it definitely wont work with a budget computer like you featured XD i had a hard time with my i7 4790k 16gig of ram and a gtx 1060 turbo (6gigs) i'll give a try with the non cuda version of Colmap ... maybe the 1060 is just not enough
We have released a sequel - Photogrammetry 2 that shows even better results and is much easier to do! :) (But needs an Nvidia GPU to run, unlike this tutorial) ua-cam.com/video/1D0EhSi-vvc/v-deo.html
10 mins ago I thought to myself - I wonder if you can use 3D scanning to help create Virtual Reality worlds quickly yet affordably. Not only do now know that the answer is yes but that I can come back here to follow step by step. Awesome video. Excellent production values and delivery of guidance. I'm off to the local antiques market to buy cool stuff to snap from every conceivable angle. Thank you for the time and effort you put into this. Legend!
it's what i have been already using for YEARS!!!! ua-cam.com/video/y6KRve6NOAs/v-deo.html there are so many iOS apps out there since 2012 already... how this complicated method is better? I don't know!
Thanx, man. After watchin this video. I went out into the yard with the phone (the weather is cloudy and the light is nicely diffused) and I took a hundred photos for the test. Very inspiring.
You are awesome. So now I understand how this rotating round stand for small models works. Its just a motor, sheet of round material with some non repeating pattern on it. So you put your camera/phone on the stand and do fast shots while the round stand rotates.
I recommend Meshroom. I had a lot of problems with the software used in this video. Meshroom lets you modify your result within the programm and you can work with nodes too. It combines all into one and works way faster and better than this outdated software.
You can rotate the object as long as background is in solid color. There are several solutions on internet, mostly by programming Arduino to rotate turntable with the object, then sending remote shutter signal to the camera. This whole process can be automated with very good results. Also if you don't exactly know if you did enough pictures, you can simply record video instead with you walking around the object then extracting video as separate .jpgs.
That is true, BUT a solid background makes it harder for the program to figure out the position of the camera, so in some cases the background helps in the process Often it is recommended to put the scanned object on a newspaper or similar to get more reference points in the images
Not sure if Meshlab has it, but with you can further remove noise from the point cloud using a Statistical Outlier Filter (CloudCompare has it). It will remove most of the stranded points that are close to the surface. This will make the initial computed mesh cleaner and easier to work with and smoothing results will look better.
Josef. Once again a great video, short, to the point and plenty of information. But please, either lower the volume of the background music or eliminate it. It's annoying and is not needed, IMO. Thanks for all of your hard work. I have purchased four of your printers and have been responsible for at least a dozen of other sales to my friends and people I have demonstrated them to in workshops I present, and articles I have written.
The common reason you find background music that is a bit too loud in relation to the speech is because people don't have professional facilities and tons of editing time to record their voice overs, so the music hides the annoying background noise - the microphone and preamp hiss, the computer rumble and fan noise, the furniture noises, breathing and mouth sounds, the gargling sound of noise suppression plug-in...
Wonder if taking a burst (a burst of pictures at a rapid rate) would work too, if not better but would probably take a lot more processing power to create the 3d model, since bursts take a lot more pictures at a faster rate then we do manually. My phone can take bursts by holding down the shutter button, it might be especially useful for something like this actually because a lot of phones have similar features and some have things like optical image stabilization which is not only really helpful when taking picture with shaky hands, but for taking still, unwarped, accurate, images for a 3d model as demonstrated in the video.
Love it! I could see building a dedicated featureless turntable except with landmark "X"s and a featureless, blocking, concave background with a stationary camera mount. Note doubt a low-tech Lazy Susan and poster sized card stock would do.
No, this is photogrammetry since it’s still camera shots meshed together thru data points to generate a model... the structures he shot were never in any motion
Cheesecake This seems like a cool idea. Easy to execute in other software too if you can find a way to slice a video into frames (ffmpeg can do it I think). Question is, would the resulting scans be feasible to process on an average desktop PC... I imagine they would have a LOT of points.
Videos are more compressed then video, and also have rolling shutter and motion blur. You can render our individual frames with quick time or vdub easily.
3DF Zephyr is probably the most feature filled and user friendly photogrammetry software out there. Also completely free for a 50 photos and less if anyone wants to learn. I train all my students on Zephyr, definitely my recommendation!
not only is video compressed intra-frame, but it is usually less than half of the resolution of the sensor ... ig you have a phone, you can take really crisp photos, but a FHD video is still just above 1 megapixel while a frame from a 4k video is just under 5 MP ... let that sink in
VisualSFM is also powerful and free, it uses additional CSM tools you need to install for as dense recontruction; it works great with different imager sources.
