3D Photo Scanning Tutorial | Blender 2.8 | Meshroom | Instant Meshes
HTML-код
- Опубликовано: 22 авг 2019
- *Notice Description Contains Affiliate Link
*I should note that this video assumes some prior knowledge of Blender.
In this video, I go step by step through the process I use to create 3D models/assets via photographs. This is known as photo scanning or photogrammetry. In this tutorial, I will go through the entire workflow of scanning an object, retopologizing the object, and baking the high poly detail to the retopologized model. All of this will be done with completely free software such as Meshroom, Instantmeshes and Blender 2.8. If you enjoyed this video please slap a like on it, and to see more videos from me, consider subscribing. Be sure to leave any questions or comments below!
Links:
Blender 2.8: www.blender.org/download/rele...
Meshroom: alicevision.org/#meshroom
InstantMeshes: github.com/wjakob/instant-meshes
Camera Sensor Database Info: github.com/alicevision/meshro...
A great tutorial about Baking from cages: • Baking Perfect Texture...
!!Afffiliate Link!!
Want to level up your learning while also helping the channel? Sign up to skillshare with my affiliate link below, and use code ANNUAL30AFF to get 30% off an annual membership!!
Link: skillshare.eqcm.net/B0VDaL
Small Note for Blender 2.9 users- everything in this tutorial worked for me EXCEPT the diffuse bake I needed to CTRL+ CLICK instead of SHIFT+CLICK the high and low poly objects in order to work!
You made my day
This is pure GOLD! Thank you! :)
This is the exact video I was looking for! I didn't know how to use all 3 programs in unison like that, and I didn't know about cages for texture baking, thanks a ton!
Perfect Tutorial, my search is finally over on how to do all of this. You just earned a Sub, please keep it up. Thanks a mil.
DUDE! Awesome video. I love that you move quickly enough that we get to the final product without having to sit through too much explanation. Some of it I knew how to do. Some of it I didn't. But that was the beauty of it. I'll figure it out. I just needed to know the steps. I hope you continue to make more videos like this!
You seriously rock! I especially liked that you are also a Blender user. This video really helped me learning how to use Meshroom. Great. Thanks.
Nice quality presentation. It says what you need to know without wasting time!
This was a really good tutorial. It’s exactly what I was looking for
This is a much better video than the one I saw on meshroom from one of the major blender channels. They just go through it to impress their audience but never explain what can go wrong, how to actually apply it to practical cases (they never mentioned you'll end up with a model that's so high-poly you'd have to retopo it anyway) and what the overall workflow should be. This video on the other hand, is actually useful.
This is a very useful tutorial, I hope I can use it in the future when I have a better pc and camera
Thank you for the complement! I wish you the best results when the time comes!
This is gold. Thanx my man
Subscribed, most artist will skip the retopo part but it's really crucial because it's basically free detail without affecting performance if you bake normal map, and this makes modelling game assets so much easier
just rad!) Thanx, man. This is the most informative video for now!
Amazing Tutorial..!!! Really Appreciate All The Useful Information You Provided... Thank You :)
Cool tutorial, thanks man! Much appreciated!
Great job. Thx, Tyler.
When baking to diffuse in this case you don't have to check alpha - it increases file size
Great job! Thank you! 🙂
Great tutorial, thanks Tyler.
Wow dude, nice detailed process!
so glad i found your video :3 fighting around whit my photo scanned objekts for months.. watched tuutorial after tutorial and never came to the result i wanted to have.. well until now :)
Thanks for this video! Sparked my creativity for a project.
Thanks so much for this video
I can't tell you how helpful it was 👍
Awesome! This is certainly useful 🙂
Great tutor , Thanks and keep going
great video mate
Thank you for this video!
Great video! Thank you!! I'll get a try!!
great tutorial I have tried a few times the photogrammetry game but my results weren't that great thanks I cant wait to try this out again
Excellent tutorial. Subscribed.
Great thanks Man!
NIce Work! I just saw a similar workflow for Cinema4d but I love this...it's FREE!
Thank you, very practical
great tutorial
Very nice tutorial. Thx guy for this, it was very interesting. :-)
for normal maps 32 bit float will give you the best normal bake, as far as i know, but i haven't done this in 2.8, only 2.79b and older. nice videos, would love to see more!
thank you for the very nice tutorial
Dude, this video is literally perfect. Shows anyone how to create 3d models and it's exactly what I needed. Have to take product pictures and make into 3d models to insert into environments.
Maybe another video on this with other stuff would be great or troubleshooting, etc... Thanks!
This is a good video.
I have seen a lot of tutorials on this but I have to say that yours is the best one so far.
Did you change any setting in meshroom other then the camera info?
good job. thanks a lot for the info
Great Video! Realy well explained...
For some reason, I have to select the lo poli first, then the hi poli, or it wont bake....
