PTX + Photos
Answered
Hi,
Finding this software interesting, however I have not had any luck being able to merge both laser scan data and photos.
Can someone provide me a step by step for aligning a point cloud in one component to photos in another component?
I'm guessing that the alignment process did not automatically align the scan and photos. I have followed the tutorial video but no luck.
My steps:
Import PTX files (already registered, but used draft setting just in case)
Once import is complete, I import photos.
Click align images
Several components are generated, but the largest component (449 images) does not seem to be combined with the laser scan.
I have added control points between a photo from the laser scanner and a corresponding image - Do I need to do this for several photos? If so what is the correct process from start to finish.
Finding this software interesting, however I have not had any luck being able to merge both laser scan data and photos.
Can someone provide me a step by step for aligning a point cloud in one component to photos in another component?
I'm guessing that the alignment process did not automatically align the scan and photos. I have followed the tutorial video but no luck.
My steps:
Import PTX files (already registered, but used draft setting just in case)
Once import is complete, I import photos.
Click align images
Several components are generated, but the largest component (449 images) does not seem to be combined with the laser scan.
I have added control points between a photo from the laser scanner and a corresponding image - Do I need to do this for several photos? If so what is the correct process from start to finish.
-
26forum_align_problem.jpghttps://www.capturingreality.com/forum/download/file.php?id=262097281imageFile jpgforum_align_problem.jpg
-
Hi Jon
Import PTX files (already registered, but used draft setting just in case)
= OK but if have them properly registered can use its own registration
Once import is complete, I import photos.
Click align images
= OK
Several components are generated, but the largest component (449 images) does not seem to be combined with the laser scan.
= can post fullres image so can see the components , can describe what the biggest components there are ?
I have added control points between a photo from the laser scanner and a corresponding image - Do I need to do this for several photos?
= yes you need more CPs, say 1 CP is on at least 2-3 images properly placed.
take a look how im work with the CPs and how they are spread over image, so it properly and precisely align
if they would be concentrated to small area it can cause not very precise alignment
https://www.youtube.com/watch?v=d8naLEtLqDY
If so what is the correct process from start to finish.
= yes you are correct in rough workflow description.... -
Thanks will give this a try.
So just to be sure, I'm matching control points between a photograph and the scanner photo? (each scan has 6 photos) -
Hi Jon
As see in the video how i set CPs in one component and then in second, so say 3 images in one component and 3 in second component.
so if have lasers there try put on say 3 images and 3 scans.
Best if set in different scan positions so its better conditionally set.
Try it out. after alignment check the re projection errors of the CPs -
I have had some luck with using PTX + photos and have attached the result.
The main thing I notice is that the laser scanner destroys reflective surfaces (which is expected).
Is there a way to export the depth maps (exr preferably) I might be able to correct this using a workflow I'm testing.
Thanks again for the help, Its just a process of understanding the software as it is slightly different to what I'm used to.-
29wheelhouse_withlasermeshing.jpghttps://www.capturingreality.com/forum/download/file.php?id=292097281imageFile jpgwheelhouse_withlasermeshing.jpg
-
-
Hi Jon
result look not bad :D
Is there a way to export the depth maps (exr preferably) I might be able to correct this using a workflow I'm testing.
not yet, but its faster and very well know workflows in Zbrush or 3Dcoat to clean the models out as a solid mesh.
The depth map workflow need much more cleaning steps to get good results...
next try disable images for reconstruction and get only the mode form laser scans
(select camera(s) and there is a option ENABLE IN MESHING and set FALSE/TRUE )
then make opposite so disable laser scans and use imgs for recon
and so can see where need more imgs or laser data.
the shiny stuff on laser data can be compensated with image coverage... -
Thanks for the kind words, I'm trying many experiments at this stage.
I have the option to use the depth maps in Photoscan, which I then use to remove problematic areas using some tools.
Zbrush and and 3D coat are good for fixing things and I use both packages quite a bit, but they tend to change a mesh from being accurate.
Internally does RC use EXR depths? it would be very helpful to see them - I'm also using these depths to generate nice point clouds when I dont want to use meshing. -
Hi Jon
I have the option to use the depth maps in Photoscan, which I then use to remove problematic areas using some tools.
