Tutorial - How to export 3D photogrammetry meshes from Google Maps

[@SanningArkitekter]

Update (220530): With Maps Models Importer version 0.5.0 (release candidate) being released a month ago, we were eager to try out an updated workflow with the new software versions (Maps Models Importer 0.5.0 / RenderDoc 1.19 / Blender 3.1). However, we personally had no luck in capturing the data with RenderDoc 1.19 (albeit succeeding to launch Chrome directly from RenderDoc), so we refrain from recommending the updated versions as of now, and still use the previous versions for the time being. However, it works perfectly fine to capture data with RenderDoc 1.13 (old version) and import the saved .rdc file with the new Maps Model Importer version 0.5.0 in Blender 3.1, so the versions we personally use are: Maps Models Importer 0.5.0 / RenderDoc 1.13 / Blender 3.1. We wish you all the best in your mesh extraction! /S&A

[@SanningArkitekter]

Update (220709): We just wanted to say that the RenderDoc 1.13 capturing method that nowadays works best for us is to click on "Capture After Delay" with a delay set to 5 seconds, ALT+TAB back into Google Chrome, and very gently pan (by clicking and dragging with LMB) the map. The capture will trigger when the delay timer is finished. This has more consistently yielded higher-coverage captures for us, after a frustrating period of numerous captures of only thin slices or erroneous fragments. You might not have had these problems recently, and we recommend that you use the method that works best for you :) /S&A

[@SanningArkitekter]

Update (220710): It seems as if this particular method, while increasing the likelihood of adequate coverage, it is prone to producing duplicate meshes, which both makes the model unnecessarily complex/heavy and can produce rendering errors. Luckily, thanks to the below script by batFINGER on GitHub, it is an easy feat to remove all duplicates. Simply select all imported objects in Blender, go to the Scripting workspace, and paste the below script into the Python Console. Press Enter twice to execute and delete the remaining selected objects (the duplicates). import bpy from mathutils import Vector context = bpy.context scene = context.scene view_layer = context.view_layer TOL = 1.0e-3 # reduced tolerance. # routine to text matches on bound box. def test(ob1, ob2): # test if ob1 and ob2 which have same face/edge/vert count are same bbox1 = [ob1.matrix_world @ Vector(b) for b in ob1.bound_box] bbox2 = [ob2.matrix_world @ Vector(b) for b in ob2.bound_box] diffs = [v1 - v2 for v1, v2 in zip(bbox1, bbox2)] return all(d.length < TOL for d in diffs) #return sum(diffs, Vector()).length < TOL obs = {} keys = [] for ob in context.selected_objects: ob.select_set(False) if ob.type != 'MESH': continue # make a key from faces, edges, verts mesh = ob.data key = "%d_%d_%d" % (len(mesh.polygons), len(mesh.edges), len(mesh.vertices)) obs[ob.name] = key keys.append(key) possible_dupes = set([x for x in keys if keys.count(x) > 1]) print("-" * 50) while possible_dupes: p = possible_dupes.pop() dupes = sorted(name for name, value in obs.items() if value == p) while dupes: original_ob = scene.objects.get(dupes.pop(0)) for dupe in dupes: dupe_ob = scene.objects.get(dupe) if dupe_ob.select_get(): continue if test(original_ob, dupe_ob): dupe_ob.select_set(True) view_layer.objects.active = dupe_ob print("possible dupe: ", dupe, " of original: ", original_ob.name)

[@DarwinsGreatestHits]

I had success with Model importer v0.5.0, RenderDoc v1.19 (enable process injection in settings), and Blender v3.2.2.

[@SanningArkitekter]

@@DarwinsGreatestHits That is great to hear! We'll be sure to try it out ourselves at a later date! Thank you for your report /S&A

[@IndigoKidda]

hey man what ram do you have to be able to bake them textures at 16k? and is there any way of getting the exact quality of the google doc files render onto a mesh?

[@SrValeriolete]

Finally a step by step tutorial with all the details and pitfalls and that don't rush through it

[@SanningArkitekter]

We are very glad that it helped you! And we thought about editing out the pitfalls, but found that it was more pedagogical to leave them in :) Thank you, and Merry Christmas! /S&A

[@euclois]

@@SanningArkitekter How do we scale the model on blender to real world dimensions? A 195,0m real world reference comes out as a ~2,45m model after exporting it to blender. I'm not sure if I should always scale the model with a fixed factor, or it's a case by case.

[@SanningArkitekter]

@@euclois For us the scale has not been consistent between projects, so what we do is simply: 1. Go to Google Maps, and locate the area you have exported, 2. Right-click on a distinct point in the map (which of course also is in your extracted mesh), and choose "Measure distance", 3. Left-click on another distinct point, and take note of the distance (A). 4. Go to the Blender file, and locate those two distinct points, 5. Use the Measure tool (icon looks like a right angle), and CTRL+drag (which will snap to vertices) from any vertex closest to the first reference point, to any vertex closest to the second reference point. Take note of the distance (B). 6. With a calculator, calculate A/B (make sure that units are consistent [e.g. both in metres], otherwise, first make sure to convert to same unit). This is the factor you want to scale the mesh in Blender with. We wish you all the best, Sofia & Alexander

[@j.overstreet3666]

If I could give you a 1000 thumbs up, I would. Thank you for this video. It has been immensely beneficial for my project.

[@SanningArkitekter]

Thank you very much, Mr. J! We are so glad that this has helped you in your project :) Best wishes /S&A

[@RoadwayVR]

Here is very very simple workflow to get same and better result: 1. Buy Lily Texture Package for 6$ 2. Run it on all objects (this will combine all textures) 3. then Ctrl+L > Link Materials (this will combine all materials) 4. Ctrl+J to combine all meshes, 5. M > By Distance to merge nearby vertices.

