Woohoo! Orthomosaïcs! Thanks for this new feature.
I tried it and it seems to work pretty well. However, sometimes it yields some strange colorations at the edges. These tend to be red or green (see pictures. Also in the final render btw).
Greetings,
Robin
Thank you for reporting this, have you tried drawing a region around the problematic areas to see which images were used to make them?
Thank you so much for the quick reply! I tried your suggestion on a new run (same data) and something weird happens. The red spot dissappears in the viewport when the polygon has been drawn. However, the export remains the same: a red spot is still there!
The active images don’t seem to contain any red adjacent or at this spot (a bit red is needed to the left just outside this trench).
I’m not sure what’s happening here. Is the colour correction not working properly? After this test a second question arises: why is the viewport showing a different result after drawing this polygon?
Let me know how I can provide you with data / info!
Robin
Hi Robin,
what are your settings, when you are creating the ortho?
Hi ,
These are the settings.
Thanks in advance!
Hi, these problems could be related with your settings in Rendering method. As I assume, this is some kind of archeological spot. Do you have just aerial/drone images or also terrestrial? If you have just drone images and want to create ortho from the side of the hole, you can use Image mosaicing (general) or True ortho from textured model, which should create better results.
Hi Ondroj,
I will try it as soon as I can. I’ve just received my laptop back from the repair shop. I’ll let you know!
Hi Ondrej,
I´m back on track.
It seems the seems occurs with another dataset. A green glitch. I used the experimental setting: Image mosaicing (general). True ortho from textured model would work, I guess. However I want to stick with the mosaics. I really like this fast feature. It is one of the reasons I would prefer RC over others.
Greets,
Robin
Hi Robin, what kind of dataset was used for this ortho?
Hi Ondrej,
A set of 32 pictures were taken from about 3,5 meters height with a remote controlled camera on a stick. The object was a parking place with a crate. I was testing this technique and capabilities of RC.
I made several ortho´s with this dataset. Apart from this one, the results were great. It is odd though this happens: no green things can be found on any of the pictures in this dataset. Perhaps some ‘seams’ in the orthomaic are not rendered well in some cases?
Let me know if you need more info!
Greetings,
Hi Robin, as it is still an experimental tool, this could happen. It there a green line also on Image mosaicing - aerial? As this could be consider as some aerial images.
Hi Ondrej,
It seems to depend on resolution: when I choose ortho pixel size to be 0.002 the glitches appear. When the value is equal to 0.003 everything is fine. I tried some values in between and there seem to be a thresshold value:
0.000696 - 2 glitches (estimated optimum value)
0.002000 - 2 glitches
0.002125 - 2 glitches
0.002175 - 1 glitch
0.002500 - 0 glitch
0.003000 - 0 glitch
I understand it is experimental. It is no big deal these errors occur. I’m grateful these new features have been added to RC and the RC team is committed to solve any issues. Maybe this feedback helps tracking down a bug and improve RC. That would be great.
Greetings,
Robin
Thank you for the testing, we appreciate it. Will inform our developers about it.
I have not experienced the above issue (yet).
I am modelling the face a straight wall approximately 100 metres long using a Leica BLK3D and Panasonic Lumix G100 camera and I am very happy with the mosaic orthos I am getting out of 1.2.0.16905 RC.
My model is essentially locally geo-referenced by the Leica BLK3D. It is dimensionally scaled correctly and up is up but it does not have a geo-referenced origin or plan (x,y) rotation.
Consequently I do not get the Type option in the Ortho Projection Tool. That is not a problem as all I do is select the side face of the reconstruction region and orientate the reconstruction region by eye using the Top View.
To speed up this part of the process is it possible to rotate the reconstruction region by picking two points when viewed from the Top rather than the method I am using above ? I will look more into the geo-referencing
I would like to know broadly what the difference is between image mosaicing general and aerial and/or whether this is relevant to my application described above?
Thanks, Kev