Thank you very much!
Really awesome stuff !! Best video on meshroom and instant meshes.
What do people do to make their roughness maps ?? I've seen people kind of reuse their diffuse map with a color ramp or something ??
nice!
Great and very well done video tutorial!
As a complete re-starter with 2.83, I found meshroom can use 2 or 4 reduction on the node (densemesh?) - then as far as I can tell, if that poly is sufficient then resulting .obj is similar to your more advanced method, ie low poly with lots of original photos detail? plus the job is quick on older computer and one step.
Super helpful tutorial man! If I were you I'd try to add more keywords to the video title and info like Meshroom, Blender 2.8, Normal map baking, etc so you can get relevant viewers! Great workflow and far more concise than other tutorials out there, keep it up!
Thank you so much!
Hi Tyler. Thank you for the great tut. Can I do this from images saved from an unknown camera source, or a hires image pc screenshot?
Great video !! Thank you, can you try something similar ? For example, do all the same for the statue but also photo scanning a cloth of yours. And the combine the clothe in the statue
Moving an object like shown and not having variance in the background will diminish the results. Those extra reference points help.
Well I'm definitely not saying you're wrong, I was initially under the impression that you cant move the object whatsoever. However, I find that if I take enough photos it usually works fine for me. Results are going to vary from project to project though, and this method of doing it definitely seems to require more photos.
So what is suggested? Lazy Susan?
14:52
Polly: Hi Tyler!
PARA LOS QUE NO SON ANGLOPARLANTES EN EL MINUTO 19:06 DICE ALT+S !!!!!! me costó un montón entenderlo. Mil gracias por este tutorial. thank you very much😍😍😍
I should have turned screencast keys on, my apologies
@@TylerSerino you do not have to apologize. The tutorial is perfect and has been of great help to me. Thank you.
really awesome video, one question: why use instant Mesh instead of using a modifier in blender to simplify the mesh?
In my opinion, It's much faster and cleaner than using the decimate modifier. Also, if I want to keep the high poly, and create a low poly to bake too, that would involve duplicating the highpoly in blender and then decimating that mesh, which is going to be heavy. If I were dealing with something like a character, I would manually retopologise it, but overall I'm not a huge fan of the decimate modifier.
New to 3d stuffs and I saw people selling pre photo scanning stuffs that is on the market for a lot and I just wanted to see if recreating a photo scanning thing is difficult or not and it's not. I like keeping things original for my self.
Thank you..... :-)
Great tutorial Tyler! Still I have one problem - same as several people mentioned in the comments bellow. After baking the first map I noticed that the result is just blank white shapes as I prepared them with the seams - so the shape is ok, I just missing the colour information (original texture). I think it's because I used another photogrammetry software (reality capture) on my second computer and then I just copied my .obj file to my first computer where I have instant mesh and blender. Could be this blank texture result of missing .mtl file in my second computer? Is somehow .obj file connected to the .mtl file without any further importing into the Blender?
I love how you skip some details that are easy to find anywhere, like installation process for instance, and keep the focus on what we need to know, I failed some times with others videos about it because of the lack of information and for the first time I could really understand what I did wrong and how the process really works, but Unfortunately you skiped too much in Blender and make it really difficult to understand if you're new in Blender.
The only detail that I learned in others tutorials and I liked is the way of how you get the photos, you could record a video, import in Blender, export as animation to get the images and use it, in this way you can manager better the amount of photos and save a lot of time.
Thanks so much for that video, even with some parts that's confusing, it's still far away better than any other for this propose.
The downside of using a video and converting it to an images is that there might be some motion blur and could lose texture quality. That's why taking a bunch of still photos is the more recommended option.
@@aldenlopez24 That's can happen with photos too, that's why you always need to review the photos. Personally, I think video is better because you save more than 90% of the time recording and exporting, than taking the pictures and if happens to have any kind of issue like motion blur, with videos you have a lot of frames to extract another photo, but with pictures you need to go back to the place and make more, but of course, that's for me, someone with less experience in video may have so many motion blur that will need to redo the video.
Guys! That's super cool, but when I export the mesh, it loose all the UV maps, so it's useless for me... any suggestions?
Amazing vid but i have one question why did you not turn the tri's into quads as that would have made the model look a ton better ?
But still a good vid ..😉
Supper Tutorial-- Tyler Serino you can really tach. Many Thanks. only thing some times I get models upto Chest -- how can I shift Origin
Thank you! You can shift origin within blender, by selecting the object in object mode, then going to the "object" dropdown in the top left and choosing "set origin". There are few different options for how you want to set the origin, all depends on where you want to set it.
It looks very easy the way you instruct it,
I want to use many cameras at once , Is there a possible method to identify my camera scene and position ( Map ) to meshroom to decrease the processing time or having better quality?
Are 13 MP cameras enough ?
Yes, but the more the better.
