Forum

Author Topic: dense/moderate model more faceted than dense aggressive?!  (Read 6493 times)

andyroo

  • Sr. Member
  • ****
  • Posts: 436
    • View Profile
dense/moderate model more faceted than dense aggressive?!
« on: January 25, 2014, 12:21:38 AM »
My latest model run produced unsatisfactory results (too faceted a surface) with dense/moderate settings, which I thought were the same as the default/recommended in 0.8.5 or 9.1 (I forget the details of each version, sorry).

I re-ran the model with aggressive, but it clipped out/overly smoothed areas of the model that I don't want clipped.

I've attached two images showing close-up and further out views that I hope illustrate the issues I am having.

I am confused about dense cloud generation and depth filtering. Especially confused about why the moderate filter produces a more faceted surface than dense (is it because I "used up" my 40 million faces elsewhere in the model in "moderate"?

I continue to struggle to produce the same quality DEM that I did in 0.8.5, and I find the settings less intuitive and less exposed for me to tinker with so I can optimize them for flight conditions (exposure quality mainly, primarily due to sun angle and cloudiness)

Also I don't understand why depth filtering step is applied in dense cloud reconstruction and not in mesh. Are the points generated but flagged with a filter value, or are they just not generated (where I have the clipped values in the Dense/Aggressive mesh)?

Marcel

  • Sr. Member
  • ****
  • Posts: 309
    • View Profile
Re: dense/moderate model more faceted than dense aggressive?!
« Reply #1 on: January 27, 2014, 12:44:56 PM »
Also I don't understand why depth filtering step is applied in dense cloud reconstruction and not in mesh. Are the points generated but flagged with a filter value, or are they just not generated (where I have the clipped values in the Dense/Aggressive mesh)?

That is actually an interesting question. I know that Photoscan creates actual depth maps for each camera (you can export them as EXR if you have 'keep depthmaps' enabled), so I always figured that the depth map filtering was done one these images. But when it is done with the depth map filtering stage, it does not take much time for to complete (point cloud and all). So either the conversion from depth maps images to dense point cloud is very fast, or the the point cloud is build during the depth map filtering?

Does anybody know the exact process?

Alexey Pasumansky

  • Agisoft Technical Support
  • Hero Member
  • *****
  • Posts: 14753
    • View Profile
Re: dense/moderate model more faceted than dense aggressive?!
« Reply #2 on: January 27, 2014, 03:36:57 PM »
Hello Andy,

Actually, Aggressive filtering is recommended for most cases and it is equal to the P1 = 40,  P2 = 2000 parameters used in very old version (we're not sure that those parameters are more intuitively understandable than current filtering options).

As far as I can see from the screenshot, Aggressive filters out forest, while Moderate keeps the forest but during meshing a lot of faces are used to create the mesh in the area of the forest and too little are left for the rest of area.
Best regards,
Alexey Pasumansky,
Agisoft LLC

andyroo

  • Sr. Member
  • ****
  • Posts: 436
    • View Profile
Re: dense/moderate model more faceted than dense aggressive?!
« Reply #3 on: January 29, 2014, 10:27:03 PM »
... Aggressive filtering is recommended for most cases and it is equal to the P1 = 40,  P2 = 2000 parameters ...

As far as I can see from the screenshot, Aggressive filters out forest, while Moderate keeps the forest but during meshing a lot of faces are used to create the mesh in the area of the forest and too little are left for the rest of area.

Hi Alexey,

Thanks for the quick reply. If that is the case then I don't think it is functioning the same as P1 and P2 in 0.8.5. In 0.8.5 it seemed like the polygon budget was more evenly spatially distributed over the whole model area.

I was able to have texture for forest and reservoir with ~25M polys at that point, and now I have upped my polys to 40M.

If I had a better way to clip the dense pointcloud data (or constrain where it is generated), I could test that on my full datasets, but I will check out a small subset and see what I can learn.