Ok. Ran up on this ..
Free program.
Takes a series of photos of an object and converts to a 3D mesh. I am watching the last of some tutorial videos and am about to find me a suitable object.
http://www.123dapp.com/catch
The video tutorials ..
http://www.youtube.com/playlist?list=PL2F8EF70230430FC2&feature=plcp
Looks like this is a seriously kick-ass program guys. I mean .. you should be able to take your Barbie Doll and make a 3d mesh you can then upload to Shapeways etc.
It's not clear to me how well it might work, especially since the program is still in beta testing. Ed, do you plan to try it out? If so, your feedback would be fascinating. -- Russ
Yep. Will do so and report back.
-- Russ
Ok. Here we go.
Back in the 80's there were some Paleontologists working in the Gobi desert in China. They found at the 'Flaming Cliffs' a bunch of remarkable fossils preserved by the desert. One of them was this Oviraptor still in it' shell. An artist took the photographs and re-created the creature in resin. This is what I used to test the program.
This is one of the 29 photos that I took of the model. The program suggests 40-50 photos. What can I say .. I was happily taking photos and drinking beer. I figure .. if I can do this beer'd up .. then says something about the program.
(https://www.finescalerr.com/smf/proxy.php?request=http%3A%2F%2Fimages116.fotki.com%2Fv107%2Fphotos%2F2%2F1709102%2F10681290%2FOviraptorPhoto-vi.jpg&hash=9e45bad063183296ceddd8aabf6b16008eb9420f)
Anyhoo. Took the photos (Aperture priority mode and tripod). Ran the program .. easy. Whole process prob took an hour.
This is the completed screen.
(https://www.finescalerr.com/smf/proxy.php?request=http%3A%2F%2Fimages53.fotki.com%2Fv665%2Fphotos%2F2%2F1709102%2F10681290%2F123Dscreen-vi.jpg&hash=3cb472dea01bb320960391c5b34185ccb6738068)
Exported the file as an .obj file
Loaded the .obj file (mesh) into MeshLab and then exported as a .stl file
This is the screen from MeshLab
(https://www.finescalerr.com/smf/proxy.php?request=http%3A%2F%2Fimages61.fotki.com%2Fv380%2Fphotos%2F2%2F1709102%2F10681290%2FMeshLab_obj-vi.jpg&hash=0ae380baebe60cf2d648befa8c0f1870bd113894)
Imported the .stl file into netfabb Studio Basic to automatically repair any problems.
Exported the repaired file.
This is the netfabb Studio Basic screen
(https://www.finescalerr.com/smf/proxy.php?request=http%3A%2F%2Fimages15.fotki.com%2Fv587%2Fphotos%2F2%2F1709102%2F10681290%2Fnetfabb-vi.jpg&hash=0f48e3911393f91b2cf4c37b3ed47beb259156ac)
Imported the repaired .stl file into Sketchup
(https://www.finescalerr.com/smf/proxy.php?request=http%3A%2F%2Fimages109.fotki.com%2Fv786%2Fphotos%2F2%2F1709102%2F10681290%2Fraptor-vi.png&hash=b22aa31397dfa57af6bc1c639960188008ee4bf6)
Just for the heck of it, rendered the Sketchup object in Kerkythea
(https://www.finescalerr.com/smf/proxy.php?request=http%3A%2F%2Fimages53.fotki.com%2Fv665%2Fphotos%2F2%2F1709102%2F10681290%2FKerkythea-vi.jpg&hash=5752d2613c6b75768b5d45d793073805e4062d3f)
Summation. It works. I thinking that you could get your GI-Joe .. mock up a character with clay .. then process him and create your own 1:24 figures.
If you could keep from moving . .you could have someone take YOU photo(s) and replicate yourself.
All the software I used was FREE. This was the first try and it worked. To me .. that's a couple thumbs up.
