Photo to 3D object
DUDU
Posts: 1,945
This technique exists since a long time, but with several photos of the same object.
Here a software wich seems easy to use to reconstitute a 3D object starting from only one photo:
https://www.youtube.com/watch?v=Oie1ZXWceqM
Did somebody already use it?
Comments
similar to Google Sketchup modeling tequniques
Metacreations used to offer "Canoma" with similar features, apparently some of it is now part of Photoshop:
http://en.wikipedia.org/wiki/Canoma
You could have a look to that program.
http://www.alecrivers.com/3dmodelingwithsilhouettes/
The only issue I've found is that, when importing into carrara, you have to go in the vertex room and merge vertices on a short distance (or use something like meshlab to do it).
What dis program does not do is getting the texture from the photo.
Very interesting Phil, it's a sophisticated spline modeler , thanks for making known this to us !
I looked at a tuto of Sketchup, it doesn't seem very easy…
There is also the free package Autodesk with 123D and all the objects to be downloaded here:
http://www.123dapp.com/Gallery/meshmixer/content
(click on 3D Models tab).
Canoma doesn't exist any more but seems to be integrated partially in Photoshop (where ?)
Thank you for these links !
About 3-Sweep, after long research to download it, it appears that it is an Israeli project which is not finalized yet…
Actually, if it where possible to add backdrops in the spline editor (like you can in the vertex editor), you could do something close.
On second though, you could in the vertex editor with polylines (or curves) and surfaces (rules, coons or gordon) or extrusion.
What's nice in the demo you've shown is that the program uses image contrast to automatically detect edges (like the selector or scissors in 2d programs).
SilhouetteFX is really to expensive for what we do.
We always await the EP's tuto on the spline modeller.:)
Perhaps you know some other, we knows very few things about it's tools.
An image in backdrop in the vertex modeller, how does one make that ?
I had never noticed that, thank you.
This one is free (untried yet)
http://www.gimpel3d.com/
this one may be even better suited :
http://insight3d.sourceforge.net/
http://www.3ders.org/articles/20130910-3-sweep-extracting-editable-objects-out-of-a-single-photo.html
Interesting beta
I still have my old copy of Canoma 1.0, and to be honest....it is a little clunky trying to get perspectives correct when you use several photos on any but the simplest of objects.
I would much rather model and texture in Carrara....soooo much better:-)
3-Sweep looks to be one to watch out for when it released, it looks very clever and easy to use. Modelling for the masses!
I was curious so I checked out the website, very cool stuff.
http://photosculpt.net/
http://www.agisoft.com/
Photosculpt uses 2 photos to generate an object. Its incredibly easy to use, and it generates high resolution objects and textures. It is a 32-bit app, but it can generate some pretty large files even with the memory limitation.
Photoscan can generate objects from multiple photos, although you need a large number to get a good object. It is bundled with Stereoscan, which, like Photosculpt can generate an object from 2 photos.
The two photo modeling only gets you so far, but its worth a look if you need something quick, and you are not going to move around the object a lot.
cdordoni,
That photsculpt is interesting is there anything like it for the mac ? I have used Caroma, years ago, and it worked in a way, though not well enough to be of use. I think you need more than one photo to give the real 3d effect. If you rotate the object and you do not have the side texture it soon becomes apparent.
Thanks for the link.
Starboardtack
I would expect Photosculpt to work pretty well on a Windows emulator on a Mac, since it does not use GPU, or more than 3Gb of Ram.
AGIsoft's Stereoscan was available in Windows, Mac and Linux versions, unfortunately the info pages on the site have been removed. I just checked the AGIsoft forum and its says Stereoscan is no longer supported; however, the functionality to work with 2 photos is apparently possible in Photoscan, and there is a Mac version available.
You could install the Photoscan Standard trial (the Pro version has functions for GIS applications, so you you won't need that) to see if that would work. I saw some comments on the forums there about using Photoscan with 2 images.
Here's a small tutorial about the Spline Modeler. Not extracting buildings from photos or anything like that though.
http://www.daz3d.com/forums/discussion/31940/P30/#476706
Thank you EP, I had not seen this thread.
CDordoni, Photoscan seems interesting, I will try it.
Thanks.
There is also 123D Catch, it is a free program or App, I have used it on my iPhone to capture stuff while out and about.
Anyone wanting to use the work commercially would have to buy the Premium license (1 or 2 year subscription), though its relatively inexpensive.
I have not looked at it in a while, but at one time your work was locked up in the "cloud". Perhaps with the Premium subscription they offer ways to save your work on your own computer in standard formats.
Anyone wanting to use the work commercially would have to buy the Premium license (1 or 2 year subscription), though its relatively inexpensive.
I have not looked at it in a while, but at one time your work was locked up in the "cloud". Perhaps with the Premium subscription they offer ways to save your work on your own computer in standard formats.
I am sure that I have saved as an OBJ and loaded into Carrara. I can't offhand remember if this is standard or if it required purchase of a low cost add-on - if so, it wasn't much!
Someone made a throw-away remark about the software being integrated into photoshop. I own the latest photoshop but can't find the ability to generate 3models from 2d photos. I note someone else asked where this feature is, but no-one responded, is this just disinfo?
I have certainly exported from 123Dcatch to obj but it is a messy mesh of triangles
http://www.sharecg.com/v/63265/gallery/11/Poser/Sunnyspot
http://www.daz3d.com/forums/viewreply/767520/
I spoke about Canoma which was to be integrated in PS because I read that on Wikipedia, but I don't find it either…
http://en.wikipedia.org/wiki/Canoma
You chose a difficult model Wendy, you rented a drone to photograph all that ?
Silhouette FX does it with satellite data (1500 Euros) :gulp: .
Thanks for your reply. I saw a video of someone making a quick model in Blender with 3 photos in orthographic view. They basically just filled the photo parts up with shapes and subdivided it - it looked suprisingly good, I guess the same process could be used in Hexagon or Carrara?