faster workflow for adding ground control points

Answered

Comments

19 comments

  • Avatar
    Wishgranter
    Hi Bill

    yes, I will prepare it as PDF for you in the next few days.
    0
    Comment actions Permalink
  • Avatar
    Bill Isenberger
    Wishgranter wrote:
    Hi Bill

    yes I will prepare it as PDF for you in next few days.

    Great! I have a couple new projects I took photos for to try this on.
    0
    Comment actions Permalink
  • Avatar
    martinb
    One more video. Here I'm using arrows to navigate between images.

    Rules:
    1. Click ALIGN\Control Points button to activate control point tool.
    2. Click-and-hold mouse over a 2D image view to place the control point. It will zoom in the image, so that you can precisely place the CP. Use mouse wheel to change zooming. Place the point.
    3. Use left-right arrows to switch between images in 2D image view.
    4. Place the control point in some other image. It will place the same control point in this image.
    5. If you have a layout with more 2D image views, then system will automatically zoom all images which contain the same control point so that you can more precisely place the CP.
    6. Right click to start placing another control point. You will see in 1Ds under control point node that right clicking will deselect the control point. Control point tool places the point which is selected in 1Ds.
    7. Another method to start placing another control point is simply to click again into the same image - as if you were trying to place the same control point into an image twice. System will detect that and it will start placing a new point.

    Drag & drop
    You can ctrl+drag control points from image to image. You can drag&drop control points from 1Ds to images. You can drag&drop image measurements between control points (merge tracks).

    This is a short example:

    0
    Comment actions Permalink
  • Avatar
    Bill Isenberger
    Hi martinb,
    I have seen that video. Really helpful for adding components.

    My project was flown in 8 sections. Each section has 10 GCP targets recorded with a RTK-GPS. When all of the sections are combined the project has 1000+ images and 80 GCPs. Is there a faster way to mark the GCPs in a project like mine?
    Thanks,
    Bill
    0
    Comment actions Permalink
  • Avatar
    jnowat
    Is it smarter to place more than two control points (use more than two images for CP #x)?
    Is it smarter to use more than one control point per image?

    Thank you
    0
    Comment actions Permalink
  • Avatar
    Wishgranter
    Hi Jnovat

    Is it smarter to place more than two control points (use more than two images for CP #x)?

    It is best to use 3-4 CPs per image ( as it helps to define the camera position better ) and at least in 2 images.
    More is better but you do not need to place say 5-10+ CPs per image ( but there are cases when you need to do that :D ) when it is a complicated subject like buildings ( interior + exterior )

    Is it smarter to use more than one control point per image?

    Yes, more is better...

    As you have seen in the Martin’s video or mine (https://www.youtube.com/watch?v=d8naLEtLqDY)
    Look at how I am spreading the points in images, how many CPs are used and how many projections are created...
    0
    Comment actions Permalink
  • Avatar
    Bill Isenberger
    Hi Wishgranter,
    Is the video posted by martinb a few post up the fastest way to add GCPs to large projects? Most of my projects are for topographic mapping and orthographic photos.
    Bill
    0
    Comment actions Permalink
  • Avatar
    Wishgranter
    Hi Bill

    yes and no :D

    It is mostly about what workflow is the most handy (easiest to work with), there are multiple ways to do it.
    One more, which was intended for large scale work, will be prepared but I need to discuss it with developers first, as some minor adjustments need to be possibly done.
    0
    Comment actions Permalink
  • Avatar
    Bill Isenberger
    Wishgranter wrote:
    Hi Bill

    yes and no :D

    its mostly about what workflow is most handy ( most easy to work with ) there are multiple ways to do it.
    One more what was intended for large scale work, will prepare but need discuss it with devs, as some minor adjustments need to be eventually done.

    Looking forward to this. I created an ortho on my dataset without using the GCP's. Looks pretty darn good! Can't wait to plugin in the GCP's and kick out the orhto and DSM. I've got to do some field work the next several days so it ooks like the free license will be expired before I get back. I'll be checking the forum and my email to see what direction CapturingReality takes next.
    Thanks for the help and guidance Wishgranter ,
    Bill
    0
    Comment actions Permalink
  • Avatar
    chris
    maybe there should be an option to hide 2d images?

    maybe they grouped, switch between groups?

    say if i have a scene with 2000 photos, but I only want add tie points to say 12 of them. if i could hide the rest, then it would be possible to see all the control points + images that are needed in one space.

