More technology
Posted: Fri Jul 10, 2015 9:50 pmThis post is a bit tongue-in-cheek, looking at what we might expect in the future, with current trends in technology. There's a lot of reading here, think of it as a blog post. Plus it shows what I do as a hobby, which is basically what I do for a living, only weirder...
A while ago I did a type of creativity workshop, where the idea was to define your 'ideal' app. The whole idea was to keep it pretty much under wraps, just in case you were ever in the position to make it happen, but this was never going to be a money-maker. However I've revisited this now because I've been playing with Autodesk 123D Catch. Not a name that comes tripping off the tongue, I always have to look it up as I never remember it.
By the way, my ideal app was an airport design tool, where you could visit an airport with your camera/GPS/compass/accelerometer-equiped tablet, take photos of everything, and by the time you get home you have a complete 3D model to drop into the simulator:)
Catch lets you create a 3D model of anything you can photograph, within reason, and although it is mainly used for smaller objects, a few people do use it to model buildings, in particular there's a real drive to 'map' and model historic buildings as a preservation project. So maybe this could be the first step in modelling an airport without any scenery design
skills, just a camera and this app! Ok, so it isn't quite the tool I envisioned, in that it doesn't 'know' where the object is, or even the size, until you tell it somehow, and it doesn't integrate with the LINZ database, for your subject building's roof textures, and it doesn't know jack about flightsim scenery, but it is a start, so worth an exploration.
By the way, this is available as an app for Windows tablets, but it doesn't work on my processor -- you'd need something a bit gruntier, such as a Surface, but it does work on an
Ipad, apparently. So I won't be finishing an airport before I get home, since I'll be doing most of the work on the desktop PC. Well, not quite, because most of the work is done in the Cloud, using Autodesk's own servers.
For my first project, I wanted to model a building, so I looked around for something of reasonable size -- and by that, I mean it should fit on a coffee table. There was only one I could find in the house, a small ceramic model of St Stephens Chapel, in Parnell, which used to be my 'local' when I lived there. One day a long time ago my first wife and I were out browsing the shops, and I pointed this out on the shelf, and my wife went back and bought it for me. That was another world, and another time, that's for sure. Marriages come and go, but cute ornaments last forever. (It has been glued once, from memory.)
However this turned out to be a disaster, as the little chapel's glaze was a bit too shiny. You can't really 'scan' anything shiny or reflective, and although you wouldn't think of this particular ornament as shiny, it really doesn't take much shine to rule it out. Any object which has a higher than 'normal' specularity would give trouble. The process relies on the renderer being able to match points from angle to angle, and having a lot of points actually change colour/tone dramatically when the angle of light changes throws the whole process out.
Having failed a 'proof of concept', I decided to go straight to a full-sized building. I looked around and discounted most options, mainly because of reflections, from glass etc. So on to plan C...
Plan C was simply walking around my yard in the rain looking for something to photograph. The obvious answer was -- a pumpkin sitting on a wall. I don't know what it is doing there, but I won't look a gift horse in the mouth...
Basically, I just walked around the pumpkin and took a photo every few degrees -- first from standing height, then from level with the pumpkin, then by holding the camera high over my head. I took a total of 44 pictures, which seemed a good number from watching the tutorials. All the same wide-angle, and all with the same settings, although I did use auto-focus and auto-exposure. I also reduced the image size in-camera, which is something I never do, but in this case I figured that the full size would be over-kill for this sort of thing.
The rest of the process is fairly automatic -- run the Autodesk 123d catch application, upload the photos, wait a bit, then view the 'finished' result. This was a lot better than I thought, although it did model a lot more of my wall than I planned, so I had to crop some off to get rid of the missing geometry which is inevitable in a subject like this.Of course normally you'd want to model an object which you can photograph from 360 degrees, but it does work with this kind of 'diarama' shot, which is restricted to 180 degrees simply because there's a hedge in the way.
So now that I know that it is possible to build an accurate, photo-real 3D model with absolutely no skills at all, the next step was to get it into the simulator. Easier said than done. I had to render it at low resolution, and crop it a lot, as it is just too complex - it has more geometry than a normal simulator version of a medium-sized city. But things change, and in no time we'll have a simulator which can not only cope with this level of detail, but can build its own scenery just by taking it to an airport. In the meantime, I managed to place it in Prepar3D, although the scale is completely out -- you really need to tell it some of the measurements, but the goal here was more to get it showing up.
So if you want to build an airport with absolutely no modelling skills, this is a good place to start -- in a year or two, when it has improved, and our computers have become much more powerful.
A few screenshots to illustrate the process.
Here's the 123D Catch workspace:
A close-up of the pumpkin mesh shows just how complex the geometry is, even after setting to the lowest resolution:
It actually loads into ModelConverterX, which allows me to export as a P3D scenery object:
And here it is placed in Prepar3D, about 50 times too big, but who cares, it works!