PLEASE HELP! I have a MacBook Pro (Catalina) and my graphics card is an Intel Iris Pro. I need to use CUDA in order to use the program Meshroom, but I found out that CUDA works only on Nvidia GPU, which I don’t have. I cannot install Meshroom without Nvidia GPU. I have no idea what to do and Google isn’t helping at all. PLEASE HELP THANK YOU!
Thank you so much! Would you suggest the same workflow to convert 3D scans of large areas to simplified meshes to be used in Unity?
Yes so long as it's not something you are animating (ie like a character) where you need more control over the geometry than what instantmeshes gives. Also for a game i'd try to retopo as low as possible without losing too much detail.
I have an issue when the processs making. In the PrepareDenseScene part it stops.The line goes red and when it click star again,it stops again.
Can someone explain me whether you need to lower down the quality of the result from Meshroom if you are going to convert the object into NFT? If you are not using it for any gaming project or CNC project.
Why not use blender's decimate modifier to lower the poly count?
Great Tutorial, but I got stuck on the Diffuse Baking. It just doesn't texture the Lowpoly Model for me, no matter what I try. Has something changed between 2.8 and 3.2 that could cause this Problem?
Nvm I figured it out... Works great! :D
How can we assign the models' measurement ?
thank u so muchfor this famo. kleen
Hello, what camera do you recommend me to start with?
INSTANT SUB
Hi Tyler.
Please, how to install Meshroom? I can't find any tutorial on youtube to help me do it. Do you know any tutorial? Can you help me please?? I'm desperate. Thank you very much and greetings.
How much iso, aperture and shooting speed? please
thanks Tyler.. Im getting an error message during bake that my hi rez model has no material WTF??
That's a wonderful tutorial, thank you! I have a question - on the scan I'm trying to achieve I keep getting the same issue - one particular element is missing in the final result. The object is a small figure holding a lance. Everything turns out great beside the lance's handle (rod, stick or whatever you call it). The handle is always missing in the end and the tip of the lance and other similar elements appear to be floating in thin air. Do you have an idea on what settings I need to play with to make it appear?
First off thank you for the comment! I'm glad you enjoy the tutorial. I'm not so sure that it's a setting, rather than a lack of necessary photos. If the lance is small and thin, this could potentially be difficult for the software to solve. It's hard to say without seeing the actual figure you're scanning, but I personally would try taking a plethora of pictures of the areas in question from as many angles as possible and add it in with the images you have that are working. Only do this though if you haven't changed the lighting/setting of the model or the two different sets of images may confuse the software. But yea, what I would honestly do is just try taking a boatload of images of the lance from as many angles as possible. Moreover, if this isn't working, you may need to start over with different lighting that helps to contrast the lance. It may be that the lance blends in with the model itself and so the depth map filter is filtering it out. Again, hard to say without seeing the actual figure, but this is what comes to mind based on what you've described.
@@TylerSerino I think it's because the handle is dark brown and it is indeed somewhat thin compared to other elements. In the 3d viewer of Meshroom it does appear that there are a very few points on those areas so maybe the software thinks they are just not enough and decides to skip them. I assume I need to retain the general composition of the scene and close-ups of that handle won't work? For reference, this is what I'm trying to scan - i.imgur.com/obhoKmY.jpg - Ornstein from Dark souls.
@@DedoPorno You can do close ups of the handle, and in fact that's what I would do. Just make sure you also have images that capture the object as a whole. I don't know the specific maths that meshroom uses to calculate everything, but it seems as though it relates features across photos. So if you have a bunch of photos showing the entire model, and then do close ups of the weapon, say with the hand in view, or part of the head etc, it will usually be able to make out what those images are in relation to the object as a whole. So take your photos of the object as a whole first and then slowly start taking images focused on just the handle from a bunch of different angles.
@@TylerSerino I see. Sounds like a plan. I'll try tomorrow. Thank you for the advice, Tyler!
@@DedoPorno No problem! Let me know how it goes!
Hi! If I don't put any number on the clampsize, when I'm importing the file, the model does not appear. Why does this happen? It also looks really low quality :(
Can these objects be used as kinetic meshes?
my meshroom got stuck in dept map. what should I do
❤
Очень полезное видео, спасибо!
Хотя есть некоторые вопросы..
I did in that way and meshroom is not able to place cameras in correct possition.
Thanks for this dopest video! I have some Question.. Can i get a low polly of photoscan of buidings and room ! The problem is can't optimize buildings it have 3500000 poligons,and its unreal to optimize in Unreal Engine 4 in game development. Howe can i get optimize? Ore i can make low polly in Mash Room? Thanks
Well, it's going to depend. I haven't tried on something like a building. The best way to find out would be to try it in instantmeshes. However, because of the fact that the overall shape of most buildings and rooms is less organic (ie sharp corners, cube shapes etc), the topology may not be to your liking. In that case what I would do, is start with a retopologised mesh from instant meshes and attempt to clean up geometry from there manually. Another option would be to use a tool like retopoflow (which is paid). It will take some trial and error in the end.
