Show Posts

This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.

Messages - Marcel

Pages: 1 ... 3 4 [5] 6 7 ... 21
General / Re: Best practices for tank scanning
« on: March 24, 2015, 11:16:10 PM »
I've attached a schematic drawing that shows what I mean about positions. Photoscan can get more information from a photo when the camera position is different (instead of only the angle being different). It is fine to rotate the camera and take two shots when the subject doesn't fit. But I don't think it helps taking 5 shots that have a lot of overlap.

ISO Low -1 might still be useful. What it does is expose the photo 1 stop brighter, but it scales back the values so it does not look brighter. The overexposed photo has less noise in the dark areas, but the highlights might suffer.   It depends on your subject if ISO Low is useful. If your subject includes a lit of white highlights (like the snow in your tank images), then ISO200 is probably better.

I played around with your project, the point cloud actually has slightly less noise when you process it at Medium instead of High, and you do not loose much detail (and as a bonus the processing is much faster).

General / Re: Best practices for tank scanning
« on: March 24, 2015, 11:32:29 AM »
Good stuff!

I see you managed to get a tripod, that makes a lot of difference.

There are still some blurred images, for example DSC_6154.JPG. Are you using a remote release cable? Because when you push the shutter button on the camera using your finger, it's very easy to move the camera too much. Even the shutter action can move the shot, so for absolute perfect results you can use Mirror Up (if your camera supports it).

I would go through all your images and remove blurred ones (like DSC_6154.JPG). This will improve the quality of your scan. Don't forget to use Optimize Cameras after you have removed those photos from your chunk, or redo the complete alignment.  Images that are very slightly blurred like DSC_6172.JPG can cause noise in your surface. You have more than enough shots, so do a very strict selection and remove anything that is not completely sharp.

Layout wise: having your camera in the same spot and shooting 'panorama style' is less useful than giving each photo a different position. You could lower your ISO to ISO100 to get a tiny bit less noise.

General / Re: lens to Sony nex 5 and Lumix GM1 and GM5
« on: March 16, 2015, 08:48:27 PM »
Why do you need two cameras? If you fly with a drone, can't you just take photos at a faster interval?

I would also think that at 150 to 300m high, you will not have enough seperation between your cameras (unless they are not pointing straight down, but at an angle so they each cover a different area).

My own experience with a dual camera setup was not a happy time. It is a pain to setup (getting exact same settings  and correct focus on both cameras). I also had a camera specific problem (shutter slap causing blurred images, but this was with a dual A7R setup). Much money was wasted.

General / Re: lens to Sony nex 5 and Lumix GM1 and GM5
« on: March 16, 2015, 07:26:37 PM »
This one is fairly light, and I think it would fit on your camera:

Not sure if that is the correct focal length for you, and it is fairly expensive.

It can be very sharp, but do purchase it at a store that has a good return policy. That way you can test it (shoot some test photos at the distance you will use it), and if the copy is slightly off you can return it for a different copy. I tested 3 of these, and one of them had a soft corner. The other two copies I tested were perfecly sharp. This was using a full frame A7R, so on your (cropped frame) Nex it's probably less of a problem.

It is a really good lens, especially considering the size and weight (120 grams).

General / Re: Best practices for tank scanning
« on: March 16, 2015, 12:08:45 PM »
Or maybe Google Drive compresses the JPEGs I am downloading, that could also be?

Tiff would indeed be lossless, but the filesize is much bigger so projects can take up a lot of diskspace.

General / Re: estimate image quality
« on: March 16, 2015, 12:07:17 PM »
The Estimate Image Quality is only a tool for the user to get an idea of the image quality. The values are not used during processing.

General / Re: Best practices for tank scanning
« on: March 15, 2015, 01:02:36 AM »
Note that the markers are interesting, but really not needed to get a correct alignment. There is plenty of small details on the tank for Photoscan to use. Unless you use them for scale reference ofcourse.

General / Re: Best practices for tank scanning
« on: March 15, 2015, 01:00:59 AM »
I've suddenly run into strange alignment problems.

I've limited model to front details of tank, since I have only 16 Gb of RAM and cannot install more in my present notebook.

project t34front

Some photos - MANY photos - are not getting aligned.