[@SanningArkitekter]

That is a really good piece of workflow suggestion you got there! That would surely make that part of the process significantly less cumbersome :) /S&A

[@wingmanmedia360propertybus4]

mate you are the only one who is using Render Doc and has given a good trick with chrome zoom. It still won't do high details for texture and objects for quite short buildings. However if you want to capture any CBD with a lot of high-rise buildings with your trick you can get a maximum resolution texture for them. If not a mess with badly defined vegetation like trees, some buildings not having hires texture and objects on the ground level are not well defined it actually looks quite real. I will try to do Eifel tower just to see if I can get the whole structure in 3D. Thanks a lot on this trick.. I would never guessed it could work and I almost gave up on this method of getting anything decent from Google Maps.

[@SanningArkitekter]

Great to hear that you also found good use with the Chrome zoom technique :) All the best to you /Sofia & Alexander

[@julesxjoules]

you have helped a lot i cant thank you enough for your in depth trial and error filled tutorial, i learned a lot

[@SanningArkitekter]

Thank you very much for your input! Initially, we thought that the large amount of trial-and-error seemed too unprofessional for a tutorial - but retrospectively, it's good that those parts are included as well, since obstacles that we encountered surely could be encountered by anybody :) Great to hear that you learned from it! All the best to you, Sofia & Alexander

[@orellugasi4721]

Nvm, I found the problem that you forgot to mention: For the newer version of RenderDoc, you have to go first to settings (Tools > Settings) and mark the button: "Enable process injection (restart required)", after you do that the "Inject into Process" option will appear at File.

[@SanningArkitekter]

It is great of you to share how you managed to get it working with a more recent version of RenderDoc! We hope that others who had the same issue will be helped by this as well! Best wishes /S&A

[@RaulTorresMorfin]

If you are working with imperial units, change them on Blender to 1ft/imperial, select your whole mesh CONTROL J to join everything and change the scale to .3 (XYZ) so your model will be the correct scale/size

[@SanningArkitekter]

Thank you for sharing your workflow for imperial unit scale conversion :) /S&A

[@aenysugianto1908]

Hi, I'm working with the 3D mesh scalling rn, would you mind to give me this tutorial? Thanks.

[@infotreetv1379]

Very nice information how to increase level of detail 👍

[@SanningArkitekter]

We are very glad that the information was appreciated :) /S&A

[@erikwest7299]

thank you for the thorough walkthrough of your process. I'm currently trying to extrapolate partial processes to use with the Blosm plugin, but having some difficulty. Would you consider producing an updated workflow tutorial?

[@SanningArkitekter]

Hello, Erik! Thank you for your kind words! We have not had much time to do further tutorials, but your suggestion is an excellent one! But like this comment is very late (two wonderful children influencing that, without them to blame) - don't hold your breath :) Best wishes to you /Sofia & Alexander

[@lipsach]

Thanks for detailed explenation.

[@SanningArkitekter]

Thank you very much for you kind words! We are glad that it helped you :) /S&A

[@orellugasi4721]

Hi, First thank you very much for the clear and detailed explanation! :) As of today: Last version of MapsModelImporter v0.6.2 compatiable with RenderDoc v1.25 When I open this version of RenderDoc, there is no option of "Inject" anymore at File, instead of it in the tabs, there is "Launch Application" that doesn't look similar.. Where can I find the starting process id in RenderDoc v1.25 that I've recieved from cmd? Thanks :)

[@SanningArkitekter]

Thank you - we are glad to hear that :) /S&A

[@Aenimae]

Hi, i'm stuck on the injection process. I removed from chrome all tabs, and extensions. But when i inject it tells "OpenGles (not presenting)" intead of active, even thou it has established connection. Cannot go on from here, cause every time i open a new tab on chrome it refreshes renderdoc with a new PID. Is it possible that i do not have the right version? i downloaded the latest one... Thanks!

[@SanningArkitekter]

Hello, Marco! Yes, we believe you are correct in assessing that the version of RenderDoc might be the culprit! As stated in the pinned comment by us, we have not had any luck using any other RenderDoc version than 1.13. So we would suggest you to try with that version (as is used for the tutorial). We have had success in using the 1.13 RenderDoc version together with the later version of Maps Model Importer (0.5.0), so the main thing is using the same RenderDoc version as in the tutorial (1.13). You can find it here: renderdoc.org/builds

[@Aenimae]

@@SanningArkitekter oh my bad, I did not see the pinned comment. Thank you very much for the answer i will try with the older versions, keep up the good work!

[@SanningArkitekter]

@@Aenimae Thank you! And good luck, Marco! /S&A

[@JayGee1]

Awsom and detailed tutorial 👍

[@SanningArkitekter]

Thank you very much for your kind words! We wish you the best /S&A

[@JoshPurple]

Brilliant! Thank You very, very much 🏆🏆🏆!

[@SanningArkitekter]

Thank you! We are really glad you like it! Sincerely, Sofia & Alexander

[@MrChristiaan13]

Does it also work with google Earth Pro? Google maps 3D is not vissible for me, the function is locked.

[@mckone9]

same

[@mckone9]

Google earth working btw

[@SanningArkitekter]

Thank you, @mckone9 for answering this question by verifying that it works :) /S&A

[@MrChristiaan13]

@@SanningArkitekter yes it also worked with google earth. Thank you

[@SanningArkitekter]

@@MrChristiaan13 That is delightful to hear! Best wishes /S&A

[@fahimehmirshafiei9347]

thanks for your tutorial, but I dont have "Inject in to process'' part , so can you help me or giving some advices to fix this problem? Thanks 🙏

[@SanningArkitekter]

Hello! Did you resolve this issue? What software versions are you using? We personally have had best success with the following: Maps Models Importer 0.5.0 / RenderDoc 1.13 / Blender 3.1 // Best wishes, Sofia & Alexander

[@jave567]

this is a 5 star tutorial thanks!!