Hey Jo, great concise tutorial. You covered all of the required steps in a short amount of time. Definitely going to try this given I saw a $5.5K handheld scanner today. Cheers, JAYTEE
at 3:31 its shows the screened poisson surface reconstruction but mine ends at the voronoi filtering, is it in a different place? was it removed? I'm using the newest version.
Thanks for sharing. You are probably aware of this, but a drone would be the best tool for capturing 3D images - it is more precise, programmable and has more spatial freedom than humans
Interesting!!! But I would like something faster, I thought that shooting on the phone or SLR would serve to simplify the 3D process ... . I would take the photos and then just with a program do everything. I've tried Meshroom and other programs but to no avail. If I attach a link with an archive of some photos of a 1:18 model you can give me two tips to get a good result? Thank you very much.
Hmmzz... software setup needs some streamlining, but I think the best way to make it better would be to first get the rough model and then continue refining it with more pictures, matching the new pictures to already existing model, for that the workflow would have to be more automated tho.
Excellent video, one thing I don't think I understand from the video though: Will the point cloud software pick up large planes that are a single diffuse colour with no variation of texture? Like the side of a green bin or something.
Thanks for a great introductory video. -- basic enough to be immediately understandable by noobs, but in-depth enough for anyone who wants to dive straight in to get fully started!
This is really awesome! Very cool that it's possible to make a 3D scan with just a phone! I have a question though: If we aren't supposed to move the object, how can we exactly get a bottom angle of the object? Because Groot in your example is still sitting on the white surface.
in 3d graphics modeling is only 30% time And even with scan u can give a lot problems! For me faster create by hand this model than dancing with tambourines and recive some models which not good editable
Try Meshroom. It's free and outputs a pre-uv obj and texture image file so you won't need to use two separate software. You can import it into blender and it's good to go.
Very interesting. I tried to do exactly the same, but I still have errors with the filter applied in Meshlab... and I had to export manually the ply file from COLMAP. Maybe that's where the problem is. I'll try to fix it...
Yeah, I used to do that in Autodesk 123D in 2012. The models turned out a lot better. Nowadays it's no longer available. It used to be free. Made a lot of scans of my fridge magnets back then. I can print them now. 🤪
Hi Josef, I was hoping you could make a video on octoprint. I have a mk2s and I plan to get a mk3 and i'd like to add octoprint to both but I have no idea how to set it up with either or how to use it. I know a little bit about electronics but the whole idea of it is a little bit intimidating. Thanks!
You say take photos of bottom of the model but when you move the model to make it upside down you move it but you also say never move the image? So what to do?
You'll have to get creative and literally get under the model. Placing it on something tall and thin helps. Or in the case of the statue shown in the video, just trying to get shots from a very low angle to get at least some data from the bottom helps :) -Mikolas
Very interesting technique. I was wondering if the background is appearing in the images taken, would it not be better to block out the background by setting up chroma colour prop. The image would perhaps be sharper. Just a thought. I suppose a steady hand is a good thing too. Thanks for posting.
Well, it is still easier to 3D sculpt from the scratch than to scan using photogrammetry in some cases. As some 3D sculpting will be needed when cleaning and patching anyway.
That depends if you have 3D sculpting skills. You can't fault photogrammetry for it's sheer accuracy in capturing objects, even if the topology isn't the greatest. There's a difference between knowing how to use zBrush and being able to replicate an object exactly without letting artistic influence interfere. I think a pipeline of photogrammetry data with alterations from a sculptor is likely to be the way forward.
i see it as with motion capture. you can get insane level of detail but cleanup can be a bit of a nightmare and always needed. as with all 3d sculpting it depends on what the end product is going to be.
Yes, but it is amazing what you can do with photogrammetry in the case of environmental preservation. Photogrammetry used as a means to give a virtual tour of historical places would be less impressive, had they only been sculpted. That's not to say that skilled sculptors wouldn't be able to do something similar, but it's not quite the same, is it?