    i think also being about to "sort by" the images so you can quickly see which ones already have control points, which have been aligned properly, which have exif info, etc...
    0
    Comment actions Permalink
  • Avatar
    lopes
    I am in the same boat as Bill. Most of my work includes creating Orthophoto, DSM and 3D point cloud. Would love to see this PDF document on speedy GCP placement because this is, at least currently, a rather cumbersome workflow for large ortho projects (1000+ images). I also agree with chris in the fact that it would be much easier to place GCP in photos if they were filtered to only show the images where the GCP is located.
    This product is incredibly fast at other workflows and is a stellar package in total. I am very impressed and look forward to learning more.
    0
    Comment actions Permalink
  • Avatar
    Wishgranter
    Hi

    I have done a lot of testing on this stuff lately and we have slightly changed some workflows ( UI work ), so now I just need to prepare the PDF with a step-by-step tutorial.
    I need few days to get it done and make it ready to publish...
    0
    Comment actions Permalink
  • Avatar
    Bill Isenberger
    Looking forward to this. :)
    0
    Comment actions Permalink
  • Avatar
    peterwatson
    Any further progress on this?
    I am in the same boat as these other guys and trying to create large ortho photos and DTMs. I find the photoscan workflow for this pretty straightforward but am struggling to do same thing in RC, more than likely this is due to my inexperience with RC but the lack of a step by step guide doesnt help! I find myself just clicking various icons and buttons to see if i can get it to do what i want it to do.
    0
    Comment actions Permalink
  • Avatar
    Wishgranter
    Hi peterwatson

    Please be polite when posting things related to other products on the forum...
    The help file is almost finished and will be released soon. It covers almost every part of RC in detail.
    0
    Comment actions Permalink
  • Avatar
    peterwatson
    Yes understand, however this was last discussed in February. Not being impolite, just would appreciate a step by step tutorial as is available for other paid products.

    Looking forward to the updated help file!
    Regards
    Pete
    0
    Comment actions Permalink
  • Avatar
    peterwatson
    OK I think i have worked out the basic workflow for creating an orthophoto and DTM with ground control points. Couple of projects I have done I am very impressed with the speed, quality and accuracy of the results. However i still find things a bit clunky and I don't know if this is just because i don't know the most efficient way to do things or what.

    At the moment I am doing a construction site - about 2.5kms x 1 km wide... about 1000 aerial photos and have 20 ground control points. Is there any way of filtering photos for a particular GCP -once I have marked the GCP in 1 or 2 images?- there is a function in PS which does this- locate same GCP in 2 images then filter by this GCP and it finds all the images that this GCP appears in and places a GCP marker at calculated position in those images- then you can just go through those images and refine the marker so it is right over the correct position. Makes the process a whole lot quicker than scrolling through the image list one at a time and trying to find GCPs visually.

    Would love to have this functionality in RC if it is not in it already- this is a great product- speed is absolutely amazing in comparison to other software! And beautiful results- which I am finding generally more accurate than other software also, depending on flight altitudes etc.
    0
    Comment actions Permalink
  • Avatar
    peterwatson
    Any updates on whether this functionality will be implemented in the future. This is probably the biggest downside for me at present. Using drone footage to produce orthophotos over a reasonably large area (+ point cloud /3d model). The GCP identification and placement in images is by far the most time consuming part of the process.

    When using RC at present I am sorting thru each and EVERY image (left/right arrows) and identifying GCPs visually in each image- sometimes there may be multiple GCPs in images so I've got to go forwards and backwards adding them in- this works reasonably well as I have images re-named according to time taken so I can do a virtual fly thru of job and skip sections which don't have GCPs... but I would prefer the software did the sorting for me!

    This is one thing the competitors do better! Both Agisoft and Pix4D have filtering mechanisms to streamline GCP placement. I have only played quickly in pix4d demo but it does a pretty good job of filtering images and making GCP placement quick. Agisoft is great and I love the way it handles GCP placement-> align images- then identify GCP in 2 images and then you just sort by that GCP- then it only gives you the images that have that GCP in it... then after a few more are located it even focuses successive images in the right place- so only minor adjustments are required to locate GCPs right on your aerial targets.

    Is this something that you guys can work towards in future releases? Would make things a lot easier for us doing this sort of work. RC seems to blitz the others when it comes to processing speed, accuracy and detail (and beautiful models!) Please, please can we have a more streamlined way of aligning ground control points?

    pete
    0
    Comment actions Permalink
  • Avatar
    yoz
    +1 for this

    Workflow for adding GCP is very awesome time consuming in comparison with competitors Pix4D, Photoscan or Context Capture.

    I think it will be a priority in the devs plan.

    Yoz
    0
    Comment actions Permalink

Please sign in to leave a comment.