[@SanningArkitekter]

We are very glad that you like it! /S&A

[@fernandoalonsova]

Increible. Muy muy bueno.

[@SanningArkitekter]

Muchas gracias! /S&A

[@MonsieurCr0w]

Hi, I followed the instructions but i'm stuck at the part where you import your .rdc in blender. My file is 150Mb and it only imports a random glitchy rock. I looked at every comment but no clue...

[@SanningArkitekter]

Hello, we have had this same issue when using another Blender version than recommended - have you made sure that the Blender and RenderDoc versions are the same as are stated to be compatible with the Maps Models Importer version that you are using? Best wishes /S&A

[@MonsieurCr0w]

@SanningArkitekter Hi and thanks you for replying! Yes i used the exact same versions that you use, without luck. Wondering if some GPU could be the issue here... Anyway I ended up using BLOSM, way faster and simpler. Next step is to learn how to make it compatible with a racing game 😅

[@SanningArkitekter]

​@@MonsieurCr0w Great to hear that you've got it working with BLOSM! And good luck with the map for the racing game - there are some really great examples/showcases of Assetto Corsa maps on KZbin :)

[@omarmohammed9571]

Thank you so much! !!

[@SanningArkitekter]

We are very glad you like it! Sincerely, Sofia & Alexander

[@WeAreConsciousExplorer]

Hey please please help me I am trying to do this from last 2 days 😭😭😭 I am able to get Chrome to launch and give me the GPU number and inject it, but AS SOON as I hit OK to close that little popup window, the injection window blanks out. My chrome doesn't have any debugging text on it either. But no one has said anything about what version of chrome to do this with, and I have the most up to date version. I'm not sure what I'm doing wrong or what I need to do to fix it. After clicking ok new pid appears continuesly 😭

[@WeAreConsciousExplorer]

Renderdoc v. 1.28 chrome latest version

[@SanningArkitekter]

Hello! Thank you for already answering what would have been our first question - what versions you are using. For the workflow to work, it is really important to use the older version RenderDoc 1.13 - this is the only version that we have had consistent success with. Please let us know if it works for you with the recommended version :) /S&A

[@WeAreConsciousExplorer]

@@SanningArkitekter Thanks brother ❤️❤️❤️❤️❤️❤️❤️

[@seanli7565]

Amazing video, thank you very much! However, my 'API:' Status remains "None" after following the steps and pressing 'Ok'.

[@SanningArkitekter]

Hello, Sean! Thank you for the encouraging words! If you would be so kind to explain a little more specificly the circumstances of your issue (at which step does it happen, preferably with a timestamp reference to the tutorial video?), then we'd be glad to try to help, or if we cannot help, we would at least try to replicate the issue on our side :) /S&A

[@seanli7565]

@@SanningArkitekter It happens from 5:25-5:27. My API remains "None" even when I click OK on the chrome tab.

[@seanli7565]

It appears to not work on my computer but works on my other computer.

[@SanningArkitekter]

@@seanli7565 Okay, we're glad to hear that you got it working, although it was on another computer. Then for identifying the issue on your computer, perhaps you could try to identify any significant differences between the two computers. Windows versions, graphics drivers could be the culprits, but we do not know what it could be :) /S&A

[@ahmadmohammadi2396]

Turn off your AntiVirus, then you would be fine

[@alexanderchiknaev356]

Thanks a lot…. But RenderDoc v1.19 doesn’t have “Inject into Process”. I used v1.13 and MapsModelsImporter-v0.4.0 for Blender 2.93 and MapsModelsImporter-v0.4.1 for Blender 3.0. And used the “Capture after delay” button 2sec in RenderDoc when I were moving in 3d view GMap. By the way THANK YOU VERY MUCH!!!

[@SanningArkitekter]

Great to hear that you got it to work! And thank you for the information about the versions which you got it to work with! All the best to you /Sofia & Alexander

[@wingmanmedia360propertybus4]

I am not familiar with RenderDoc apart from the part for doing it for getting 3D data from maps but I have fallen into a trap of updating RenderDOC to the latest version and lost that inject option under File too. I think now in a new version it can be done through application interface. So you can start Chrome with all parameters for injections directly from RenderDOC and it technically should inject into a process without looking and selecting for a proper process ID. I will try to figure it out tomorrow.

[@gabortabi]

@@wingmanmedia360propertybus4 that option can be enabled in the settings and then the menu item will also appear. Still, it does not work for me.

[@ozstockman]

@@gabortabi I deleted an updated version of RenderDoc and installed the one that used to work. It is back to normal and it works.

[@watnou346]

does not work does not work does not work it keeps closing the window inside Renderdoc after pressing "ok" on the Google chrome pid number window. im so angry now

[@zigaoberstar1039]

same here, it keeps refreshing chrome giving me new PIDs, only when i close render doc it opens the chrome tab

[@SanningArkitekter]

Hello, just to verify the issue, are you starting Chrome with all three arguments in the Command prompt? set RENDERDOC_HOOK_EGL=0 "C:\Program Files\Google\Chrome\Application\chrome.exe" --disable-gpu-sandbox --gpu-startup-dialog --disable_direct_composition=1

[@danaritter9012]

When I import the .rdc into Blender, it only imports one small mesh section of the entire scene..Any tips? Using Blender 3.4 and RenderDoc 1.25

[@SanningArkitekter]

Hello! Are you sure that the Blocks setting during import is set to -1 (infinite limit of imported blocks)? If it is set to 1, then it will only import 1 single block. We have also had this error when trying out the workflow with Blender 4.0, and have succeeded with the same input in Blender 3.6, but you are using 3.4, which is the recommended one, so software version should not be the issue. We hope that you find a solution to your problem! Best wishes /S&A

[@stop-motion-verse]

I have created an RDC file successfully, but when it comes time to import it into Blender. I hit import and nothing happens, no loading or anything. The file is 27,000 kb. The plugin is installed correctly it seems.