It does, though current phone cameras are actually quite good and very good at exposing the image well, which is also important. :) So with a decent phone camera, we achieved very nice results. -Mikolas
Cool , i was wondering if one can uses just a video clip because all the software would have to do is beable to take the .avi or mpeg and split it into a frame by frame images that it then build the .stl file from ? It be easy to just video it rather then take tons of pictures so curious if you can uses videos / camcorders instead. But this picture method is really really cool.
Well, what do you do with a hollow object? Let's say a basketball that has been sliced in half (and we just want to scan one of the halves), and we need it to be detailed on both the outside and the inside? I've been trying to figure this out for months...
This is so dope. I wonder what happens if you just record a video, break it down to frames and import them into the program? Will it improve the quality? Gotta play with this thing...
I've tried that and it's for sure an option! :) BUT, you have to keep some things in mind, shutter speed needs to pretty high, otherwise the individual frames will be blurred and even if you do so, you'll still have to go through the frames and delete the ocasional blurred one.
Nice information thanks I will give a try to this software . And I have to say 3D Creator from Sony xperia is really easy and awesome and avoid alot of step compared to this traditional method.
Hola, disculpame la molestia, estuve mirando un tutorial tuyo sobre el scanner para 3d, mi pregunta es la siguiente, me ofrecieron el sense v1, queria saber si es buen equipo a tu criterio. gracias
wouldn't it be faster to just film going around it with an infinity focus lens (so the background is not blurry) and have some program make images from every frame.
hi can you import somehow this code in c# to do it work with intel realsense d415? Hope you have some idea or code example THX - and cool video by the way!
hello, I have a question: should the photos shoot horizontally or can they also shoot vertically? Ps: excuse my bad English, it's not my native language, I'm using google translator....... Grazie (Thank's)
Running a brand new PC with Windows 10 (fully updated). Hit a brick wall within 2 minutes of this demo, Once the Colmap software is downloaded, the initial scan to test the computer fails multiple times, saying 'system Error, code execution cannot proceed because VCOMP120.DLL was not found', this happens multiple times... 3hrs later, many searches, Microsoft advise not to download the ,DLL file on its own, but to reinstall the entire PC! I later found a forum post saying "If a package requires any DLL that is not part of the operating system the developer should add the DLL to the setup program, failing to do so is the developers fault, a bug that nobody should ignore before publishing", not sure if this is true as I'm not a computer Developer, but I decided NOT to reinstall my 1week old install of Windows 10… The path later lead to various old posts telling me to download old software to fix the issue (Visual C++ Redistributable Packages for Visual Studio 2013), I also looked into the updated Visual Studio 2017 software, which lead to an install of hundreds of various tick boxes, and more wasted of time, in the end I gave up!!! Way to confusing for my computer knowledge and should not be this difficult anyhow!!!., Since price isn't a factor, I'm now looking into paying for some software that works without a degree in computing. willing to pay rather than have this pain. ANY RECOMENDATIONS PLEASE for software with a one off payment plan preferable (not monthly, as I will only use in infrequently)... I can only guess the developers of software haven't updated the software yet? Why do they make what should be a relatively simple Photogrammetry package so difficult?
This type of content is what UA-cam should be about.
Thanks for all of your time, effort and knowledge!
Chris Kaye it is in this UA-cam circle
its all one things like skill share now. good video now i know why my shiny leather couch came out so bad.. time to find a statue.
ok
Do u think able to 3D scan the lift/elevator ?
I was the 666th like
COLMAP is sooo cool, I've just started playing with it. Great video!
it definitely wont work with a budget computer like you featured XD
i had a hard time with my i7 4790k 16gig of ram and a gtx 1060 turbo (6gigs)
i'll give a try with the non cuda version of Colmap ... maybe the 1060 is just not enough
In theory it should just affect how long it takes, if you are patient you could do this with a very basic computer.
Try 3DF Zephyr free. It does all the things in the video but in one tool. Much easier to use as well. Free is caped at 50 photos.
how to install colmap??
Bonjour 🤝 Excellent travail bravo 👏 et merci 🙏
We have released a sequel - Photogrammetry 2 that shows even better results and is much easier to do! :) (But needs an Nvidia GPU to run, unlike this tutorial)
ua-cam.com/video/1D0EhSi-vvc/v-deo.html
10 mins ago I thought to myself - I wonder if you can use 3D scanning to help create Virtual Reality worlds quickly yet affordably. Not only do now know that the answer is yes but that I can come back here to follow step by step. Awesome video. Excellent production values and delivery of guidance. I'm off to the local antiques market to buy cool stuff to snap from every conceivable angle. Thank you for the time and effort you put into this. Legend!
