Shooting scenarios (or tips) for making 3D model of building on narrow street

[jkova96]

Hi dear Agisoft users,
Here's my question for all of you, I want to make a 3D model of building which is on narrow street, I mean so narrow, maybe 2 meters from building to building.
I don't want to use 360 camera cause I get really bad result. Could you tell me what would be best practice to achieve that.
BTW..I made similar scenario, only for test, not near actual building which I want to make in 3D but similar to that scenario - between 2 buildings.
Only, I don't know if there clear guidelines how to do this maybe using just one camera, like phone camera or drone. I shoot photos like for making panoramas.
I'll show you my shooting strategy below, I mean...I get only sparse point cloud but I think that's good one. Just used drone in my hands and rotate around it's vertical axis until making whole circle.
I used few stand points, actually five of them.
Could you tell me if that's good approach, for test I did not change altitude for shooting rows. I leave PDF Model in which you can clearely see approx. shooting positions and PDF report from Agisoft. In Agisoft User Manual I found that this method is not good, if I understand well... can someone tell me why? (https://www.agisoft.com/pdf/tips_and_tricks/Image%20Capture%20Tips%20-%20Equipment%20and%20Shooting%20Scenarios.pdf)
THANKS! I APPRECIATE ANY HELP!

[Paulo]

Hello Jkova,

I think that maybe you can try using rolling shutter for your camara. Using same position for drone and rotating it to take many pictures will make it very difficult to solve the camera position but using many stations to do this could give you good results..

If you can share a data set I could have a look at it....

[Bzuco]

In user manual the 360 degree method is marked as incorrect mostly from reason that it cannot produce enough photos from different angles from only few shooting positions.
There is universal rule: every surface part you want to have in pointcloud in good quality should be taken from at least 3 different angles. Distance from surface also matters and it depends on how big is the resolution of photos, what is the lens focal distance and how precision final pointcloud you need.
In your case you are limited by the narrow street between buildings. If you need only one building, then set the drone position as close as possible to the opposite building facade and take only 5 shots in range <90°(at 18°steps), another 5 shots in the same range but looking slightly up, and another 5 looking slightly down. Then move horizontaly and verticaly 1-1.5m to next shooting position.

According to the screenshots you provided, F2.8 and 1/15 - 1/30s might be not ideal. Try to increase ISO to 200 or 400 and set F to 4.0 for sharper images in corners. Sparse pointcloud is already looking good so far.

[jkova96]

Hi @Paulo and @Bzuco,
Firstly, sorry for late reply...
Thank you, both of you for providing me how to capture building when looking to the narrow street.
Here's my data set - via this link: https://mega.nz/folder/RwxD0K6L#tUQxy3IYSAqxVFnbfsF1kg  (including photos, psz file (for version 1.0), agisoft report, my measurements on approx. stations for shooting photos and excel table exported to PDF with calculated coordinates of these stations)

To Paulo, you said:

If you can share a data set I could have a look at it....

Answer: Definitely, just download psz file - here: https://mega.nz/file/A94jzRQR#kbHmUJs_GuSuw5WdFbkQekcAihXY6kNJSha79Iox0nI  Or if you need everything available you can find it on link mentioned at beginning. I did not mention, this is just a test, so I did not move my drone vertically - just for test whole thing was captured approximately on same level.

[Paulo]

Hi kova,

I looked at your dataset and was able to align with Z up by disabling your marker reference and using rotate model to walls vertical and ground horizontal. Attached is a creen copy of model created at medium...To get a correct scale, I used distance  between marker 2 and 3.

But as Bzuco said, a better scenario would be having 2 cameras on a 2m telescopic tripod and taking 2 pictures straight at wall, 2 at -45 deg left and 2 at 45 deg right and then moving 1.5 m and then 6 more ....  And be aware of Depth of Field in vey narrow streets like the Callejon del beso in Guanajuato where legend says a man and a woman kissed from opposing windows in the street...

PS. your input marker coordinates were in cm instead of m.

[jkova96]

Hi @Paulo,
Sorry for late writing!
Thanks for your advice, but I'm little bit curious, or maybe a skeptic.
How did you, both of you, came to this conclusion? Cause, I did in "my way" with a lots of pictures around one point...maybe I'm doing something wrong...I don't know.

Did you have some scenario in your carrier when you done capturing process according to the scenario you and @Bzuco described?
If so, I wonder can you share with me some dataset, cause I wanted to know if that's still good enough - but in practice.
Please, if you have some project with narrow streets - I want to know how exactly.

BTW...I have blueprints and coordinates expressed in cm not in meters and I forget fact that Agisoft units were in meters. I see that you figure it out...well done:)

Have a nice day!
Thanks!

J.K.

[Paulo]

Kovacevic,

I think you can look at our suggestions and do whatever you want with it but it is our advice according to  our experience...and it is free!

[Bzuco]

@jkova96
Check this video  https://drive.google.com/file/d/1xYyYWlCdZf_9hIYI_eBCAQwtl9y5Jk-T  , it is from my project. Most of the time I am shoothing facades. There is 1m grid, so ~2m distance between positions and ~2m far from facade. I always use this approach. If you can't get 2m far from facade in narrow street, then distances between positions will be shorter ~1.5m...it depends also on the lens you are using. I am using 24mm lens.