[@SanningArkitekter]

Hello! Are you using RenderDoc 1.13 and Maps Models Importer 0.4.0? There have been reports of the workflow working with Maps Models Importer 0.6.2 with RenderDoc 1.25 & Blender 3.4, but 1.13 and 0.4.0 is the only versions that we ourselves have verified to work :) /S&A

[@jamesrancho]

Thanks for the great detailed video. I find it hard to find the right parameters for different site size, Meshlab returns warning "Inter-Triangle border is too much" for 2 px, only 1px is accepted in my case; Bake takes 5 hours in my blender, my margin size is 1px. Then I tested with a smaller size terrain, no warnings but my image are always almost black.

[@SanningArkitekter]

Hmm, that the UV unwrap in Meshlab has its limit for large meshes with finite texture resolution and the necessity to fit the unwrap with regards to margin sizes is reasonable, but we wonder about the output being always almost black for smaller mesh sizes - did you resolve this issue? Best wishes /S&A

[@razvanalx4748]

it doesn't work for me, after I give it an injection in chrome and then press ok, that code is reset

[@SanningArkitekter]

Hello! An updated tutorial has been published, and is available via the following link: Tutorial - How to extract Google Maps meshes and import into Blender/Rhino/UE5 for reference (2024) kzbin.info/www/bejne/aqK8fmZ9rMuqZpI

[@haithamalkilany3785]

Hello I have a question, I want to use this in order to bake the buildings texture in Maya from High poly to low poly in order to take the textures is it illegal to use it this way

[@SanningArkitekter]

Hello! We recommend you to read through the Usage Guidelines for Google's map data on this link: about.google/brand-resource-center/products-and-services/geo-guidelines/ As well as the terms of service: www.google.com/help/terms_maps/ We will paste here a couple of excerpts from these, but we are not lawyers, and you are responsible for how you use these data. "You may not use output, or use third party tools to capture output, from Google Earth, Google Earth Pro, or Earth Studio to reconstruct 3D models or create similar content, or to create other content, products, or services that may violate our Terms of Service." "2.b [...] you may not [...] copy the content (unless you are otherwise permitted to do so by the Using Google Maps, Google Earth, and Street View permissions page or applicable intellectual property law, including "fair use")" Our interpretation - which is not legal advice - is that for usage on your own personal computer, with no motive of profit, or educational non-commercial use, the concept of fair use applies, which is an exception for this workflow ("capture output") stated in the terms of service. However, if you have any commercial motive, then it is clear that such usage violates the terms of service. We hope this clears up the question a little for you :) Best wishes /Sofia & Alexander

[@MonsterJuiced]

No matter what I just can't seem to get that renderdoc overlay on the chrome browser and the buttons remain greyed out in renderdoc so I cannot capture. I have no idea why

[@SanningArkitekter]

We just wanted to add here (replied to your other comment) that if you haven't already, try with the 1.13 version, that is if you have tried with 1.19 and 1.21 unsuccessfully. If this problem also occurs with 1.13, and you follow the steps in the tutorial, then we sadly do not know what the issue could be. Sincerely, Sofia & Alexander

[@togasimulations6456]

Hi, I'm very new to blender and I was wondering if it would be possible to edit the mesh once it's imported and be able to delete parts of the mesh and put my own custom textures over it before baking and joining everything together? Thanks

[@SanningArkitekter]

Yes, we think that this would be entirely possible; however, due to the fragmented nature of the photogrammetry meshes, where there are many small mesh objects comprising the scene, it might be a challenge to make a deliberate selection of the area to replace - and depending on how large the mesh is, it might be challenging for the computer to edit the mesh (at least it has been from our experience - in these cases we have found it fruitful to work in smaller chunks). But experiment away, and see what you might find! /S&A

[@michaelhughes3293]

yes, that is possible, its just an object like any other in Blender. I often adjust vectors after capture.

[@darkizofficial]

I don't have inject option in render doc

[@SanningArkitekter]

What software versions are you using? We personally have had best success with the following: Maps Models Importer 0.5.0 / RenderDoc 1.13 / Blender 3.1 // Best wishes, Sofia & Alexander

[@darkizofficial]

@@SanningArkitekter I'll try again, thanks!

[@matiaspose5106]

Each time my GPU starts to render the map Google Chrome starts a new API.. Does anyobody know how to solve this?

[@SanningArkitekter]

Hello, Matiás! Hmm… We have no success in replicating your issue. Just to verify, are you using RenderDoc 1.25, and making sure that no Google Chrome process is running before executing the two commands in Command Prompt? Best wishes /Sofia & Alexander

[@gunterstunter]

Is this higher resolution than the google api for the 3d tile set that they released this year?

[@SanningArkitekter]

We do not know if there are any differences in terms of resolutions between the datasets, but with this workflow, what you see is what you get - you will capture the level of detail that is showing on screen for Google Maps, so if you already have access to the 3D tile set, you can visually compare yourself with Google Maps, and see if there seems to be any difference :) Best wishes /S&A

[@AromaticGold]

Excuse me, maybe you can help me solve my problem. I installed the versions you mentioned(0.5.0 , 1.13renderdoc, 3.1 blender) but after inject the process is run and click ok for PID GPU for chrome GPU, the card to capture sequentional frame disappearing with tab and must do process again, but whit no any sucess.I even tried other versions but the result was the same

[@SanningArkitekter]

Hello! We are sorry for the late reply, but we wanted to test the workflow ourselves before answering, and now we have had the time to do so - unfortunately for your issue, everything works 100 % for us on our end, following the workflow as demonstrated in the tutorial, with the same software versions that you used, so we have no idea what could be the issue. Our intuitive suggestion would be for you to test on another computer, if possible, to see if the issue persists. Sincerely, Sofia & Alexander

[@AromaticGold]

@@SanningArkitekter yourwelcom, thank you for your reply I found my problem, the problem was that I had an extension on Chrome and that no other Chrome window should be open when you run the program.