Just what i have been looking for...for YEARS!
it's what i have been already using for YEARS!!!! ua-cam.com/video/y6KRve6NOAs/v-deo.html there are so many iOS apps out there since 2012 already... how this complicated method is better? I don't know!
yeah but it doesn't come out too good now does it??
@@honestview quality
Thanx, man. After watchin this video. I went out into the yard with the phone (the weather is cloudy and the light is nicely diffused) and I took a hundred photos for the test. Very inspiring.
You are awesome. So now I understand how this rotating round stand for small models works. Its just a motor, sheet of round material with some non repeating pattern on it. So you put your camera/phone on the stand and do fast shots while the round stand rotates.
Videos like this make a very happy Prusa customer. Keep up the great work.
quick, clear, right to the point,the kinds of tutorials I love! good work
I recommend Meshroom. I had a lot of problems with the software used in this video. Meshroom lets you modify your result within the programm and you can work with nodes too. It combines all into one and works way faster and better than this outdated software.
ua-cam.com/video/1D0EhSi-vvc/v-deo.html ;)
Some time ago my interest / curiosity in 3D modeling / photogrammetry increased and this video helped a lot, Thanks !!!
it makes me so happy that there are people who care about Linux distributions of useful programs
Super cool! Nice work. I bet allot of ppl are excited about this. Especially the ones that need this type of software. Thanks!👍
You can rotate the object as long as background is in solid color. There are several solutions on internet, mostly by programming Arduino to rotate turntable with the object, then sending remote shutter signal to the camera. This whole process can be automated with very good results.
Also if you don't exactly know if you did enough pictures, you can simply record video instead with you walking around the object then extracting video as separate .jpgs.
That is true, BUT a solid background makes it harder for the program to figure out the position of the camera, so in some cases the background helps in the process
Often it is recommended to put the scanned object on a newspaper or similar to get more reference points in the images
*Has a 3D model*
*prints it*
*Scans it back into a 3D model*
Profit??
and repeat :)
That would actually be an interesting experiment (or art project) - you could observe and measure the losses between each step.
Juho Leiniö 3D inception! :D
Repeat that enough times and you'd probably just end up with a blob, lol.
The perfect lesson as video was intended.
Not sure if Meshlab has it, but with you can further remove noise from the point cloud using a Statistical Outlier Filter (CloudCompare has it). It will remove most of the stranded points that are close to the surface. This will make the initial computed mesh cleaner and easier to work with and smoothing results will look better.
Josef. Once again a great video, short, to the point and plenty of information.
But please, either lower the volume of the background music or eliminate it. It's annoying and is not needed, IMO.
Thanks for all of your hard work. I have purchased four of your printers and have been responsible for at least a dozen of other sales to my friends and people I have demonstrated them to in workshops I present, and articles I have written.
And I was sitting here like, "where do they find all these beats". To each their own.
The common reason you find background music that is a bit too loud in relation to the speech is because people don't have professional facilities and tons of editing time to record their voice overs, so the music hides the annoying background noise - the microphone and preamp hiss, the computer rumble and fan noise, the furniture noises, breathing and mouth sounds, the gargling sound of noise suppression plug-in...
As somebody who suffers from cochlear synaptopathy, I struggle to hear the narration in videos like this with loud background music.
I quite liked the music. No issues with the volume either.
Wonder if taking a burst (a burst of pictures at a rapid rate) would work too, if not better but would probably take a lot more processing power to create the 3d model, since bursts take a lot more pictures at a faster rate then we do manually. My phone can take bursts by holding down the shutter button, it might be especially useful for something like this actually because a lot of phones have similar features and some have things like optical image stabilization which is not only really helpful when taking picture with shaky hands, but for taking still, unwarped, accurate, images for a 3d model as demonstrated in the video.
I love your tutorial format, please keep up the great work.🤜🏼🤛🏼🍀😎
This is such a good video (two weeks into my i3 Mk3.. Having a lot of fun and this takes it up a notch).
Love it! I could see building a dedicated featureless turntable except with landmark "X"s and a featureless, blocking, concave background with a stationary camera mount. Note doubt a low-tech Lazy Susan and poster sized card stock would do.
Josef is the 3D printing hero we need, but not the one we deserve. Awesome video, man!
Awesome video! I've been looking ages for something like this!
This is actually called Structure from Motion, a combination of computer vision and photogrammetry.
