search

alicevision / meshroom-manual

Meshroom manual and documentation
https://meshroom-manual.readthedocs.io
39 stars 30 forks source link

Video sequences versus individual shots #20

Open skinkie opened 4 years ago

skinkie commented 4 years ago

After figuring out that video sequences 1080@15p have a very good reconstruction percentage I wonder if this could be a first thing for everyone. For me I recorded 60 seconds of video, by walking around a statue and created 75 high res (16MP) photo's afterwards.

The photo's can do 66/75, 18k points. The movie+photo's (and some quality impromevents) do 100k points for 705 photo's, all image in the image sequence can be located. I think the documentation should suggest this workflow, but might also suggest what kind of settings for featurematching should be set up.

diffence-video-image

jonititan commented 3 years ago

I also use this workflow with ok results. I used ffmpeg to split but the files but am now looking at importing the videos directly as it looks like that does now work. However with video i'm not sure if exposure and focus are in the metadata? I thought that was the advantage of photos?

Faceless-Fan commented 3 years ago

Can someone help with adding a camera to the DB please? I am using my phone for both Photos and MP4 these are the specs https://www.devicespecifications.com/en/model/c29352e7

natowi commented 3 years ago

@Faceless-Fan This is not the right place to post this. See https://github.com/alicevision/AliceVision/pull/1009