[@SanningArkitekter]

@@AromaticGold Great to hear that you solved the issue! And the shortcoming that no other Chrome window cannot be open before starting the Chrome process to be injected by RenderDoc can be frustrating, especially when watching a tutorial for reference (in which case we recommend to use another browser in parallel) :) /S&A

[@geun4068]

i dont know whats matters Globe is not working, I've searched and tried many ways, but it didn't work. let's find another way. Can I pick Google Earth in 3d in the same way?

[@SanningArkitekter]

We have not tried ourselves, but the mesh data should be the same. It would be great if you tried and reported back whether it worked or not :) /S&A

[@geun4068]

@@SanningArkitekterits was work thx

[@SanningArkitekter]

@@geun4068 That is great to hear! Best wishes, Sofia & Alexander

[@asphaltpoet]

Hello! I’m running into an issue with the initial injection into chrome where it cannot find it’s path, does this method still work with the current version of chrome or should i rollback to a different version

[@asphaltpoet]

Resolved!

[@emanuelemastaglia]

@@asphaltpoet how you solved it?

[@SanningArkitekter]

Great to hear that you got it working! To answer Emanuele's question, our perhaps inaccurate guess would be that perhaps he was using a 32-bit version of Google Chrome, and per these instructions in the description, one needs to change the directory to Chrome accordingly. Best wishes /S&A -------------------------------------------------- Copy and paste these two lines into a command prompt and execute to start Chrome in injection-compatible mode - if you intend to use the 32-bit Chrome application instead of the 64-bit one, change 'Program Files\' to 'Program Files (x86)\': set RENDERDOC_HOOK_EGL=0 "C:\Program Files\Google\Chrome\Application\chrome.exe" --disable-gpu-sandbox --gpu-startup-dialog

[@devioxs5355]

Could you tell me if it is possible to remove buildings and cars from google maps

[@SanningArkitekter]

Hello, Kevin! To our knowledge, not by any non-manual means to do - one of the drawbacks of any photogrammetry workflow is that you might capture more than what you want within your bounding box of interest, simply because it was there in the footage. And since the photogrammetry mesh is constructed from the footage, if you want to remove a certain part, you must do so manually [in a photogrammetry software such as RealityCapture or Metashape, or in a 3D modelling tool such as Blender] (and even if you close the holes of any such topological manipulations, the texture is still bound to include the omitted items when you reproject, since what you have removed topologically still exists in the underlying footage, and to fully remove it you will have to manipulate the texture as well, which might be even more tedious than manipulating the topology). So the workflow showcased here is mainly for simple referencing. What we do recommend if you are not interested in the buildings (and cars) is to use another digital representation of a landscape, such as a DTM (Digital Terrain Model, excluding man-made objects) or a DEM (Digital Elevation Model, only including the surface of the Earth). These are generally consisting of a heightmap constructed from a fly-by laser reading of an area, or by other means, with non-natural (or in the case of DEM, non-topographical) elements filtered out and the ensuing blanks interpolated. The highest-resolution DTM/DEM data will most probably be authored by your national surveyor's agency (Lantmäteriet here in Sweden, supplying a DEM with a 1-metre resolution, as well as a 0.5-metre DSM [Digital Surface Model, with vegetation and buildings included]), but the licences and allowed usages will vary. In the public domain, NASA has captured and published a global DEM model with a resolution of about 30 metres per pixel, named "NASA Shuttle Radar Topography Mission Global 1 arc second", or "SRTMGL1". This will of course yield a significantly less high-resoluted digital model in comparison to the photogrammetry mesh, but if you do not want buildings, and cannot access your national surveyor's agency's data, the accessibility of the public domain might be convincing enough to try it. You will need both a USGS (United States Geological Survey) account and a NASA account (both free to create) to download the data, which you most conveniently do via the following address: earthexplorer.usgs.gov/ In the first tab, click in the map to define a polygon area roughly aligned to your area of interest. In the second tab, go down to "NASA LPDAAC Collections", and in the subcategory "NASA SRTM (STRM 3) Collections", check the option called "NASA SRTM3 STRMGL1". In the fourth tab, click the thumbnail icon under any entry to show an image preview of the height data. When you have identified the DEM part that you are interested in, click the download button to download it (make sure that you are logged in to both USGS and NASA to access the download link). With the DEM downloaded (a .hgt file - do not rename this since the title contains its only metadata [coordinates]), import it into QGIS, optionally reproject its coordinate reference system from the file's native WGS84 to the one you use in your project, and export the image data as 16-bit TIFF - this can now be used as a Heightfield in Rhino or a Displacement map in Blender, to create a topographical digital model. If you are interested in how to export the DEM as a TIFF (PNG works as well, but JPEG is only 8-bit and will lead to topographical banding) in QGIS, we have a more in-depth tutorial on this (the first 17 minutes of the linked tutorial): kzbin.info/www/bejne/jp-4oqugeduajck We wish you all the best, Sofia & Alexander

[@devioxs5355]

Thank you so much guys for the help I asked because I want to make a ww2 RAF game where you can fly around the whole or most of Britain and Germany and I didn’t want modern building or cars being in it If you could give any tips on how to do this I would be grateful otherwise have a great day

[@SanningArkitekter]