No, this is photogrammetry since it’s still camera shots meshed together thru data points to generate a model... the structures he shot were never in any motion
Nice video. Zephyr offers a function where you can upload an mp4 / video file. Pretty cool feature.
Cheesecake This seems like a cool idea. Easy to execute in other software too if you can find a way to slice a video into frames (ffmpeg can do it I think). Question is, would the resulting scans be feasible to process on an average desktop PC... I imagine they would have a LOT of points.
Videos are more compressed then video, and also have rolling shutter and motion blur.
You can render our individual frames with quick time or vdub easily.
3DF Zephyr is probably the most feature filled and user friendly photogrammetry software out there. Also completely free for a 50 photos and less if anyone wants to learn.
I train all my students on Zephyr, definitely my recommendation!
not only is video compressed intra-frame, but it is usually less than half of the resolution of the sensor ... ig you have a phone, you can take really crisp photos, but a FHD video is still just above 1 megapixel
while a frame from a 4k video is just under 5 MP ... let that sink in
можно ли использовать зефир для съёмки предметов (товары для интернет-магазина) ?
i am ready to do this on a car and 3d print it
remember the add ''you wouldn't download a car''
well i will
VisualSFM is also powerful and free, it uses additional CSM tools you need to install for as dense recontruction; it works great with different imager sources.
Hey Jo, great concise tutorial. You covered all of the required steps in a short amount of time. Definitely going to try this given I saw a $5.5K handheld scanner today. Cheers, JAYTEE
you are my hero, that's exactly what I needed for a long time!!!! awesome stuff!
A seriously good tutorial. Well done, quite professional.
I've been wanting to learn how to do this for quite some time. Very helpful tutorial. Thank you, sir!
Awesome. Great overview of all the software steps.
at 3:31 its shows the screened poisson surface reconstruction but mine ends at the voronoi filtering, is it in a different place? was it removed? I'm using the newest version.
You have to pick Surface reconstuction: Screened Poisson
Thanks for sharing. You are probably aware of this, but a drone would be the best tool for capturing 3D images - it is more precise, programmable and has more spatial freedom than humans
Interesting!!! But I would like something faster, I thought that shooting on the phone or SLR would serve to simplify the 3D process ... . I would take the photos and then just with a program do everything. I've tried Meshroom and other programs but to no avail. If I attach a link with an archive of some photos of a 1:18 model you can give me two tips to get a good result? Thank you very much.
That's noting new 123D catch exist for a long time already and it works pretty well
and now i m using my drone to capture photos of big objects. Good tutorial thanks😎
Dog #2 Full Body scanned 3d model
www.desirefx.me/dog-2-full-body-scanned-3d-model/
Hmmzz... software setup needs some streamlining, but I think the best way to make it better would be to first get the rough model and then continue refining it with more pictures, matching the new pictures to already existing model, for that the workflow would have to be more automated tho.
Thanks for all this information. I will definitely give it a try. So powerful with a 3d printer.
Absolute legend you are Mr. Prusa!
Excellent video, one thing I don't think I understand from the video though: Will the point cloud software pick up large planes that are a single diffuse colour with no variation of texture? Like the side of a green bin or something.
Thanks for a great introductory video. -- basic enough to be immediately understandable by noobs, but in-depth enough for anyone who wants to dive straight in to get fully started!
Pěkný video. Nikdy bych neřekl že mobilem můžeš dosáhnout takových výsledků.
This is really awesome! Very cool that it's possible to make a 3D scan with just a phone!
I have a question though: If we aren't supposed to move the object, how can we exactly get a bottom angle of the object? Because Groot in your example is still sitting on the white surface.
Finally, I can put myself in Blender without modeling skills XD
in 3d graphics modeling is only 30% time And even with scan u can give a lot problems! For me faster create by hand this model than dancing with tambourines and recive some models which not good editable
You can use Make Human which is a free app for creating random human models and try to recreate yourself
In your set of photos for photogrammetry (@1:00/7.31) I do not see the photo of the bottom of the model (you suggested that at 0:45/7:31)
Very good introduction, i didn’t know about that process
Thank's for sharing - you know how it works, I am very grateful for minds like yours!
Try Meshroom. It's free and outputs a pre-uv obj and texture image file so you won't need to use two separate software. You can import it into blender and it's good to go.
ua-cam.com/video/1D0EhSi-vvc/v-deo.html ;)
sony xperia xz1 also has 3d scanning depth sensing camera and software.
