Messages

1

General / Re: HANDHELD photogrammetry, high ISO, AI denoise... Doable/Advisable...?

« on: May 02, 2023, 12:56:53 AM »

I shoot with a Panasonic GM-5 at ISO 3200 - plenty of noise which is tamed by denoising. All hand held - hence the high ISO for a high shutter speed and good depth of field. I take lots and lots of pictures. The results are always fantastic. Just so long as the image is sharp - that's the most important thing.

2

General / Re: Wide angle lens for full frame Nikon mirrorless camera

« on: December 11, 2022, 04:58:18 PM »

Thank you Kiesel.

Any additional links for photogrammetry are welcome!

One question: Does sensor resolution (e.g. 44MP vs 24MP) matter a lot for architectural photogrammetry or lens quality is more crucial?

Oddly (or perhaps not so oddly) I have sometimes had better results resampling the original images at a lower resolution (60% to 40% and stripping the metadata) and performing the photogrammetry with those. In my experience (confined to sculpture), higher resolution is no guarantee of good results. It comes down more to how many pictures there are and how much overlap there is. The number of pictures is quite crucial.

3

General / Re: Model export quality

« on: November 28, 2022, 11:46:30 AM »

I always use Wavefront for the export format. I've never had a problem importing into Blender (v. 2.90.1).

4

General / Re: Building a mesh from Dense Cloud vs Depth Maps vs Tie Points

« on: November 28, 2022, 03:38:45 AM »

It's a lot faster without the extra step of generating a dense cloud and the results are superior - at least in my usage. I haven't used a dense cloud since Metashape standard edition was released with depth map capability.

5

General / Re: Arrgg.. Can we get proper 0,0,0 and 0 rotation position for exporting to blender

« on: July 03, 2022, 04:40:50 PM »

The thing that annoys me the most is that MS and Blender have their X-axis at 90 degrees to each other. I always have to do a rotation of -90 in Blender to get the model to the correct orientation.

6

General / Re: Build Mesh - Out of Memory

« on: November 18, 2021, 12:38:18 PM »

I always found this to be a problem using the dense cloud.

You can export the dense cloud, reduce the number of points in a cloud editor, and then import it back in to MS.

Or use depth maps and dispense with the dense cloud.

I use the latter because the results are superior, it takes less time, and there are no OOM problems.

7

General / Re: Photo Pane Show or Hide Points Question

« on: October 04, 2021, 09:13:58 PM »

I assume they are points that were not matched.

8

General / Re: When to run "Reduce Overlap", and what to follow it with?

« on: September 08, 2021, 07:22:38 PM »

I've also been running Optimise Cameras afterwards... Is this needed?

Yes...but the primary advantages of accuracy and the outputs used to measure the changes recursive optimisation brings can only be fully realised and exploited in the Professional version.

For representative shape models the Std version creates I think it's of limited value - unless anyone knows differently?

I've used the optimisation (standard edition) and it appears to make little to no difference. So perhaps there is an element of truth to this.

9

General / Re: When to run "Reduce Overlap", and what to follow it with?

« on: August 28, 2021, 02:23:04 PM »

I can't say I've found it useful for sculpture. I have huge amounts of overlap in order to make sure I have no holes or badly meshed parts of the model. If I use "reduce overlap" on any setting the results are nearly always worse. I've decided it is not worth my time experimenting with this feature and just put up with extended processing times using all the photos to get a result I know will be accurate.

Perhaps it works better on UAV work?

10

General / Re: I-PHONE XR PICTURES FORMAT TO LOAD IN METASHAPE?

« on: July 06, 2021, 07:15:52 PM »

Remove the EXIF data from the images. It's only used as a guide by Metashape and so long as you have a lot of images it isn't really required.

11

Bug Reports / Re: Processed thumbnails load too slowly

« on: July 06, 2021, 10:00:06 AM »

I've only just seen you reply.

Thank you very much for considering a possible fix.

cbn

12

General / Re: Looking for papers about the process of photogrammetry

« on: June 24, 2021, 12:17:08 PM »

You could have a look at the Articles section of this website - quite a few things there which I think would be of use to you.

In particular, my article on photogrammetry on the process and application applied to English parish churches and their contents:


https://www.academia.edu/41196210/The_application_of_computational_photogrammetry_to_the_study_of_churches

13

General / Re: image stabilisation - yes or no

« on: June 16, 2021, 11:53:42 AM »

Good points!
But 400 ISO is the max for my APSC-cameras (and weight + depth-of-field are constraints for higher mounted cameras).

Yes indeed, maybe someone could come up with a cool concept for IS-comparisons:
i.e. a reproducible robotic motion rig

I was thinking along those lines. The rig would have to jiggle the camera like a human would. Not beyond possibilities - but it would be expensive. There must surely be a better way to test this.

Possibly determine how many pixels shift there is and then crop a series of images with random offsets by up to that number of pixels to make sets of pictures to use to make the models. One would need to strip the Exif Metadata from the images - but that's no big deal. I usually strip it from mine and find MetaShape works just as well.

14

General / Re: image stabilisation - yes or no

« on: June 16, 2021, 12:38:39 AM »

I'm sometimes confronted with projects, where there is little chance to optimize the image capture process (i.e. using a long monopod without the possibility for additional artificial lighting).
Images with motion blur in such a situation almost certainly introduce more reconstruction errors than IS-images?!

What about increasing the ISO?  I have mine very high and most of the noise is removed when I process the raw files to jpg.

As to IS, it would be an interesting experiment, but probably hard to do because you can't really replicate the exact camera positions IS/non-IS if hand-holding the camera. So any models will not really be directly comparable for accuracy.

15

General / Re: image stabilisation - yes or no

« on: June 11, 2021, 10:45:32 AM »

Yeah if it's not on a tripod there is no reason to not use IS.

IS shifts the image. I would have thought this would be detrimental because the parameters of the lens would change shot to shot.