Just a point. There is an option under 123D to download a file with higher resolution. I didn't worry about as I was interested in seeing how the program worked. The file I show here is their 'standard' mode .. or whatever. You may also notice that I wasn't too careful cutting the mesh around the base of the stand as again .. the was test.
Very interesting. Thanks for the time looking into this and sharing it.
I wonder if a way to capture the object could be done by placing it on a turntable (like the Tamiya spray stand, or one of those cheaap plastic ones for the center of a table); and then shooting it in 1080HD video, while turning the turntable, then in your computer, using one of the cheap/free film editing programs, that allows you to extract/break down the video into individual frames/images, aand then selecting/ussing from those images for the neede ones.
Good idea, Marc.
Thanks, Ed. That was really inspirational. The possibilities of the program seem almost without limit.
If this thing works well and if the 3-D prints are of sufficiently high quality, it may revolutionize modeling. Then the final frontier will be Marc's specialty: The finish.
Russ
@Marc .. if I understand it correctly, you don't want to turn the object. The reason .. and this is from one of the video tutorials .. is that the 'stitching process' .. uses the background beyond the object .. to help with the stitching process. It really doesn't seem to be that a big a deal. On one of the tutorials this guy is just walking around an object with camera held in hand and snapping pics. The one problem I do see isn't the software .. but just physical limits given to the photographer. If you think about it .. if you were walking around and snapping pics of a Civil War general at a park .. it would work .. except for the top. In the video .. and what I did .. was walk around the object and snap pics at level .. then .. up higher and around. The statue in the park .. how are you to snap pics from above. I would think that the stitching would be problematic say at the top of the statue's head possibly.
@Russ .. Later today after coffee and I wake up I'm going to download the higher detail mesh and see how that goes.
I need to find something that will make a good subject. I have a drill press .. might use that since that could be something I could actually use when 3d printed.
Re-Meshing: I asked for the High-Resolution mesh. It took 5 minutes to completely ask, download and display the higher resolution.
Additional Stictching: There were 4 photos that hadn't been stitched into the Photo Scene. Adding them requires finding 4 points on that photo and the same 4 points on two other photos that HAVE been stitched into the Scene. When you do that you should find something else to do for a bit as it needs to re-compute everything.
Detail Shots: I was just re-watching one of the videos and see where they said that after you do a loop around the object .. you go back for detail shots. Huh. Have to try that.
Ok. Here's the screen-grab of the 123D Catch after bringing up the Hi-Res mesh
(https://www.finescalerr.com/smf/proxy.php?request=http%3A%2F%2Fimages16.fotki.com%2Fv363%2Fphotos%2F2%2F1709102%2F10681290%2F123DHD-vi.jpg&hash=719aa114f15b90a020d01291080a81959e6cc509)
Brought up in MeshLab (so can convert to STL)
(https://www.finescalerr.com/smf/proxy.php?request=http%3A%2F%2Fimages53.fotki.com%2Fv665%2Fphotos%2F2%2F1709102%2F10681290%2FMeshLab_obj_HD-vi.jpg&hash=5a5977e3d2812770fa1c5d051c26cd8b2e6c59a3)
netfabb Studio Basic .. correct any errors
(https://www.finescalerr.com/smf/proxy.php?request=http%3A%2F%2Fimages51.fotki.com%2Fv282%2Fphotos%2F2%2F1709102%2F10681290%2Fnetfabb_HD-vi.jpg&hash=995d6e0b298130d3f7ba96ebffe4628ac62578bd)
AccuTrans 3D. Check for Water Tightness. It's easy to see the level of detail in this photo by looking at the brass plate just below the critter. This is where I looked back and saw that you are supposed to take detail photos afterwards. Ohhh .. I think .. that if I bring the camera in close .. like to this plate it would greatly improve.