Very interesting. I tried to do exactly the same, but I still have errors with the filter applied in Meshlab... and I had to export manually the ply file from COLMAP. Maybe that's where the problem is. I'll try to fix it...
thanks man, really appreciate this clear concise presentation
Great video. This opens up so many possibilities!
Nice video. Will be usefull. I will give a try to someof softwares you mentioned.
Yeah, I used to do that in Autodesk 123D in 2012. The models turned out a lot better. Nowadays it's no longer available. It used to be free. Made a lot of scans of my fridge magnets back then. I can print them now. 🤪
Hi Josef, I was hoping you could make a video on octoprint. I have a mk2s and I plan to get a mk3 and i'd like to add octoprint to both but I have no idea how to set it up with either or how to use it. I know a little bit about electronics but the whole idea of it is a little bit intimidating. Thanks!
You sold me to subscribe with that 3D Sculpture snap at the end.
You say take photos of bottom of the model but when you move the model to make it upside down you move it but you also say never move the image? So what to do?
You'll have to get creative and literally get under the model. Placing it on something tall and thin helps.
Or in the case of the statue shown in the video, just trying to get shots from a very low angle to get at least some data from the bottom helps :)
-Mikolas
Very interesting technique. I was wondering if the background is appearing in the images taken, would it not be better to block out the background by setting up chroma colour prop. The image would perhaps be sharper. Just a thought. I suppose a steady hand is a good thing too.
Thanks for posting.
Well, it is still easier to 3D sculpt from the scratch than to scan using photogrammetry in some cases. As some 3D sculpting will be needed when cleaning and patching anyway.
That depends if you have 3D sculpting skills. You can't fault photogrammetry for it's sheer accuracy in capturing objects, even if the topology isn't the greatest. There's a difference between knowing how to use zBrush and being able to replicate an object exactly without letting artistic influence interfere. I think a pipeline of photogrammetry data with alterations from a sculptor is likely to be the way forward.
i see it as with motion capture. you can get insane level of detail but cleanup can be a bit of a nightmare and always needed. as with all 3d sculpting it depends on what the end product is going to be.
Well, you can also paint a photorealistic image in photoshop from scratch, but taking a photo as a first step is usually faster.
@@kpbendeguz that said, photogrametry is not good enough yet for a lot of things (for the video exemple for instance it was easier to just sculpt )
Yes, but it is amazing what you can do with photogrammetry in the case of environmental preservation. Photogrammetry used as a means to give a virtual tour of historical places would be less impressive, had they only been sculpted. That's not to say that skilled sculptors wouldn't be able to do something similar, but it's not quite the same, is it?
will the resolution of the camera ( comparing a mirrorless camera , against a cellphone camera) , improve the quality of the scan? amazing tutorial!
It does, though current phone cameras are actually quite good and very good at exposing the image well, which is also important. :) So with a decent phone camera, we achieved very nice results. -Mikolas
Seems once I get that Prusa Mark 3 I3 I ordered i will need to learn a lot more :) Thanks for the video!!!
Another great option is using a Kinect for Xbox 360 or One or Kinect for Windows 1.0 or 2.0.
Cool , i was wondering if one can uses just a video clip because all the software would have to do is beable to take the .avi or mpeg and split it into a frame by frame images that it then build the .stl file from ? It be easy to just video it rather then take tons of pictures so curious if you can uses videos / camcorders instead. But this picture method is really really cool.
Thanks, I need to make the wheel cap for old car, I hope I can do it.
You are savage dude I am having my great morning today!!!
Awesome information. By the way as a beginner which home 3D printer is best?
Nice tutorial! Now I'll just tape my shiny car!
Hmmm... I personally think meshroom provides better results but this looks ok too.
This is an epic video! Thanks for making this!!
How can you compensate for white powder coated objects? Nice video. Well done.
Well, what do you do with a hollow object? Let's say a basketball that has been sliced in half (and we just want to scan one of the halves), and we need it to be detailed on both the outside and the inside? I've been trying to figure this out for months...
Good info...clearly explained...fyi agisoft is not free anymore
Great video! But I can’t find the file to put into mesh lab ... any help? The ply file doesn’t come up
Very interesting and informative, thank you for putting your time into creating this video!
Wow! This is an awesome tutorial! Thank you!!!
HOO.... radiologists and medical professionals is going to love this ....