[CheeseAndJamSandwich]

As already mentioned, the manual suggests to not stand and shoot a bunch of photos from one position, then moving on to another, to shoot another bunch...


But practically, i guess it's easier to not miss an angle when actually out in the field shooting...

We know that photogrammetry loves seeing thing from different angles, so how about you just move along the wall a small, but meaningful amount, say 10-20 cm, between each of your 16 photos (Bzuco's number of shots)... I.e. enough that when you've done the 16 shots, you're along the wall enough to where you'd be shooting the next 'batch' of shots...  Easy to manage, remember.  Shoot, shuffle, shoot, shuffle, shoot, shuffle, shoot, shuffle...  You'd get in the swing of things pretty quick, i'd guess.

And not only doing the 'shuffle' sideways, you could also vary the heights of each shot too...  Perhaps splitting the photos into 4 main levels, knee-height, waist-height, head-height, and overhead height...  And why not alter those by 10cm up and down as you take each shot...

This would mean that even two sequential shots at slightly different angles, slightly different distance along the wall, slightly different height above the ground, so greatly helping the algorithms work their magic working out the camera positions and 3D shapes.

For jkova96, from the manual, it looks like you should stand against one wall, and shoot the other... then when you get to the end of the wall, switch sides.


A while ago i tried some turntable scanning of models of insects made of leaves, twigs, seeds and seed pods...  Really not easy with those incredibly skinny legs and wings!  So now i'm thinking that it'd help a lot if i'd changed the height of the camera subtly, not just turning the turntable, between each shot at each 'plane' i shot from.  Do the turntable scanning folk do this?  Easy to program the automated rigs to do i guess!

If any of you try this shoot n shuffle, can you please get a phone/gopro to shoot a timelapse of you doing it!  It'd look awesome/hilarious i reckon! lol.
Edit:  Aaaand set it to and in sync to some excellent techno music!  lol.   Probably a good workout!

[CheeseAndJamSandwich]

I guess, doing the standing and shooting from one position/height is kinda just simulating a very wide angle lens, but one with really crap, varying lens geometry!
So you're only kinda getting 10 photos along the wall, instead of 100s.

And...  depending on the FoV of the lens, could you get away with taking one 'nadir' (looking straight at the wall) and just two 'oblique' (looking 30-45 degrees each way), and repeating these 3 shots from say 3 different heights?  Possibly giving you the detail required with drastically fewer photos taken, faster scans.

[Paulo]

OK,

it does seem weird...

[Bzuco]

From manual:
Interior(Correct) picture - there is one big downside - this will work only for small rooms e.g. 3x4m and only if you need just basic shape of room. If you want to capture also details from walls you need to be closer to that wall you are shooting, not on the opposite side of the room(that is too far for getting details from walls).

Facade(Correct) - this approach is advised, because it comes mostly from drone scenario where the most important is capture position of objects and how tall they are is slightly less importatn. Keeping 60% overlap is critical. If you miss that, you lose continuity...small downside.

Facade(Incorrect) - angles which are parallel with wall are useless(for facade itself). But angles 30-45° from nadir angle are very useful for capturing holes/window deep niches, because those angles can capture better the depth. These angles are also great as a fix for 60% overlap, which is in this case not needed. All shots are in focus and with aperture f7 and more are also sharp in corners, so not problem at all, it is faster solution and also more comfortable. Better than use wide angle lens.

Interior(Incorrect) - that example with only 2 positions and shots at 45°steps will also work in small room, but is not advised, because people could start to think that several shots in 360° from one place is enough for photogrammetry  .

There are also scenarios, where 360° shots from several positions are the best choice. It depends on the environment, the objects, and also the time you have for shooting. Metashape is very powerful and flexible, so it is not always necessary to follow the recommended practices exactly .

[CheeseAndJamSandwich]

How do the scans look if you change the View Mode to Model Confidence?
I'm guessing it'll be a series of 'good' hotspots along the walls corresponding to the shooting positions?
My underwater scans are obviously just very similar to drone scans, with only nadir shots... And the confidence under, along the 'flight' paths are very uniform, as expected. 

Note, i'm using 2x GoPros, on a 3m pole, 4-5m height, 2 second timelapse.

For this close proximity work you both have, a hybrid 'Façade - Correct' but with 2 (or more) oblique shots added, should work the best.  As you would then get much better geometry out of the door and window frames, etc.

But still, if you can shuffle, move down the wall a little bit between each shot, it can only create better, more accurate geometry.  And is really easy to do in the field.  Even if you're using a tripod, it's not like any of the shots have to be perfectly horizontal, or positioned... Quite the opposite!  They only need to be static...  And with the differing shooting heights i suggested, you could just move down the wall set at low height, then come back at medium hight, and down again with high height.
And you might be able to get away with just using a monopod?  Might give sharp enough photos, whilst making the shoot n shuffle much faster, especially on uneven ground.  And even easier to do high shots.  Using a cable release.

Wondering if it might also reduce processing times... not sure..
But i do think you could get away with a lot fewer photos and get better geometry just by moving a bit between every shot.

[CheeseAndJamSandwich]

So like this:

[Bzuco]

How do the scans look if you change the View Mode to Model Confidence?
I'm guessing it'll be a series of 'good' hotspots along the walls corresponding to the shooting positions?

Oblique angles contribute to several nearby shooting positions, so whole facade is easily in dark blur color in confidence view mode, which means ~100 combined depth maps.