RealityCapture Tutorial: How to take photos for RealityCapture
ฝัง
- เผยแพร่เมื่อ 11 ก.ค. 2024
- In this video, you will find recommendations on how to take photos for RealityCapture.
This video is mainly intended for beginners who do not know where to start, but also advanced users may find certain tips useful for their workflows.
It will cover basic tips, schemes, and examples on how to capture various objects:
00:46 Do not limit image count
1:27 Use Highest Resolution Possible
2:09 In app model resolution calculator
2:29 Do not lose focus
2:54 The exposure triangle
3:35 Each point should be visible in at least two images
4:11 Shoot a full loop around the object and also from several elevations
4:44 Do not change the viewpoint more than 30 degrees
4:59 Follow the shape of the object
5:32 Maintain high overlap between images
6:17 Do not shoot panoramic pictures
7:22 Avoid object movement (e.g. due to wind)
7:46 Avoid transparent or shiny objects if possible
9:32 Avoid featureless textures
10:38 Keep constant lighting conditions
11:42 Photographing interiors
12:44 Drone mapping
13:36 Using masks
Links to tutorials mentioned in the video:
RealityCapture tutorial: Scan objects from all sides using Masks.
• RealityCapture tutoria...
RealityCapture tutorial: Merging exteriors and interiors of buildings
• RealityCapture tutoria...
RealityCapture Webinar: Combining Z+F laser scan with images inside of RealityCapture
• RealityCapture Webinar...
Download RealityCapture (PPI) for FREE here: bit.ly/2sotr7h
Try it yourself with free sample datasets: bit.ly/3cRBfQp
Follow us on:
Instagram / capturingrealitycom
LinkedIn / capturing-reality
Facebook / capturingrealitycom
Twitter / realitycapture_ - วิทยาศาสตร์และเทคโนโลยี
![[TH] Esports World Cup : Knockout Stage Day 1](http://i.ytimg.com/vi/ES-vr-FarFg/mqdefault.jpg)
Thank you and have a great day.
Nice video! helped me a lot
Estrela brilhou lá no auto tô mar
If you have a drone/flight mapping project, which camera angle do you recommend, straight down/90degrees or an angle like 70degree. In the case of angle, what is the recommendation for the flight plan to maintain the overlapping? Maybe there is already some documentation you can point me to?
Hi, it depends what your results will be for. If you want to have just a map, then you can use 90 degrees. When you want to have there also some 3D objects, then the angle is good to use. The overlap between images should be minimally 75%.
helpful video, thanks. does the sensor of the terrestrial camera have to be identical to the one of the drone ?
Hi Willy, it is not necessary to have identical camera sensor. You can use different cameras in your project. If you do so, there is good idea to use Group calibrations w.r.t. by Exif in Inputs.
@@CapturingReality And does that include using differing datasets (in the EXIF) with and without GPS data? For example, mixing sUAV images and DSLR with no GNSS antenna?
@@vkbellis Yes, it could be used also in this way.
🎯 Key Takeaways for quick navigation:
00:31 📷 *Shoot more photos than you think you need for a project; environmental changes or missing parts can't always be easily reshoot.*
01:34 📸 *Use the highest resolution available, but even lower megapixel cameras can yield good results for smaller-scale photogrammetry.*
02:32 ⚙️ *Adjust camera settings for exposure, focus, motion blur, and noise; manual mode provides more control.*
03:56 🌐 *Ensure multiple images capture each point of the object; more images improve reconstruction quality.*
04:58 🔄 *Maintain consistent angles and distances while capturing images; follow the object's shape for uniform detail.*
05:58 🔍 *Aim for at least 60% overlap between images; avoid shooting panoramic pictures for 3D model creation.*
06:56 🌐 *Ensure camera poses encompass the entire object; constant lighting conditions aid in better model alignment.*
08:00 💡 *Adjust photographing techniques for shiny objects; use different sets of images to capture textures effectively.*
09:49 🎨 *Utilize markers or added texture on featureless surfaces for better alignment and reconstruction.*
10:49 🌞 *Maintain consistent lighting conditions to avoid gaps or incorrect alignment caused by shadows.*
12:52 🛩️ *Utilize flight planning applications for drone mapping, adjusting altitude, speed, and settings for proper photo capture.*
Made with HARPA AI
Great video, it helped me a lot. But I have a question the video mentions scanning a reflective object, you can shoot two sets, one spray to generate the model and one to display the texture; but how do I do that in the software.
There you can see, how it was processed: th-cam.com/video/D7iY146cU2Y/w-d-xo.html
Basically, both sets are used in the alignment process, sprayed images are used in meshing process, original images are used in texturing process.
Oh, thank you.
Nice Video! how to align the two set of photos of the shinny object?
Hi Raku, shiny objects are still a problem for photogrammetry. Simply, you need to reduce this shininess. You can use some kind of spray or other texture creator. Also it depends, how is the light condition during capturing and scanned material.
would you recommend 'Flight planning application' for realitycaputre? thank you
Basically it doesn't matter which one you will use. It more depends on the settings which you set there and they should correspond with the mentioned parameters from the tutorial.
Hello, how does it work to capture the bottom of an object for example like a shoe? Can i flip the shoe to take pictures of it? Thanks!
Hi, you can watch this video, which should help you with this problem: th-cam.com/video/xxsdbzBwWLE/w-d-xo.html
@@CapturingReality thank you!!
One other question please, I took 300 pictures but the software is only recognizing 115 of them after alignement. How do I make sure all the pics I take will be taken into account? Thank you very much
@@Norma-so7mu Hi, it depends, how are your images taken. If there is insufficient overlap, then the images won't be aligned together. You need to have minimally 75% overlap between images. It also depends, what type of object are you capturing. If it is very shiny or featureless, then it is hard for the software to find tie points on the images and then it is impossible to align them.
V minulom videu bola nejaka chyba ? Asi ten zvuk mikrofónu inak super video dobre vysvetlené
Ano iba zvuk sa opravil. Obsah sa nemenil a dakujem.
Can i have record around object and export png for scan ???
Dear Duong, thank you for your comment. As an input you can also use a video. Regarding exporting files, you can view it at this link support.capturingreality.com/hc/en-us/articles/115001529012-RC-s-output-export-file-formats
It would be amazing if the software took in 360 videos
There is a feature request for this.
Should you change exposure of the camera or keep it constant? E.g some areas might be in shadow, should you expose for that or not?
Hi, you should. Basically, it is connected to Keep constant light conditions part of the video. Or the images can be modified after the capturing.
@@CapturingReality great, thanks.
@@CapturingReality Sorry, one more question. Would a 24mm lens be ok, or will the barrel distortion be too much? I was thinking of using a 50mm, but wonder if 24mm is ok.
@@tomlaw1507 It should be OK to use (24 mm - 50 mm (full frame))
🇧🇷🇧🇷🇧🇷🇧🇷🇧🇷🇧🇷🇧🇷🇧🇷🇧🇷
I get a lot better quality with just one circle