Do you think it could work to model characters? Can I make my characters from clay in a T-Pose and scan them? Or the topology will be too hard to rig for animation? I can sculpt much better physically than digitally.
I would say it's possible, but you'd have to manually retopologise, whether completely by hand or with a tool like retopoflow/others. I'm not a character artist but I know it's super important that geomtry follow certain rules if you intend to animate. The only issue is that it may be a painful process trying to retopologise a photoscanned mesh by hand, as I could see it causing blender to lag pretty hard.
I clicked start, and when it was done I couldnt load my model, its asking me to start again
@Aron Jabari happy for you
Thanks for the video. So what your telling with this video, is that you can't use the meshroom result directly into a slicer ? I'm new to 3d printing by the way.
Im not entirely sure how well the model would work for 3D printing. I am also not an expert at 3D printing, although I have done some. It's hard to say. When 3D printing if you're going for detail, you'd rather that detail be in the actual model rather than faked with normal textures. However, like I said its a huge file. The only way to know for sure is to drop it into a slicer and see if it crashes or not, and if not how long the eta is for the print.
@@TylerSerino Ok, thanks for the reply.
The process is completely different. You export the meshroom output into Blender, manually build it into a solid (fill all holes like in the base of the statue in this particular video), ignoring all the texture stuff. Then I generally use the remesh modifier. Set this to the depth of detail you need and press 'apply modifier'. Then export as Obj or Stl and open in your slicer of choice.
I use 2.81 and somehow when baking to the cage, i need to select first the Low Poly and then the High Poly.
Interesting, someone else said that too. I guess its either a bug or they changed it in 2.81
Yep, you're right ! Same here, I selected the Low Poly and then the High Poly and I worked
I have the same issue. I have to change their order how am I clicking on them or I will lose the image texture node...
How much time do you spend to do that mesh?
Can you give me some information about translucent, thin and transparent object can't be used as an object? Like leaf for example. Thanks
So, back when I first uploaded this video, I had no success with objects of that description and had not seen any others. I will say that I have seen people come up with creative ways to photoscan leaves since then. If you check out the photogrammetry subreddit you'll see some. However, they get really creative with how they get their photos. So I wouldn't say that these things are 100% impossible, but shooting them in the simple method that I did in this video is not likely to yield good results. You'll have to get creative.
@@TylerSerino Thanks mate🙏
I've did everything in this video 5 times already. I still get "circular dependency" error and I cannot go past this for hours . I added UV Map to the image texture node of the low poly and selected every image texture node of every of the 7 texture atlases I have in the high poly. I am a little bit desperate at this point. If you know something that I can try to get to baking I will be thankful to hear. Thanks in advance.
The circular dependency issue from my recolection happens if the object you're trying to bake from, contains the image node that you're baking to. It has been a while since I've seen this issue, and I am currently not at a computer with blender unfortunately, but to the best of my memory, it means that your high poly mesh has the image node that you're trying to bake to, instead of the low poly mesh.
unfortunately Meshroom does not work on MacOS :( ... do have an alternative ? thks
For some reason the texture baking never works for me, it just comes out as blank, anyone else have this issue and know a fix?
I have the same issue right now and I'm trying to resolve it. I think it's because I made photogrammetry reconstruction on my second computer and imported just .obj file without material file .mtl to the Instant Mesh on my another PC where I keep working with Blender to finish this tutorial. So you have model in Blender without any material info. We are not importing .mtl to the Blender but they are connected...
For some reason, I couldn't get past DepthMap. Any suggestions?
Hmmm, what happens when you get to the depth map? Does it error out and stop or does it just hang there? Also what does the log say?
My UV editing tab just comes up as black after I hit bake, anyone gt any solution for this?
Same issue, do you have your material file in the same computer where you are working in Blender? I think .obj file and .mtl are somehow connected.
This doesn't work for me. I've tried multiple times with different photos. It will only produce the front of my figure. How to you make it work with your photos?
It's always going to vary from project to project, but what are you shooting photos of, about how many photos are you shooting, what is the setting you're shooting in? Out of the total number of photos you're shooting, how many are meshroom using?
Rad. Unfortunately no Meshroom release for osx
I didn't realize honestly. I'm sorry to learn this :[
Mesh works only works with nvidia chips and nothing else.
Why not use decimate in mesh room to make it lower poly, saves using another program
I know a lot of people do that, and I'm sure you can get away with it, but I never liked how the topology looks when using that method.
Hello nVidia boys!
LAZY SUSAN - i have a flat disk that spins
You should not use shortcuts in a tutorial without at least putting them up on screen. otherwise great video.