​@@devioxs5355 ​ With your usecase (a RAF pilot game), it might only be advantageous to use a lower-resolution DEM as a source - with a very large area, a too high resolution might be too much to handle for one's computer. If you are using Unreal Engine, you can use the Landscape feature to cover a large area with topographical data, which adjusts its topological density based on distance for performant rendering. And with the World Partition system in Unreal Engine 5 (released in sharp yesterday) you can create really large worlds. But that system in particular we have not yet tested ourselves :) But anyway you do it, our best tip from our experience, with regards to creating large worlds, is to initially approach it with one small part of it, so that you can experiment with your workflow and work iteratively with a small, handy chunk, and then when you have developed your own optimal workflow for the small part of it, you can apply the same workflow for the whole. This makes the initial threshold to overcome much smaller, and when you're working iteratively, testing out different methods, it also saves a lot of processing time. Then later, when you've found the optimum method to handle the issue, you can apply it to the whole. With the NASA STRM, the DEMs comes in chunks anyway, so even if your final project has the endeavour to cover hundreds of these DEMs, you can begin experimenting with your methods with only one :) Good luck with the project! One of us, Alexander, was a big fan of Combat Flight Simulator 3, especially the RAF campaign (it's mind-boggling think that it was 20 years ago now - time flies!) Sincerely, Sofia & Alexander

[@ahmedsufwan551]

Hey I followed it step by step, also injected it, but the 3d option does not appear in my google maps

[@euclois]

Btw, this isn't stealing data from Google, it's more of a friendly exchange. As if Google isn't already stealing loads of personal data from you. ;)

[@SanningArkitekter]

Now, this is a very dry response to a tongue-in-cheek comment, but what Google deems fair use of the data is described (with many examples of do's and don't's) in this link: about.google/brand-resource-center/products-and-services/geo-guidelines/

[@ahmedsufwan551]

Hey I followed it step by step, also injected it, but the 3d option does not appear in my google maps

[@SanningArkitekter]

If you go to the 1:01 timestamp in the tutorial, when we disable the annotative text, you see another checkbox right next to it, which should say "Globe view" (or the applicable translation in your locality). Make sure that this is checked, and see if that solves your issue. If you do not see a 3D view (photogrammetry buildings and vegetation, etc.) of the area, it might be that Google's associates have yet to capture it - try if the workflow works for Paris (as an example), and if you are able to enable 3D view for Paris, but not for your designated area, then that might be the case. Sincerely, Sofia & Alexander

[@chopito555]

Hello, when i click capture frames immediately nothing happens

[@SanningArkitekter]

We just wanted to say that the RenderDoc 1.13 capturing method that nowadays works best for us is to click on "Capture After Delay" with a delay set to 5 seconds, ALT+TAB back into Google Chrome, and very gently pan (by clicking and dragging with LMB) the map. The capture will trigger when the delay timer is finished. Best wishes /S&A

[@marcelobaus]

Can't even install RenderDoc, nothing happens when I click on the file.

[@SanningArkitekter]

Hello, Marcelo! We are aware that this is a late response, and we are sorry for this. While we sincerely hope that you solved your issue, we want to point out for future reference, that in the RenderDoc downloads page ( renderdoc.org/builds ), you may choose between 32-bit Windows, 64-bit Windows, and 64-bit Linux, and for all these three alternatives, you have the option of downloading an installer file, or a compressed archive. Windows 11 is the first Microsoft operative system to be exclusively 64-bit, but since the advent of 64-bit processors, there has been a 64-bit version for each Microsoft OS since Windows 7, so chances are that you have a Windows 64-bit OS (since this tutorial is not for Linux). Make sure to download the installer alternative, since that executable file will guide you through the installation. We wish you all the best, Sofia & Alexander

[@simracingenthusiast4769]

Hello, i followed everythign and it works well up until i import it into blender. The imported data does not cover the whole area of the shot that i took. Do you know the issue? Thanks. (Renderdoc 1.13, blender 3.2.2, maps-model-importer 0.5.0)

[@SanningArkitekter]

Hello there! The versions of the softwares that you use should be compatible, and since you followed everything, it should work, but perhaps the import setting for Max Blocks have been erroneously set by mistake? With [-1] it imports all segments, and with another integer value, it only imports that many segments - so perhaps it might be that? When you capture the data, do you click and drag with the mouse (we have personally had the problem that when we don't move the cursor while dragging, the capture is less than expected). We wish you good luck! /S&A

[@SanningArkitekter]

The timecode for the import setting in Blender is 15:00

[@simracingenthusiast4769]

@@SanningArkitekter hey, sorry for the very late reply. Yes i have -1 on the max blocks and when i capture i press capture, alt-tab to the chrome window and drag around for 3 seconds using the left mouse button. Thanks again

[@SanningArkitekter]

Have you tried the Capture with Delay option instead of the Capture Immediately option? /S&A

[@jenniferkalouche8060]

Hello! I had the same issue and i thought I'd share how I worked it out in case anyone faces the same issue in the future. So instead of clicking and dragging to capture the data, i actually click left + ctrl ( for some shift ) to actually rotate around the region i wanna capture, and that actually captured much more data than it initially did when i was only clicking and dragging.