Yo. Kolonel Joseph. Can you do one on how to capture sprues from model kits?
This is so dope. I wonder what happens if you just record a video, break it down to frames and import them into the program? Will it improve the quality? Gotta play with this thing...
Александр Болбат it is indeed an interesting idea to use a movie.
And You do not have to make one but to use existing youtube hi quality video :)
I've tried that and it's for sure an option! :) BUT, you have to keep some things in mind, shutter speed needs to pretty high, otherwise the individual frames will be blurred and even if you do so, you'll still have to go through the frames and delete the ocasional blurred one.
Nice information thanks I will give a try to this software . And I have to say 3D Creator from Sony xperia is really easy and awesome and avoid alot of step compared to this traditional method.
Awesome work, Brother
How about we using string to lift the object the use white/green background for all sides?
work if i don't move my camera but i,ll move the rest of the universe?
:D :D :D That'll work
Yes, in theory it should work
Thats actually what happens when you move your camera
Yes it works but avoid it if possible, as the program could use the cam position
There is no spoon
Great, concise video as always. Thanks Jo.
Hola, disculpame la molestia, estuve mirando un tutorial tuyo sobre el scanner para 3d, mi pregunta es la siguiente, me ofrecieron el sense v1, queria saber si es buen equipo a tu criterio. gracias
Ok test! In Meshlab, the cleaning part is the hardest. Unwanted points make ugly blobs. I have used also Regard3D.
A better way to deal with glossy objects it's cross-polarization, I've build some cheap led rings with cross-polarization for macro photogrammetry.
Yeah, for smaller objects, controlled environment with polarized light is better. Maybe something for us to check in a future video :) -Mikolas
Great video man. Say, would it work if my stick tape around an object with a glass material like a glass cup or ceramic mug>
Thanks and yes it would :)
wouldn't it be faster to just film going around it with an infinity focus lens (so the background is not blurry) and have some program make images from every frame.
How to add 3d preview on webpage so you can look around the object on the web page?
Great video by the way!
hi
can you import somehow this code in c# to do it work with intel realsense d415?
Hope you have some idea or code example
THX - and cool video by the way!
Great video and great advices !
But did you try 3DF Zephyr Free ? I find it easier to use and more stable !
Could you use something like a green screen, key out the background, and make a better model?
hello, I have a question: should the photos shoot horizontally or can they also shoot vertically?
Ps: excuse my bad English, it's not my native language, I'm using google translator....... Grazie (Thank's)
The whole object must be in the frame, camera orientation isn't important.
Yes, you can mix horizontal, vertical or any angle in between, it doesn't matter.
Running a brand new PC with Windows 10 (fully updated).
Hit a brick wall within 2 minutes of this demo, Once the Colmap software is downloaded, the initial scan to test the computer fails multiple times, saying 'system Error, code execution cannot proceed because VCOMP120.DLL was not found', this happens multiple times... 3hrs later, many searches, Microsoft advise not to download the ,DLL file on its own, but to reinstall the entire PC! I later found a forum post saying "If a package requires any DLL that is not part of the operating system the developer should add the DLL to the setup program, failing to do so is the developers fault, a bug that nobody should ignore before publishing", not sure if this is true as I'm not a computer Developer, but I decided NOT to reinstall my 1week old install of Windows 10… The path later lead to various old posts telling me to download old software to fix the issue (Visual C++ Redistributable Packages for Visual Studio 2013), I also looked into the updated Visual Studio 2017 software, which lead to an install of hundreds of various tick boxes, and more wasted of time, in the end I gave up!!!
Way to confusing for my computer knowledge and should not be this difficult anyhow!!!., Since price isn't a factor, I'm now looking into paying for some software that works without a degree in computing. willing to pay rather than have this pain. ANY RECOMENDATIONS PLEASE for software with a one off payment plan preferable (not monthly, as I will only use in infrequently)... I can only guess the developers of software haven't updated the software yet? Why do they make what should be a relatively simple Photogrammetry package so difficult?
Try 3DF Zephyr free. It does all the things in the video but in one tool. Much easier to use as well. Free is caped at 50 photos.
Dude that was super cool thank you!
I learned SO MUCH! Thank you. I am a new subscriber.
Wait, does that mean I could technically do that with a car and downscale for a nice 3D print of it(given reflections are removed)? Great video!!
Incredible tutorial..... and ITS FREE...! 😁