I'm wondering if constraints from IRL measurements and April Tags will improve accuracy of image alignment and mesh, and improve the accuracy of fine details. I'm not bothered about the scale accuracy, just the detail of the model (building) as I have noticed the some features like roller shutters on shops tend to be terribly aligned. If it's worth the effort I'll go all out...
@@CapturingReality so they must be added and the distances between them measured, and then they will improve the accuracy of alignments, in particular fine/intricate details. For example repeating fine details like roller shutters, guttering, door handles, fencing / railings etc? I'm not bothered about scale / relative accuracy, but if the effort is worth it for that sort of accuracy then I'll go out and purchase the necessary equipment. Ie laser range finder etc.
@@PaulChapman-j6v The improvement of them is also just using them as control points. It is not necessary to measure the distance between them (that is for scaling) or their coordinates (for scaling and georeferencing). For such objects only better capturing will help (also not always, as those are not ideal objects for photogrammetry).
Hello, the scaling doesn't seem to work. I detect the markers first and then align the images and then use a .txt for defining my distances. But the distance gets overridden with Calculated Distance and also my orientation is completely off even though the 4 markers are detected. What's your advice?
Hello Vishal, When using detect markers and scaling via txt file you should 1st detect your control points, then add the txt file with the distance constraint and then align images, not vise versa. If you for some reason need to import the text file after alignment you need to click on alignment again, You can also use the update button which is also under the workflow tab, but it will not be as precise as aligning images again.
@@CapturingReality Hi another question just occurred. If want to use the scaling markers, should I generate a binary mask with or without the masks? I would like to use these masks for considering only the pixel within the mask for model generation.
@@vishalanime If you want to use masks you would need to disable masks for alignment so RC could find any features in the parts where markers are placed. Enable masks only for the reconstruction. Alternatively, adjust you masks so markers will be masked in and could be also used in alignment if you need masks for alignment.
How do I get the Distance Constraint option to appear in the Align tab? I've watched the video, read the instructions, created control points, used markers, and I've still never seen the Constraint tab. Any ideas?
Hi Dana, video is from older UX. Now the Detect marker option is under Alignment tab/Optional and Define distance is under Scene 3D tab/Tools/Registration & Sparse point cloud/Define distance. For this option you need to have done alignment.
I have a small scene like your one at about 4 minutes in. With the Detection Markers on the ground. When I Calculate the model it comes out at a strange angle. How do I use tell CR that the control point from "Detect markers" are all on the ground plane (i.e. zero in height / z axis). I have tried forcing the (detected) marker control points to be Ground Controls, but this causes corrupted control points (Red Triangles)
Hello, i have question. When i done scaling and exporting model to .obj a then i opened in another 3D software, its showing me different scale than is in the reality capture. Can you please help? Thank you!
I had the same problem, but after switching the obj export preset in reality capture from 3ds max to maya, the object has the correct size even after importing it into max
I'm wondering if constraints from IRL measurements and April Tags will improve accuracy of image alignment and mesh, and improve the accuracy of fine details. I'm not bothered about the scale accuracy, just the detail of the model (building) as I have noticed the some features like roller shutters on shops tend to be terribly aligned. If it's worth the effort I'll go all out...
Hi Paul. Yes, on some places it can help.
@@CapturingReality will just having the tags / targets in various places help too?
@@PaulChapman-j6v But they need to be measured as control points.
@@CapturingReality so they must be added and the distances between them measured, and then they will improve the accuracy of alignments, in particular fine/intricate details. For example repeating fine details like roller shutters, guttering, door handles, fencing / railings etc? I'm not bothered about scale / relative accuracy, but if the effort is worth it for that sort of accuracy then I'll go out and purchase the necessary equipment. Ie laser range finder etc.
@@PaulChapman-j6v The improvement of them is also just using them as control points. It is not necessary to measure the distance between them (that is for scaling) or their coordinates (for scaling and georeferencing).
For such objects only better capturing will help (also not always, as those are not ideal objects for photogrammetry).
It's possibile to scale with a simple widget like in bleder?
No, just with mentioned methods.
Hello, the scaling doesn't seem to work. I detect the markers first and then align the images and then use a .txt for defining my distances. But the distance gets overridden with Calculated Distance and also my orientation is completely off even though the 4 markers are detected. What's your advice?
I might have figured it out. I set the pose hardness for camera priors under the alignment settings to zero.
Hello Vishal, When using detect markers and scaling via txt file you should 1st detect your control points, then add the txt file with the distance constraint and then align images, not vise versa. If you for some reason need to import the text file after alignment you need to click on alignment again, You can also use the update button which is also under the workflow tab, but it will not be as precise as aligning images again.
@@CapturingReality Hi another question just occurred. If want to use the scaling markers, should I generate a binary mask with or without the masks? I would like to use these masks for considering only the pixel within the mask for model generation.
@@CapturingReality , any comments on this?
@@vishalanime If you want to use masks you would need to disable masks for alignment so RC could find any features in the parts where markers are placed. Enable masks only for the reconstruction. Alternatively, adjust you masks so markers will be masked in and could be also used in alignment if you need masks for alignment.
How do I get the Distance Constraint option to appear in the Align tab? I've watched the video, read the instructions, created control points, used markers, and I've still never seen the Constraint tab. Any ideas?
Hi Dana, video is from older UX. Now the Detect marker option is under Alignment tab/Optional and Define distance is under Scene 3D tab/Tools/Registration & Sparse point cloud/Define distance. For this option you need to have done alignment.
@@CapturingReality Thanks for helping me find it. The information in the built-in help could use an update.
@@DanaWilliams You're welcome. We are working on updating the Help, but it needs some time.
I have a small scene like your one at about 4 minutes in. With the Detection Markers on the ground. When I Calculate the model it comes out at a strange angle. How do I use tell CR that the control point from "Detect markers" are all on the ground plane (i.e. zero in height / z axis). I have tried forcing the (detected) marker control points to be Ground Controls, but this causes corrupted control points (Red Triangles)
As you did, you need to set them as Ground Controls, add its coordinates and then align again or update the alignment.
Hello, i have question. When i done scaling and exporting model to .obj a then i opened in another 3D software, its showing me different scale than is in the reality capture. Can you please help? Thank you!
I had the same problem, but after switching the obj export preset in reality capture from 3ds max to maya, the object has the correct size even after importing it into max
Ah man ridiculously buggy which works once in a while, I hope after this acquisition by EPIC, there is documentation somewhere :O
I want to use for my city in my future game !
2 viewer love from india
From india