Worried with geometric distortions

[David Cockey]

The head is tilted back further in the image on the left compared to the image on the right. Compare the relative positions of the bottom of the nose and top of the lips to the bottom of the ears. Also, the differences in the amount of hair and neck in the images can lead the brain to see a large difference so I cropped the top and bottom.

[Mfranquelo]

Thank you for your help David!

I've made another comparison in Agisoft by creating markers and aligning chunks based on markers.

* I created the markers on places as close as possible to the non-movable anatomy of the human skull so the distortions between each face are anatomically correct.

* There seems to be something strange with the nose and the eyes. It might be a psychological impression.

* I've attached a .GIF comparing both scans after applying Botox and before.. So changes on the physiognomy of the face may occur, but i dont know to which degree.

Do you think there are really no distortions ? i will test this week with very well known size objects to make sure of this however.

[Wishgranter]

Just try export the mesh and render it in some 3D app, so can set SAME cam settings for rendering, lens type, distance from object and etc..... if have some free time wil try it from the dataset......

[gulipin]

Hello,
I'v compared your 2 models, made them almost identical to normal world and aligned using medical chosen points on both faces. You can see that deviations can be tolerated, in 1st scan teeth are in normal occlusion, in 2nd scan they are not, what is more, forehead is a little bit different (guess it was dependant on mood)

[Mfranquelo]

Gulipin,

Thats just what i wanted, could you specify which software is the one you used ?
Could you do something similar in Meshlab ?
 

- Those values are totally correct, the patient was submitted to botox inyections causing the usual traumas and changes on the face volume.

[gulipin]

Mfranquelo,
Rapidform XOR3, guess you could make it with Geomagic Qualify, Meshlab or any other software with mesh deviation function

[Mfranquelo]

Im grateful for your help gulipin 

I've downloaded a trial version of Geomagic qualify.
Im having the following doubts:

1. My mesh is not scaled (its measuring 4mm x 3mm x 6mm) or something similar... so im struggling to find the correct values for the 3d compare tone spectrum...

2. I cannot find a good way to scale my models as real life size on Geomagic... i'll have a look at lee's tutorial for this i guess.

3. It keeps failing using the best fit alignment... both models are different in scale and this algorithm doesnt seem to match scales. I tried to use Manual alignment placing points on strategic points where the mesh corresponds to "bones" which seem to work better, however the 3d compare will vary depending to the quality of this alignment, if im correct.

4. As you can see my results are pretty weird, compared to yours.

[Mfranquelo]

I have performed another alignment on Meshlab with better results:

[gulipin]

Mfranquelo,
I'll try to help you a bit
1. when you make models in PS make them in real dimensions setting scale bar
2. unfortunately best fit is not acceptable for medical purposes, you have to use manually chosen points

[Mfranquelo]

I'll do this tomorrow, thank you for your help 
I guess i'll measure the distance between eyes, that shouldnt change. Should be around 7cm.

* I understand, i'll use manual point alignment.