Show Posts

This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.

Messages - cbnewham

Pages: 1 ... 6 7 [8] 9
General / Re: Poor model details from good dense cloud
« on: January 24, 2014, 10:49:08 AM »
I set the triangle count at one point to several million and still ended up with everything the same, except with far more triangles in the broken pieces.

I will try this again to double check, but in the older example that came out fine, I can tell you that my mesh will have had no more than 1 million before decimation.

I am still suspicious about build 1812...  :-)

General / Re: Poor model details from good dense cloud
« on: January 24, 2014, 10:23:01 AM »
The number of triangles makes no difference. I always have it set to about 500,000 and increasing them leaves it in the same state - only with more triangles... :-/

Anyway, overnight I re-ran the earlier set of photos that came out ok and it now looks like the later one with missing details and broken edges - so it looks like something has changed in the new build of PS.

Point cloud: 700,000 points
Dense: medium, moderate filter
Mesh: 600,000
Decimate: 500,000

I guess this problem stems from solving the texture problem of the previous build? If so, that's a shame. I'm reverting back and will run this through again. If it works I'll pass on this build - I'd rather work around the texture problem than have a mesh which is clearly not generating properly.

General / Re: Poor model details from good dense cloud
« on: January 24, 2014, 01:03:47 AM »
Here is the earlier one I built back on the 8th of January. This had less cameras overall, but exactly the same cameras with this area in view as on the later reconstruction above.

Note that the dense cloud has only 16 million points in this example, and 20 million on the later one above.

Why the difference in quality of the model?!

General / Poor model details from good dense cloud
« on: January 24, 2014, 12:32:17 AM »
I would like to know why the model becomes so degenerate even though the dense cloud has ample points and clearly defines the surface. There appears to be no way to adjust how the dense cloud to surface algorithm works.

Note the stems disappear on the crown and the curved edge nearby has a jagged section taken out of it. Strangely, I've built this one before (but with a few less cameras - not, I should add, imaging this particular area. I added them to the current version for detail elsewhere) and the stems were generated and the edge was not jagged. So something is either wrong with the build or there is some magical change that happens when more cameras were added.

This was with the latest build (1812).

General / Re: Masking and texture overspill
« on: January 22, 2014, 12:13:35 PM »
I've done some more investigation.

The example I gave is not actually an example of the overspill - which I have seen on another example (I'll dig that out at some point and post it).

However, what I do notice with this (above) example is that the mesh surface generation with "interpolate" (the default) generates surfaces that just don't exist in the dense cloud. In the example shown, the mesh is being extended past the bottom of the skirt and also past the edge on the left hand side where there are no dense cloud points. These mythical surfaces then get other parts of the photographs projected on them when texturing is done. When I generate the mesh with no extrapolation or interpolation the result is correct (albeit with all the holes not filled in).

One aspect of these examples is that the photos are mostly taken from above, with very foreshortened views of the vertical sides of objects. My intention is to generate overhead views, so I'm not too interested in missing detail from the sides of objects. This all works fine except for those non-existant surfaces that are being generated.

A work-around for my case is to use height field generation. This does produce a "correct" result, but I would have thought surface generation in the normal, arbitrary, way would produce a correct result too. I don't know much about how meshes are generated from point clouds, but I find it surprising that the algorithm would construct surfaces that have no points to be constructed from!

General / Re: Masking and texture overspill
« on: January 20, 2014, 03:56:18 PM »
1.0 Build 1795

I also see it has invented some extra surfaces (left hand edge) where there is no dense cloud. I've captured the screens shots straight from each view in PS, so I'm not sure how that has happened. I generated this model yesterday and was messing around trying to remove the spill on the texture, so the extra edges may be something I've done, although all I did was masking (I didn't edit the mesh outside PS)... but the overspill has always been there (except when I remove the last few photos you can see in the thumbnails at the bottom when I generate the texture).

General / Masking and texture overspill
« on: January 19, 2014, 11:24:00 PM »
I have noticed on several occassions I get overspill in the texturing, even though the dense point cloud clearly shows that the model is correct, and the points in the cloud have the correct colour. By overspill I mean where the edge of one surface is being projected on another which lies on top or underneath.

What is the solution to this? I can't mask the edge because it is part of what I am modelling. In one case I noticed it was happening because I had several photos taken at a different lens focal length (the cameras were in the correct positions in the 3D view). When I removed these cameras from the texturing it seemed to produce the correct result. Odd.

General / Re: Question on Texture Count and importing into Blender
« on: January 09, 2014, 12:09:20 PM »
Thanks. That works well.

General / Re: Align photos point limit
« on: January 08, 2014, 04:46:24 PM »
From what I have seen it is an upper limit to the number of points it will use for finding points and matching them. The number of points you get in total on your photos will depend on the number of places it can find features to generate points for and the number of points in the final sparse cloud (which were matched) will be lower than this.

General / Question on Texture Count and importing into Blender
« on: January 08, 2014, 12:14:16 PM »
I've noticed that in the current version of PhotoScan that I can get a bit more detail in my textures if I lower the texture size and increase the count (38,000 x 1 or more than c.30,000 x 2 seems to be the limits on my kit).

How can I use the texture UV maps in Blender? i.e.: how can more than one be mapped to the model and how does one go about it?  I know how to map one image when I've imported the model into Blender, but two has me stumped...

General / Re: parametrezing texture atlas freez at 70%
« on: January 08, 2014, 12:07:33 PM »

BTW, the new texturing is a lot better than in the old versions. I found quite a few problems with 0.9's texturing and I've gone back to 1.0 which produces a lovely result. I can live with medium resolution models and lower texture detail for now. I look forward to the fix!

General / Re: parametrezing texture atlas freez at 70%
« on: January 08, 2014, 12:19:08 AM »
Further to this:

Using the same data, I generated a dense point cloud at medium quality and then a 500,000 face model (the previous dense cloud was generated on high and then the model with 300,000 faces).

This time it textured in very little time - instead of leaping up to 70% in 30 seconds and sticking there, it took a couple of minutes to get to 70 and then continued on past the parameterisation and on to the blending stage.

As someone else noted, it appears to be related to how the quality of the dense point cloud is set. Anything above Medium and texture generation grinds to a halt at 70%.

General / Re: parametrezing texture atlas freez at 70%
« on: January 07, 2014, 07:03:39 PM »
300,000 faces in my case.

General / Re: parametrezing texture atlas freez at 70%
« on: January 07, 2014, 05:59:16 PM »
Unsetting colour correction makes no difference - it still takes a long time...

I installed 0.9 and, after much messing around to export everything from 1.0 and import it back into the older version, it works very well. Only 10 to 15 minutes to generate the textures.

Please fix this in 1.0 when you get a chance - it's obviously a bug.

General / Re: parametrezing texture atlas freez at 70%
« on: January 07, 2014, 03:44:51 PM »
Which one?

Currently I'm on the latest (v1 build 1795).

Are the previous versions available on this website? I haven't looked yet, but I guess they must be.

Pages: 1 ... 6 7 [8] 9