Lamborghini Sesto Elemento

[tom120934]

As you may have noticed, there is no blueprint available for this car. So if you want to model it, you have to deal with refpics only. So many thanks to Thomas 'AMV12' Durand, who posted his own pictures of it herer (http://www.smcars.net/forums/picture...-elemento.html) and congratulations to mcaa666 which already started the car (http://www.smcars.net/forums/work-pr...o-concept.html)

I just love this car. So it's worth the pain to learn a new software, let me introduce you to insight3d. Short story, for the elder ones : insight3d is like Canoma, but free and damn powerful.

Long story : insight3d is free and damn powerful software, allowing you to do photogrammetry (Photogrammetry - Wikipedia, the free encyclopedia) That means it helps you building an accurate 3D model from pictures. You give it pictures from a digital camera, you put some markers on pictures to identify the common points, and insight3d will 1/ compute the position of camera for each picture and 2/ put the points you marked in 3D space.

I attached a picture for a better explanation. You can see on it about 40 points created and displayed on one picture. I use 9 different pictures to create the model.

The workflow is to export from insight3d the cameras location and the vertices cloud, into my usual 3D software. Then, I can use it as an accurate basis to create a spline cage.

That's quite harder than doing LEGO, isn't it ?

[VitaminD]

its like ImageModeler...which i used for my fifth or sixth car already
hard at beginning...but even more convenient than blueprints in some cases
good luck Tom...dont forget about LEGO )))))

[johnas]

Ah so you decided to model this after all, good luck

Thanks for the info, that program sounds interesting, gotta take a look into it.

[Ambarrj]

awesome choice TOM... i think i should also model this car..... gonna start this one now....

[tom120934]

Thanks for the feedback !

VitaminD, nice to see I'm not the only one to use this kind of software. It actually works, see below for my (small) progress.

Johnas, yes I couldn't stop myself any longer About insight3d, it just works but I read some feedback telling the author just doesn't reply any incoming e-mail. I wrote him one as soon as I started the project, to thanks him and to give him some feedback, no reply so far...

Ambarrj, the choice isn't awesome, the car is Don't forget to post your own WIP as soon as possible, I'm looking forward to it !

Here are my progress. I added some pictures, I now have 12 pictures used for 3D reconstruction, with 86 points. And some areas of the car aren't mapped yet...

The first picture below shows the current look of my project within insight3d.

The second picture shows the imported VRML file into Blender : it's just a cloud of points, because insight3d doesn't export edges, only faces. As I created no face, I have only points.

The third picture shows the edges reconstructed, it's still hard to see a car because of the random orientation of the model.

The 4th picture shows some specific edges used to properly rotate the model : the yellow is for horizontal, going from front wheel center to rear wheel, and the white one is the middle of the car.

Last picture shows the model, properly aligned with a mirror modifier. It looks quite fine, except you can see the reconstruction isn't exempt of mistakes : for example, the rear wing isn't symmetric.

[AMV12]

Hi tom120934 ! you're welcome, i'm glad to see that my pics are useful did you achieve to download the archive ? i thought it will be deleted from the server after a while without any download :/

And it is a very useful software you're introducing here ! Especially when you have zero good prints of the car.

I will probably try it on the latest Bentley Continental, i will open a thread if i can create the blueprint at least.

See you !

[tom120934]

Hello AMV12, happy birthday with a few days delay

Yes your pictures were damn useful, as you were kind enough to post the original untouched files from your digital camera. Good resolution, it allowed me to be precise in placing the reference points.

I don't know how it can turn out with the Bentley, because it's much more difficult to spot reference points on smooth curves. For my project, there is a lot of hard bodylines with precise ends. So choosing the reference points was quite easy.

I've completed the model reconstruction with insight3d. I used 15 photos and 171 reference points. I imported the model in Blender, and was able to align it in a pretty clean way. Below are pictures from Blender viewport, I'm now able to start actual modeling, because I don't think doing a fully detailed spline cage will be useful.

