Hello everyone! I’m used to work with videos and photos for medium quality reconstructions but I would like to move to the next level and work on more professional reconstructions, historical sites in particular, and mainly sculpted caves (indoor).
I also plan on using these HQ models in Unity later on.
What pro hardware/workflow would you guys recommend? Is a Lidar Scanner worth it or is it better to stick with photos?
Thanks!
Usually, the best way to scan these types of environments is with a terrestrial scanner like Leica RTC 360 with some camera position to enhance texture detail.
a setup with a Nikon DSLR and Ring Flash will work too, but ideally, you should capture your images in 2 heights, 75% vertical and horizontal overlap, and in 60° to 75° stereo pairs. Try to not shoot standing on the same camera position just changing the angle of the shot (this will produce alignment problems).
Also, the issue with a ring flash will be that object that is far away will be underexposed, which can generate problems with texturing afterward in the post. Additional strobe lights on tripods would help, but it will be difficult to move them, expose them correctly so anything will be underexposed/ overexposed, and at the same time thinking about any camera positions you have forgotten to capture during moving forward. This is where a Leica RTC 360 will simplify the whole workflow.
Also if that’s possible you could generate targets through RealityCapture and place them through the whole tunnel to help with the measurements and the registration process. Will you also use a total station?
Thanks for your reply and useful insights!
Terrestrial scanner is the way to go then. I’m quite a beginner when it comes to large scans set up.
Do you have any good tutorials ressources to share that cover the whole workflow working with Lidar Scanners and importing materials in RC?
From doing some research it looks like best results would be mixing lasers and photos. But maybe a good Lidar Scanners would be enough to get accurate positions and high quality textures?
Also I’m not familiar with total stations, where would this hardware go in the workflow?
Hello,
You can find all of our tutorials here:
https://support.capturingreality.com/hc/en-us/categories/115000560989-Tutorials
These are the ones that I would recommend:
https://support.capturingreality.com/hc/en-us/articles/360013354939-How-to-merge-drone-images-with-terrestrial-laser-scans-
https://support.capturingreality.com/hc/en-us/articles/4402190575634-Control-points
From doing some research it looks like best results would be mixing lasers and photos. But maybe a good Lidar Scanners would be enough to get accurate positions and high quality textures?
I would argue that there is no such thing as a LIDAR scanner with color capture good enough for what you’re trying to do. My FARO Focus for instance captures a very high resolution panorama but it only exports a color value per point. So when you finally generate your texture maps in RC you’ll end up with a very watery, blurry mess rather than a crisp photo-real asset. The higher your scan resolution the crisper your color will be though. The other downside is that these devices aren’t like DSLR or mirrorless cameras in that you can’t set white balance, exposure or shutter speed. All settings are auto as far as I know. This ultimately leads to varying brightness levels on your textures. Using the Theta Z1 is a solution to this last part however (it plugs into the Faro Focus).
I suggest you do both if you can afford it. If you do it may not be a bad idea to look into some markers (check the ones you can generate/export in RC).
Ok thanks for the input! I will test out a mix between LIDAR and photos. What pictures resolution is the sweet spot for you?
I also started working with some drone footage for historical sites, so for future projects I will definitely test out the mix workflow.
(Thanks Matej for the tutos links by the way!) https://support.capturingreality.com/hc/en-us/articles/360012407600-How-to-merge-drone-images-with-terrestrial-laser-scans-inside-of-RealityCapture-workshop-
What pictures resolution is the sweet spot for you?
Anywhere between 20MP and 45MP I think. Anything above that and you’re running into RAM issues (or just SUPER long alignment/registration times. And why go below 20MP …Ultimately it’s not so much the size of the image but rather the quality of the sensor and lens.