Zbrush and and 3D coat are good for fixing things and I use both packages quite a bit, but they tend to change a mesh from being accurate.
what you mean for not accurate ? have done some comparison vs what sw results ?
Internally does RC use EXR depths? it would be very helpful to see them - I'm also using these depths to generate nice point clouds when I dont want to use meshing.
No we dont use just DEPTH maps for mesh creation -
Dear Jon,
Can you describe more precisely what kind of depth-map filtering are you doing? Maybe we could integrate such a tool directly into app.
Is there also any other reasonable use of depth maps, such as in a video post or vfx? If it is so, we can add the depth-map export. -
Depth maps are definitely used a lot in visual effects work. For instance having LIDAR information when doing match moving is really valuable (Matching camera from real to virtual). Importing the depth map into a compositor like nuke is common practice.No we dont use just DEPTH maps for mesh creation
I understand this, I was just interested in point cloud only.
There are some great tools like http://spartaproject.com/ and allow editing of point clouds very easily.
I do a lot of work for visual effects and geospatial, so any questions please feel free to ask.
If I were to suggest a feature, it would be the ability to edit the depth maps so that you can mask areas that are to exclude laser scan data. I know this seems tedious, but unless you balance the two inputs then the data will have problems. I have attached a comparison of photogrammetry only and then photogrammetry + laser. As you can see the laser version (bottom one) has holes and issues where there were reflective objects. This is expected from laser scan behaviour. Using photogrammetry only is not really an option because this would require accessing areas that are very difficult and vice versa for laser scan.
Another example is if I do an external scan + photogrammetry of a building, the windows will be problematic.
I do want to say that the software is really nice, I'm just suggesting things that the industry has problems with :)-
30wheelhouse_photo_laser_comparison.jpghttps://www.capturingreality.com/forum/download/file.php?id=302097281imageFile jpgwheelhouse_photo_laser_comparison.jpg
-
-
Also here is a comparison if I use photogrammetry or photogrammetry + laser.
The benefits of having laser scans are excellent, but not without their downsides as discussed earlier.-
31laserscan_and_photogrammetry_comparison.jpghttps://www.capturingreality.com/forum/download/file.php?id=312097281imageFile jpglaserscan_and_photogrammetry_comparison.jpg
-
-
could you recommend hardware for the laserscanning?
-
Hi finkbom
could you recommend hardware for the laserscanning?
Whats want to scann ? large buildings, small stuff or ? -
small stuff at moment small interior rooms is about the largest
-
Hi finkbom
Hope you know prices for terrestrial laser scanners, its up from 40k+ EUR for new device...
So in short, you want make scanns of interiors of buildings and small stuff like props ? -
hi did not know the price was that- yep small interior rooms(+sets) and props at the moment
-
Hi finkbom
Laser scanners are really expensive and require a fair bit of experience to use. Once you finish a few scans you then need to connect them together using "registration". Software that performs good registration is very expensive as well.
The good thing about laser scanners is the effective range can be 10-50m for good detail (medium quality scans).
The main scanners out there that everyone seems to use are:
- Zoller + Fröhlich (Excellent scanners, the 5010c is one of the best I have ever used)
- Leica P40 (new scanner, but in the same league as the 5010c)
- Faro scanners (much cheaper, portable scanners with 1/8 quality of the above).
There are also a few good scanners from Trimble but I have not used personally. The Leica P20 scanner (older scanner) is fairly good but does not have good imaging capabilities like the new one.
You are looking at 100k USD for the premium scanners, and about 50K USD for Faro (Could be cheaper now). -
thanks for the information
-
Hi Jon
except this one
Faro scanners (much cheaper, portable scanners with 1/8 quality of the above).
i agree with all info there
the quality degradation is not so high, would say but just roughly 1/2 of Leica,and its again about experience with the scanners...... -
I did a comparison between Z&F and Faro and its possible to get pretty good results with Faro but you have to wait (a lot) longer for a similar result. Ultimately you get what you pay for :)
Also the photo quality from the Z&F is a lot better. The Faro photos are not very usable due to the temperamental colour balance.
Please sign in to leave a comment.
Comments
19 comments