What can I do to fix the problem? More photos? From which points? Maybe some hint to PhotoScan, to help it place photos in correct places?

Not more photos, better quality photos. If you photos are sharp, then Photoscan will have no trouble aligning them.

This is how your photos look at 400% magnification (click on thumbnail):

There is a lot of noise in the photos (in the dark areas especially), to get less noise you need to shoot at ISO100 instead of ISO400. (and like I said: you need a tripod and remote shutter release cable for that).

There is also a lot of JPEG compression on your photos:

See all the blocky stuff? Photoscan has real trouble recognizing any features because of the compression.  Try re-saving your JPEGs at 100% quality.

For a quick fix you can also try to Align at "Low" quality. This will downscale the images to 25% of their original size, which has the side effect of making the images much sharper. Sometimes this helps.

Edit: The other shots you did (when there was full sun) are much better quality:

Nice and sharp, and no JPEG compression. If all your photos would be as sharp as this, there would be no problem with the alignment. These photos do have other issues however: the specular shine on the object varies with the camera position, and this is not ideal. Photogrammetry isn't easy  ;)

I'n not sure if mixing photos with such different lighting conditions will work, probably not.

General / Re: Alignment Experiments
« on: March 15, 2015, 12:43:30 AM »
Hi Marcel ,

Great test, thanks for posting your results.
Can i ask you :
How you generate the graphs of (key points & Re projection error)?
and how you covert your result into Normal map?

Thanks again.

I generated the graphs using Google Sheets, with the data I got from right clicking on a chunk and using "Show Info".

You can create a normalmap by meshing your point cloud, and then converting it using XNormal. But it's a slow and painful process and you need a ton of memory for larger point clouds.

General / Alignment Experiments
« on: March 14, 2015, 12:35:43 PM »
I did some tests to find out what the the influence of the Key Point Limit is on the alignment because it always bugged me not knowing what what Key Point Limit was high enough.

This is the project I used for testing, a scan of an rusty sewer lid:

It's photographed with a D800 and a 35mm lens, and has a total of 52 photos at 36MP resolution.

I ran alignments with the following settings:

Accuracy: High
Pair pre-selection: Generic
Key point limit: from 1000 all the way up to 320000
Tie point limit: 0 (no limit)

Quick explanation of the settings:

At High accuracy, Photoscan uses the full resolution photo  (Medium would use the image at 50%, Low at 25%).

Pair Pre Selection
With Pair pre-selection set to Generic, Photoscan will make a quick pre-scan to see which photos share the same view. If photos do not share the same view then it makes no sense to compare the points in the photos. This makes the alignment much faster (and with good quality photos it has no impact on quality at all)

Key Point Limit
The maximum number of points Photoscan will extract from each photo. For a high quality 36 Megapixel photo the maximum number of points that can be extract is usually around 240000. For a 21 Megapixel photo this is generally 180000 points.

Tie Point Limit
This setting has been added not so long ago. I am not completely sure, but I think when this setting is active Photoscan makes a pre-selection based on (visual) quality of the extract points (so it only compares the highest quality points).

For example, if your Key Point limit is set to 40000 and the Tie Point Limit is set to 1000, then Photoscan will first extract 40000 points for each photo, and only keep the best 1000 points. These 1000 points per photo are then used for the alignment calculations.

This would speed up the alignment a lot because there is only a fraction of the points to compare, but since I am not sure about this setting I have set this value to 0 (=no maximum).

I ran alignments for Key point limits from 1000 all the way up to 320000, and put the results in some graphs.

Number of Points in the Sparse Cloud after alignment:

A higher Key point limit means more points in the Sparse Cloud. It starts levelling off after 240.000 points, because the maximum number of points that can be extracted is being reached for some images.

Alignment time:

The alignment time is pretty much linear, which was a surprise to me because I expected it to be exponential. The graph varies a bit, because I was using my computer for other things as well so the times are not completely accurate.

Reprojection Error

The next graph is the reprojection error:

The Reprojection error is a measure of accuracy of the points, measured in pixels. When you think about this, these values are pretty impressive: Photoscan is able to align the cameras with a sub-pixel accuracy!

The reprojection error for an alignment with 40.000 points is almost twice as big as for an alignment with 120.000 points (0.7 vs 0.4 pixels). But we can optimize the Sparse Point Cloud and redo the camera alignment.