(https://www.finescalerr.com/smf/proxy.php?request=http%3A%2F%2Fimages16.fotki.com%2Fv378%2Fphotos%2F2%2F1709102%2F10681290%2FAcccuTrans_HD-vi.jpg&hash=473dfffef784d4e2e80543452ddca8c121dd2066)
Finally .. the mesh visible in Sketchup
(https://www.finescalerr.com/smf/proxy.php?request=http%3A%2F%2Fimages51.fotki.com%2Fv278%2Fphotos%2F2%2F1709102%2F10681290%2FsketchupHD-vi.jpg&hash=0721e90ff780e38167a225828d487196e27087fe)
Later on today I'm going to goto WalMart and pick up a cheap action figure. I think I can take clay and using the figure as a base .. create a new figure I can use in my O scale modeling. Thing is .. want something I can feel comfortable in 'wasting time on' .. including detail shots etc.
Very interesting. I don't get how the program figures accurate depth from a 2d photo.
Quote from: Chuck Doan on May 21, 2012, 07:01:14 AM
Very interesting. I don't get how the program figures accurate depth from a 2d photo.
Chuck. Here's a screen-grab showing the cameras. The program somehow knows the position of each camera .. and has identified points on each photo by each camera that matches points on adjacent photos.
(https://www.finescalerr.com/smf/proxy.php?request=http%3A%2F%2Fimages51.fotki.com%2Fv282%2Fphotos%2F2%2F1709102%2F10681290%2FCameras-vi.jpg&hash=5446483b74ce41737c8e2a85b7b26f5ae609d155)
Very helpful Ed. Marc's video and turntable might work if the turntable had some sort of detail on it's surface.
Will be watching for your next attempt.
8)
Quote from: mabloodhound on May 21, 2012, 08:24:36 AM
Very helpful Ed. Marc's video and turntable might work if the turntable had some sort of detail on it's surface.
Will be watching for your next attempt.
8)
Hmmm. Yeah. If you had something around the perimeter of the tt .. donno .. like a series of symbols that might work. I know that they showed taking photos of an art object in a white room wouldn't work well. They put what looked like post-it notes of various colors on the walls which then gave the program reference points.
This technology is incredibly cool. I hope they eventually produce a Mac version.
I would assume that the stiching program is sim to what you now get in cameras that have panorama functions. My new camera (which IMO is absolutely amazing) does completely seamles panos out of many mutiple fast burst images. This leads me to think that if the turntable and symbols idea works, instead of video, just use the fast burst/continuous shooting mode of the camera and save having to do all the image extraction from video.
Quote from: Ray Dunakin on May 21, 2012, 11:37:20 AM
This technology is incredibly cool. I hope they eventually produce a Mac version.
Ray .. they have a web app too
http://apps.123dapp.com/catch/
Quote from: eTraxx on May 21, 2012, 07:44:31 AM
Quote from: Chuck Doan on May 21, 2012, 07:01:14 AM
Very interesting. I don't get how the program figures accurate depth from a 2d photo.
Chuck. Here's a screen-grab showing the cameras. The program somehow knows the position of each camera .. and has identified points on each photo by each camera that matches points on adjacent photos.
(https://www.finescalerr.com/smf/proxy.php?request=http%3A%2F%2Fimages51.fotki.com%2Fv282%2Fphotos%2F2%2F1709102%2F10681290%2FCameras-vi.jpg&hash=5446483b74ce41737c8e2a85b7b26f5ae609d155)
"Magic happens HERE".
I can't imagine how that program would know the position of the cameras. Unless GPS is more invasive than I realized.
Quote from: Chuck Doan on May 21, 2012, 12:59:38 PM
I can't imagine how that program would know the position of the cameras. Unless GPS is more invasive than I realized.
Chuck .. think of it like this.
You are looking at a square piece of Styrene dead on .. right in front of you. Now ... you move until .. looking through the camera .. the square 'appears' to be 1/2 as wide as it is tall. That would mean you are at 45° from where the first photo was taken. If the square is 1/2 the apparent size then you are twice as far away.