[@user-vi7os2sc5u]

great video thank you very much I followed... It's stuck at the end. I keep getting the warning below... What should I do? Python: Traceback (most recent call last): File "C:\Users\user\AppData\Roaming\Blender Foundation\Blender\3.1\scripts\addons\MapsModelsImporter\google_maps.py", line 78, in captureToFiles out = subprocess.check_output([python, SCRIPT_PATH, filepath, prefix, str(max_blocks)], stderr=subprocess.STDOUT) File "F:\BLENDER\3.1\python\lib\subprocess.py", line 420, in check_output return run(*popenargs, stdout=PIPE, timeout=timeout, check=True, File "F:\BLENDER\3.1\python\lib\subprocess.py", line 524, in run raise CalledProcessError(retcode, process.args, subprocess.CalledProcessError: Command '['F:\\BLENDER\\3.1\\python\\bin\\python.EXE', 'C:\\Users\\user\\AppData\\Roaming\\Blender Foundation\\Blender\\3.1\\scripts\\addons\\MapsModelsImporter\\google_maps_rd.py', 'C:\\Users\\user\\Desktop\\새 폴더\\22.rdc', 'C:\\Users\\user\\Desktop\\새 폴더\\22-thgis8c\\22-', '-1']' returned non-zero exit status 21. During handling of the above exception, another exception occurred: Traceback (most recent call last): File "C:\Users\user\AppData\Roaming\Blender Foundation\Blender\3.1\scripts\addons\MapsModelsImporter\operators.py", line 54, in execute importCapture(context, self.filepath, self.max_blocks, pref) File "C:\Users\user\AppData\Roaming\Blender Foundation\Blender\3.1\scripts\addons\MapsModelsImporter\google_maps.py", line 306, in importCapture captureToFiles(context, filepath, prefix, max_blocks) File "C:\Users\user\AppData\Roaming\Blender Foundation\Blender\3.1\scripts\addons\MapsModelsImporter\google_maps.py", line 86, in captureToFiles print(err.output.decode()) UnicodeDecodeError: 'utf-8' codec can't decode byte 0xc1 in position 920: invalid start byte location: :-1

[@SanningArkitekter]

Hello there! We were worried by the sheer length of your error text that our ability to troubleshoot your problem would be non-existent - however, we believe that we have identified one definitive issue, and one general recommendation. First, if you go to the 2:13 timestamp in the video, you can see that each version of Maps Models Importer require a specific set of versions of the softwares it interacts with - in the case of Maps Models Importer 0.4.0 (the most recent as of April 15, 2022), it is compatible with Blender 2.93 and RenderDoc 1.13. From the error, we can see that you are using Blender 3.1 (which is an excellent choice for other usage), and you will need to use Blender 2.93 for this workflow to work, for compatibility reasons with the Maps Models Importer Blender plugin. You can download and install 2.93 in parallel (it will be installed in a separate file directory from 3.1). Then make sure to install the plugin to the 2.93 addons directory, and run the 2.93 executable instead. If you have used another version than 1.13 of RenderDoc, you might need to recapture the photogrammetry meshes - otherwise you should be able to use the same .rdc file for this new attempt. Secondly, and this is merely a recommedation, is to restrain oneself to only use, for filenames and directories which you use third-party plugins and "homebrew" applications with, only the latin alphabetical (and numerical) characters which are in the UTF-8 encoding standard, to make 100 % sure that any executables that are run are not having a problem simply because of non-latin interpretation errors. One would assume that this should not be a problem in our modern age, but for the two of us, having grown up in a computer environment where the Swedish "å ä ö" characters frequently proved troublesome, it's a perhaps outdated good practice that we still adhere to (so in this case, if you are still having trouble after using the correct versions, replace "새 폴더" with "New folder" or "1" or anything you would prefer :) Best wishes, Sofia & Alexander

[@Krylone]

Has anyone also tried using Meshroom instead of Maps Models Importer ? If yes what was your experience with it?

[@SanningArkitekter]

We personally have not done this - did you try that workflow out? Was it successful? Best wishes /S&A

[@mckone9]

1.13 0.5.0 3.3.8 working

[@SanningArkitekter]

This is great to hear :) Thank you! /S&A

[@andreravenna3488]

Anyone else having problems with the caputure only being a long thin strip no matter how you try and adjust?

[@SanningArkitekter]

Did you resolve this issue? Best wishes /S&A

[@b00lintime]

I found switching to Blender 3.3 fixed this issue

[@SanningArkitekter]

@@b00lintime That is great to hear! Hopefully it will help many others :) All the best to you /S&A

[@emanuelemastaglia]

Core 28236 16:01:19 win32_process.cpp(327) Error Couldn't create toolhelp dump of modules in process 16028 someone encouter this problem?

[@SanningArkitekter]

Sorry, we have not encountered this problem. Did you resolve this issue? What software versions are you using? We personally have had best success with the following: Maps Models Importer 0.5.0 / RenderDoc 1.13 / Blender 3.1 // Best wishes, Sofia & Alexander

[@mbayedieng1226]

est ce qu’on a le droit d’utiliser ces modéles dans un jeu videos

[@SanningArkitekter]

Non, seulement pour référence :) /S&A

[@Alejo8830]

Jefferson Sales done a few but not there yet

[@SanningArkitekter]

It is something that gets easier to do when you have tested around quite a few times :) /S&A

[@40S1_2]

is it possible to get the buildings inside aswell?

[@SanningArkitekter]

This would not be in this source data, since the photogrammetry data captured here seems to be from drone/aerial footage, which would only capture exterior surfaces. /S&A

[@gabbocxliv]

when ii try to import the rdc file in blender a error message pops up... help me please

[@SanningArkitekter]

Hello, Gabbo! We are aware that this is a late response, and we are sorry for this. While we sincerely hope that you solved your issue, if it persists, what does the error message say? Have you tried to use the exact same versions of the softwares that are used in this tutorial? Sincerely, Sofia & Alexander

[@gabbocxliv]

@@SanningArkitekteri solved my issue, but in case i get it again i will contact you! Thanks for your help

[@SanningArkitekter]

@@gabbocxliv Great to hear! /S&A

[@danielmamiasvili316]

Hello , my name is Daniel and im a young game developer , can you please tell me if its possible to make the 3d model look more detailed?