To do this, I have used Edit->Gradual Selection->Reprojection Error with a value of 0.5 and removed those points. This gets rid of all points with a reprojection error larger than 0.5 (about a third of the points in the Sparse Cloud) Then I used Tools->Optimize Cameras to redo the alignment of the cameras. After optimizing, the graph for the reprojection error looks like this.

So after optimization the reprojection error is pretty much the same for all Key point limits (and I did not loose that many points) . The reprojection error now has values around 0.25, so Photoscan managed to align the cameras with a precision a quarter of a pixel!

I tried optimizing the point cloud even further by using "Gradual Selection -> Reconstruction Uncertainty = 8", but the reprojection error actually increased slightly after these optimizations. I don't think the accuracy is actually worse (since I deleted bad points), so maybe the reprojection error is not the best indicator of the accuracy of the alignment?

Dense Cloud Quality

All this talk about Reprojection Error is pretty theoretical, what is the effect on the Dense Point Cloud?
I did a Dense Cloud reconstruction at "High" quality for the various alignments. I converted the result to a normalmap, because we know from experience that a normalmap shows problems really well.

10000 points: there are some cameras that are excluded from the Dense Cloud reconstruction, so the Dense Cloud has some holes. Also, there is some general noise all over the scan.

20000 points: looks much better, but there is a very slight noise (only visible if you overlay the normalmaps).

40000 points: scan looks good

more than 40000 points: no visible difference in quality.

I also did a comparison with the Dense Cloud build at the Ultra quality setting. There wasn't any visible difference either.


The default value for the Key point limit (40000 points) seems to be well chosen. I don't see any improvement in quality of the Dense Cloud when using an alignment with more than 40000 Key points. If you look at the values of the reprojection error after optimization, this actually makes sense. The values are all under 0.3 pixels, which is well under the size of the details in the Dense Cloud. The alignment might be 0.1 pixel more precise, but this is way below the threshold of visibility (I would estimate that details in the Dense Cloud are for structures at least 2-3 pixels in size).

I will probably run my alignments at a higher Key point limit anyway (maybe 120000 points), just to be sure that my alignment is as accurate as possible. The Alignment doesn't take that much time compared to the DC reconstruction, and it gives me that warm safe feeling of doing it right.  :)

Please note that your results may vary: this is a very specific type of scan where all the photos are in the same plane. A more 3D object might need more points for a good alignment. Also, the photos in this project have almost perfect sharpness, so Photoscan has a very good input. If the photos would be less sharp, more points would be deleted during optimization (and using more points might be useful).


General / Re: Avoid Unsharp/Stretched Textures?
« on: March 13, 2015, 10:25:56 PM »
Doe the parts that are stretched have good photos, or are those parts photographed only at an angle?

General / Re: Licensing for network processing
« on: March 10, 2015, 02:58:58 PM »
You need a Pro license for every network node as far as I know.

A separate 'network node license' that only does network processing would make more sense I think.

General / Re: Best practices for tank scanning
« on: March 10, 2015, 11:13:49 AM »
Screenshots or it didn't happen  ;)

General / Re: ARM/Embedded Support? Or Pipe Dream?
« on: March 10, 2015, 11:13:23 AM »
What would be the advantage?

Also, Photoscan needs a lot of RAM, and these devices generally don't have more than 4GB?

General / Re: Calibration/Accuracy possible with Photoscan?
« on: March 10, 2015, 01:33:49 AM »
I think it is possible, but it will certainly not be easy. Smooth black plastic is not ideal for photogrammetry. You would need to add some sort of pattern to the surface to make it easier for Photoscan to reconstruct the shape.

Because it is a fairly small object you probably need to use a macro lens, and you will run into trouble with depth of field being too shallow because you are so close to your object. You can use focus stacking to get around that, but that adds a lot of work (and more things that can go wrong).

Another problem is the deep crevices, Photoscan will have a difficult time to get a good reconstruction of those areas.

The only reverse engineering work I have seen using Photoscan was large scale (large metal parts for ships). The only way to be sure would be to try yourself? It's not just a question if Photoscan can do it, it's also the question if you can create photos that are high enough quality.

Pages: 1 ... 3 4 [5] 6 7 ... 21