And here's a link to the 'project' in the Cloud, if you press the 3D view button, you can rotate/zoom etc. There's a 'full-screen' button which is a good idea.
http://www.123dapp.com/catch/RobsFirstPumpkin/4171571
A while ago I did a type of creativity workshop, where the idea was to define your 'ideal' app. The whole idea was to keep it pretty much under wraps, just in case you were ever in the position to make it happen, but this was never going to be a money-maker. However I've revisited this now because I've been playing with Autodesk 123D Catch. Not a name that comes tripping off the tongue, I always have to look it up as I never remember it.
By the way, my ideal app was an airport design tool, where you could visit an airport with your camera/GPS/compass/accelerometer-equiped tablet, take photos of everything, and by the time you get home you have a complete 3D model to drop into the simulator:)
Catch lets you create a 3D model of anything you can photograph, within reason, and although it is mainly used for smaller objects, a few people do use it to model buildings, in particular there's a real drive to 'map' and model historic buildings as a preservation project. So maybe this could be the first step in modelling an airport without any scenery design
skills, just a camera and this app! Ok, so it isn't quite the tool I envisioned, in that it doesn't 'know' where the object is, or even the size, until you tell it somehow, and it doesn't integrate with the LINZ database, for your subject building's roof textures, and it doesn't know jack about flightsim scenery, but it is a start, so worth an exploration.
By the way, this is available as an app for Windows tablets, but it doesn't work on my processor -- you'd need something a bit gruntier, such as a Surface, but it does work on an
Ipad, apparently. So I won't be finishing an airport before I get home, since I'll be doing most of the work on the desktop PC. Well, not quite, because most of the work is done in the Cloud, using Autodesk's own servers.
For my first project, I wanted to model a building, so I looked around for something of reasonable size -- and by that, I mean it should fit on a coffee table. There was only one I could find in the house, a small ceramic model of St Stephens Chapel, in Parnell, which used to be my 'local' when I lived there. One day a long time ago my first wife and I were out browsing the shops, and I pointed this out on the shelf, and my wife went back and bought it for me. That was another world, and another time, that's for sure. Marriages come and go, but cute ornaments last forever. (It has been glued once, from memory.)
However this turned out to be a disaster, as the little chapel's glaze was a bit too shiny. You can't really 'scan' anything shiny or reflective, and although you wouldn't think of this particular ornament as shiny, it really doesn't take much shine to rule it out. Any object which has a higher than 'normal' specularity would give trouble. The process relies on the renderer being able to match points from angle to angle, and having a lot of points actually change colour/tone dramatically when the angle of light changes throws the whole process out.
Having failed a 'proof of concept', I decided to go straight to a full-sized building. I looked around and discounted most options, mainly because of reflections, from glass etc. So on to plan C...
Plan C was simply walking around my yard in the rain looking for something to photograph. The obvious answer was -- a pumpkin sitting on a wall. I don't know what it is doing there, but I won't look a gift horse in the mouth...
Basically, I just walked around the pumpkin and took a photo every few degrees -- first from standing height, then from level with the pumpkin, then by holding the camera high over my head. I took a total of 44 pictures, which seemed a good number from watching the tutorials. All the same wide-angle, and all with the same settings, although I did use auto-focus and auto-exposure. I also reduced the image size in-camera, which is something I never do, but in this case I figured that the full size would be over-kill for this sort of thing.
The rest of the process is fairly automatic -- run the Autodesk 123d catch application, upload the photos, wait a bit, then view the 'finished' result. This was a lot better than I thought, although it did model a lot more of my wall than I planned, so I had to crop some off to get rid of the missing geometry which is inevitable in a subject like this.Of course normally you'd want to model an object which you can photograph from 360 degrees, but it does work with this kind of 'diarama' shot, which is restricted to 180 degrees simply because there's a hedge in the way.
So now that I know that it is possible to build an accurate, photo-real 3D model with absolutely no skills at all, the next step was to get it into the simulator. Easier said than done. I had to render it at low resolution, and crop it a lot, as it is just too complex - it has more geometry than a normal simulator version of a medium-sized city. But things change, and in no time we'll have a simulator which can not only cope with this level of detail, but can build its own scenery just by taking it to an airport. In the meantime, I managed to place it in Prepar3D, although the scale is completely out -- you really need to tell it some of the measurements, but the goal here was more to get it showing up.
So if you want to build an airport with absolutely no modelling skills, this is a good place to start -- in a year or two, when it has improved, and our computers have become much more powerful.
A few screenshots to illustrate the process.
Here's the 123D Catch workspace:
A close-up of the pumpkin mesh shows just how complex the geometry is, even after setting to the lowest resolution:
It actually loads into ModelConverterX, which allows me to export as a P3D scenery object:
And here it is placed in Prepar3D, about 50 times too big, but who cares, it works!
And here's a link to the 'project' in the Cloud, if you press the 3D view button, you can rotate/zoom etc. There's a 'full-screen' button which is a good idea.
http://www.123dapp.com/catch/RobsFirstPumpkin/4171571