[@SanningArkitekter]

Hello, Daniel! With this workflow, what you see is what you get - although you cannot extract a mesh of higher resolution than what is supplied, you can make sure that you view the highest one available by manipulating the browser zoom level, as already stated around the 6:05 mark. Best of luck in your aspirations! /S&A

[@danielmamiasvili316]

@@SanningArkitekter Thank you for answering

[@Altitude962]

It shows failed to import module please help me

[@SanningArkitekter]

Hello, Nitin! Which software versions are you using for RenderDoc, Maps Models Importer, and Blender? Sincerely, Sofia & Alexander

[@kampfpanzer-703]

How do i get 3D view? my google map is showing 2D view

[@SanningArkitekter]

If you go to the 1:01 timestamp in the tutorial, when we disable the annotative text, you see another checkbox right next to it, which should say "Globe view" (or the applicable translation in your locality). Make sure that this is checked, and see if that solves your issue. Sincerely, Sofia & Alexander

[@SanningArkitekter]

If you don't have the layers pane open, click the "Layers" thumbnail in the lower left corner, and then "More". /S&A

[@SAYITAINTSODVD]

Have they made an workflow for MAC?

[@SanningArkitekter]

Hello! RenderDoc is not available on Mac as of now. The application itself is not Windows-exclusive, since there exists a Linux version, but the Linux-version lacks critical funtionality necessary for this workflow (inject into process of Google Chrome). So currently (220509) this workflow is only available on Windows. Sincerely, Sofia & Alexander

[@SAYITAINTSODVD]

@@SanningArkitekter thanks for the info. I’m going crazy over here lol

[@SAYITAINTSODVD]

@@SanningArkitekter if you run across anything we MAC users can use PLEASE POST. THANKS BUDDY!!!

[@smokeystover8959]

meshlab keeps crashing on me when i try this :(

[@SanningArkitekter]

We would suggest that you first test out with a very small portion of your intended mesh, to see if the crash is simply due to not having enough hardware resources available (namely RAM), or if it is a problem that happens no matter the size of the input mesh (which would mean that it's more likely a software error, in which case we would recommend reinstalling MeshLab instead). Sincerely, Sofia & Alexander

[@WeAreConsciousExplorer]

Please help me bro 🙏😭

[@SanningArkitekter]

We replied to your other comment :) /S&A

[@isaacvl95]

How much gb would this take

[@SanningArkitekter]

Hello! This exact model, as shown, is 282 MB as the RenderDoc capture file, and 427 MB as the exported OBJ from Blender, excluding the file size of 255 MB for the 16K image texture. So around 1 GB for these three files (but we have also kept the intermediary Blender file, close to 1 GB in itself). Best wishes /S&A

[@isaacvl95]

@@SanningArkitekter is there any possibilities on adding zones that aren't 3d, some zones on my city are not 3d on Google earth, plus google hasn't updated my city for like 8-9 years, could bing maps be an option? But I don't see it has a 3d option

[@SanningArkitekter]

@@isaacvl95 If you are primarily interested in 3D buildings, perhaps send a query to your city council with a question about their availability of 3D building data (the city of Stockholm, Sweden, offers free data in low resolution, while one can license higher-quality data). If you are content with terrain data, our suggestion would be to contact your national survey agency (in Sweden, it's Lantmäteriet) and ask about if they supply digital terrain models (DTM) or digital elevation models (DEM). If you want to try out a different workflow in Blender, then there is the plugin Blender GIS (download for free on GitHub) - Nicko16 has a great introduction via the following link. All the best to you /S&A kzbin.info/www/bejne/j3_XfKGjh72JotE&ab_channel=Nicko16

[@chooseadventure2311]

Ironic how google map data is copyrighted if i wanted to use for business. Yet i never gave permission to google to put my house on google earth.

[@SanningArkitekter]

Of course copyright is a more complex issue than can be briefly dissected, but one key aspect of copyright is "originality", and this is not (as we understand it) primarily about "how original/unique something is", but more of "to what degree is the creation uniquely influenced by its creator", and in this case, Google and their photogrammetry partners are surely (by ways of their programming and technological methods) influencing the end result (Google Earth) in a way which could not easily be replicated without copying the source code, so in that way, we think that Google Earth lives up to the requirement of originality, although competitors (e.g. Apple Maps) are doing a similar thing with similar methods (but achieving originality for their separate product, since they are uniquely influencing it themselves). But the question at hand, and the one you bring forward, is more akin to the dilemma when someone depicts something (let's say a photo of a house) and commercializes that depiction. The legal status of this is surely to vary across borders - in Sweden, it can be really simplified into this: anyone is free to depict anything (except buildings and places which are explicitly forbidden to depict, such as military complexes, etc.) from where they stand (and claim rights for that depiction), but the owner of a property (or the one with leasing rights) has the right to decide what someone visiting that property is allowed and forbidden to do. In effect, this allows for exterior photography, shot from the public street, with no need to ask for permission, while interior photography, or any photograph shot while visiting the property of others, is subject to the permission of that property owner/lessee. In other words, the exterior of the house is subject to the public, and the interior of the house is subject to the private, and this notion (if legally applicable in your country) might be why Google did not need permission to include your house on Google Earth. Best wishes /Sofia & Alexander

[@sgasolacewolf4930]

Dang it this copyright stuff🤦

[@SanningArkitekter]

Yes, the photogrammetry meshes are protected by copyright and property of their respective copyright holders. However, Google allows - in some cases - non-commercial use of Google Maps data, as long as proper attribution is given. If you are interested, please read this article by Google in their Brand Resource Center, where they give examples of copyright fair use and prohibited use: about.google/brand-resource-center/products-and-services/geo-guidelines/ Sincerely, Sofia & Alexander

[@chenmoney1920]

who really care copyright

[@SanningArkitekter]

We think it is important to adhere to, but if you are interested in the philosophical concerns about copyright, the Wikipedia page on it is a fairly light read, and showcases proponents' and opponents' views on it: en.wikipedia.org/wiki/Philosophy_of_copyright