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 - David Cockey

Pages: 1 ... 10 11 [12] 13 14
I have a new HP laptop with an Intel I7-4700MQ processor and NVIDIA GeForce GT 740M graphics. Graphics driver is the latest from NVIDIA. Open Hardware Monitor shows very heavy use of the GPU during Build Dense Cloud with PhotoScan 1.0 Build 1748, and the log shows the GPU contributing as much as the CPU. Clearly the GPU is not "crippled".


Feature Requests / Re: OBJ Texture Names Chunk Specific
« on: October 29, 2013, 12:03:30 PM »

General / Re: Best post processing in Lightroom
« on: October 22, 2013, 04:43:16 AM »
I adjust the images so that none of the areas which are needed are either saturated or "black". Setting of the controls depends on the images.

I use the Chromatic Aberration correction and vignetting correction in Lightroom since there are profiles available for my lenses. I make sure to turn off distortion correction.

General / Re: Accessing RAM
« on: October 09, 2013, 09:18:00 PM »
My experience is PhotoScan uses as much memory as it needs, up to using all the available memory.

The amount of memory needed depends on settings and can vary greatly. What settings are you using?

Generally alignment requires less memory than creating the dense point cloud and meshing. My strategy with large numbers of photos (up to 600 photos) has been to align all the photos together, and then create duplicate chunks. The bounding box in each chunk is adjusted to cover a portion of the object. The resulting meshes can be merged before exporting if desired.

Feature Requests / Re: recalculate selection
« on: September 27, 2013, 09:22:16 PM »
Simple to do.

1) Duplicate a chunk which has the desired cameras/photos aligned.

2) Adjust the size and orientation of the bounding box in the duplicated chunk to fit the area of the object where the higher quality is desired.

3) Optional - In the duplicated chunk disable the cameras/photos not needed with the revised bounding box. This is not required but will speed up building the geometry in the duplicated chunk.

4) Build geometry in the duplicated chunk with the desired quality.

5) Because the alignment of the photos between the two chunks was not changed the two chunks will have the same alignment. To have the chunks show in PhotoScan as aligned run Align Chunks for the two chunks using the Camera method. This should only take a few seconds.

General / Re: Best F-Stop for the best Data?
« on: September 24, 2013, 10:20:33 PM »
"Best" F-Stop depends on the shape of the subject and the lens characteristics, and can be a tradeoff between sharpness at the focused distance and depth of field. Most lens tests use flat subjects for testing purposes and the reported f-stop for best sharpness applies only to flat objects.

The optimum f-stop for overall sharpness for objects with "height" variation will be smaller (numerically higher) than the f-stop for flat objects. How much smaller is optimum depends on the amount of height variation, the sharpness of the lens and the variation of the sharpness with f-stop. More variation in height means depth of field is more important.

Lens test results and calculations could be used to determine optimum f-stop but experimentation is probably simpler.

I recall reading that PhotoScan varies GPU usage to maximize overall speed, and the amount of GPU usage depends on GPU speed, CPU speed and memory speed. But that's only a recollection.

General / Re: fixed focal length ?
« on: September 04, 2013, 02:11:06 AM »
I've had success using "blue tape" to hold a zoom lens in one position while taking a set of photos.

I also usually turn off auto-focus and tape the focus ring in one position for all photos taken at approximately the same distance from the subject..

General / Re: The Transform between length in 3D model and the truth
« on: August 28, 2013, 04:12:51 AM »
Geometry created in the standard edition of PhotoScan has arbitrary units and the scale will be different in each project. The scale of the PhotoScan geometry can be determined if the actual distance between two points of the geometry is known. Measure the distance between those two points in PhotoScan, and the scale will be the ratio of the distance in PhotoScan to the actual distance. This scale will work for all distances in that particular geometry.

General / Re: exporting model to CAD
« on: August 20, 2013, 09:43:56 PM »
As discussed in another thread several days ago

For the texture on an .obj mesh imported from PhotoScan to be seen in Rhino you may also need to edit the .obj file and .mtl file.

-  In the .mtl file edit the first line "newmtl Textured" Replace "Textured" with a unique name not used by any other texture in Rhino.

-  In the .obj file edit the second line "usemtl Textured"  Replace "Textured" with the same name used in the .mtl file

Agisoft is going to try to change PhotoScan so the file name is used instead of "Textured" which should eliminate the problem.

General / Re: measure drawing?
« on: August 20, 2013, 09:40:07 PM »
When photographing a building for Photoscan identify several features which should be easy to identify after a mesh is created and textured. Measure and record the distances between these features. Process the images in Photoscan and obtain a mesh with texture. Export that mesh (I use .obj) and import it into AutoCAD, etc. Once in AutoCAD align the mesh as desired. Then select two of the features on the mesh which you previously identified and measured the distance between on the actual building. If you have several sets of features to select between use the ones which are largest distance apart which can be clearly identified on the building and which are the longest distance apart. The mesh can then be scaled so that the distance between the features in AutoCAD matches the actual distance. The scale ratio is (Distance on building) / (Distance on mesh as imported).

General / Re: Rhino doubts
« on: August 14, 2013, 11:04:51 PM »
I have scaled and rotated .obj meshes in Rhino which originated in Photoscan and were imported into Rhino. The texture is not affected by the scaling or rotation and continues to work properly. I've also trimmed the meshes and the texture continued to work properly.

General / Re: Export for Autocad
« on: August 13, 2013, 03:32:01 AM »
.obj export ??? It's what I use for exporting to Rhino.

General / Re: Rhino doubts
« on: August 13, 2013, 03:30:00 AM »
The chunk name cannot have any spaces for the .obj files to work properly. I don't think special characters are allowed either.

You may also need to edit the .obj file and .mtl file.

-  In the .mtl file edit the first line "newmtl Textured" Replace "Textured" with a unique name not used by any other texture in Rhino.

-  In the .obj file edit the second line "usemtl Textured"  Replace "Textured" with the same name used in the .mtl file

Agisoft is going to try to change PhotoScan so the file name is used instead of "Textured" which should eliminate the problem.

Bug Reports / Disabled photos used to Align Chunks
« on: July 23, 2013, 06:18:39 AM »
Disabled photos are used by Align Chunks when the Camera Based method is selected. This has caused problems for me when I have disabled an obviously incorrectly aligned photo/camera in a chunk.

Is "Disable" for a photo in a chunk intended to only apply to certain tasks? If so what tasks use photos which have been set to "Disable"?

Pages: 1 ... 10 11 [12] 13 14