dataurl+"images/dog.jpg") # this is 3D
img = img.clone2d("top-right").ontop(True) # this is a 2D copy of it
show(s, img, bg='wheat')

Also check out this example.

[marcomusy]

I think what you are looking for is texturing a mesh (?)
If so you need to define you (u,v) texture coordinates, then simply:

from vedo import *
img = Image(dataurl+"images/dog.jpg")
cube = Cube()
cube.texture(img, tcoords=None) # or simply cube.texture("dog.jpg")
show(cube, axes=1).close() 

[baba-yaga]

Well, it would be a too easy task just to texture the cube ) Actually, the primary task is photogrammetry: restore 3D object from images shot from different angles. The first step is alignment of the 3D model to the image using a small set of positioning landmarks. Next comes adjusting the other landmarks by moving them in the focal plane and warping the model. And here I need a precision with ability to zoom in without breaking the alignment of the image and the 3D model. Here how it looks in real:

You see a grey 3D model coming out of the image which cuts through it. So I need to move the image forward to make it visible at all places and then adjust the projections of the model vertices to fit the image. If I change the opacity of the model, I get a mess of all inner parts (teeth, nostrils, etc.) showing too

[baba-yaga]

Then texturing is also a complicated thing. I do not want to transfer the texture on the eyeballs, for instance, so again precision is needed. The same type as zooming the whole computer screen

[marcomusy]

OK - to be honest i'm not familiar at all with the subject of texturing a polygonal mesh :)

My suggestion (which may be completely stupid or naive)

• you can create an image and make a mesh out of it with img.tomesh()
• assign to this mesh x,y arrays by normalizing it the way you like (so this would serve as (u,v) later)
• write a callback function that allows you to click and select landmarks
• snap the 3d image mesh to some landmarks
• in the same callback you can reset/slide along an axis the position of the image wrt the model
• read out the values of the original [u,v] onto the mesh (which will not have changed)

Hope this helps!

[sergei9838]

Thank you, Marco! This is about what I am planning to realise. But it appears that I do not know how vedo treats obj-files. But this will be a separate topic.