[helicase]

those images have natural foreshortening to them from what your directly copying off of. I don't see how you can get away with doing blueprints like that without messing up the proportions of the car.

[tom120934]

I'm not sure to have fully understood what you mean but the picture I've shown aren't blueprints traced from some side picture or top picture. These 3 pictures are orthographic screenshots of a 3D model, reconstructed from a set of real pictures from the real concept car. This model and the source pictures perfectly match, perspective induced deformations are corrected by the reconstruction algorithm.

If I got you wrong, please reformulate, remember I'm just a French guy

[helicase]

No, You understood my question. Well it matches perfect. Nice job! You picked a great car to do blueprints for too!

I just wish real life cameras where orthogonal, that would make it easier on us.

[mcdronkz]

I just wish real life cameras where orthogonal, that would make it easier on us.

You can't get rid of perspective, but with a long focal length lens and some correction work afterwards, you can get pretty close .

[tom120934]

This insight3d software does the job of getting rid of perspective. How does it works ?

Consider a 3D space, where the object stands, and a set of cameras around it. Each camera took a picture. Put each picture in front of its camera, on a transparent sheet. Look from the location of the camera, the picture will perfectly match the object behind it (if you don't consider the optical distorsions induced by camera defects)

Consider a specific point on your object. E.g. a corner of the windshield, or the center of a rim. You can see this point from some of your cameras, so it shows up on some of your pictures. You can draw a straight line doing from a camera to this point : it will pass through the related picture (standing in front of the camera on its transparent sheet, remember ?) precisely on the location where this points stands on the picture. It's the notion of projection of the real world on the picture.

insight3d works using this projection, by reverse computing the projection. Give it the pictures, give it a set of specific points on your model. It will be able to compute the location of the different cameras who shot those pictures.

Once the camera located, it can compute the 3D location of the specifics points, by throwing lines from the camera location to the location of each point on each picture. For a given specific point, all the lines will converge into the real place of this reference point on the model.

So you're able to compute a 3D model from a set of pictures. Of course it's a simplified 3D model : e.g. I only used 171 specific points.

The good thing is to use many pictures. Theoretically, 2 pictures are enough to perfectly locate a point in 3D space, because there will be only one intersection of the 2 lines. But as placing a point on a picture perfectly is not possible, many pictures leads you to a beam of lines, to compute an average position, much more reliable because the errors are minimized.

End of the explanation. Sorry if I was too long but I really think this method is a good tool, even if it's much longer than using blueprint. When there is no blueprint, and when you don't want to create model by eye, I think it may come in handy.

[helicase]

This insight3d software does the job of getting rid of perspective. How does it works ?

Consider a 3D space, where the object stands, and a set of cameras around it. Each camera took a picture. Put each picture in front of its camera, on a transparent sheet. Look from the location of the camera, the picture will perfectly match the object behind it (if you don't consider the optical distorsions induced by camera defects)

Consider a specific point on your object. E.g. a corner of the windshield, or the center of a rim. You can see this point from some of your cameras, so it shows up on some of your pictures. You can draw a straight line doing from a camera to this point : it will pass through the related picture (standing in front of the camera on its transparent sheet, remember ?) precisely on the location where this points stands on the picture. It's the notion of projection of the real world on the picture.

insight3d works using this projection, by reverse computing the projection. Give it the pictures, give it a set of specific points on your model. It will be able to compute the location of the different cameras who shot those pictures.

Once the camera located, it can compute the 3D location of the specifics points, by throwing lines from the camera location to the location of each point on each picture. For a given specific point, all the lines will converge into the real place of this reference point on the model.

So you're able to compute a 3D model from a set of pictures. Of course it's a simplified 3D model : e.g. I only used 171 specific points.

The good thing is to use many pictures. Theoretically, 2 pictures are enough to perfectly locate a point in 3D space, because there will be only one intersection of the 2 lines. But as placing a point on a picture perfectly is not possible, many pictures leads you to a beam of lines, to compute an average position, much more reliable because the errors are minimized.

End of the explanation. Sorry if I was too long but I really think this method is a good tool, even if it's much longer than using blueprint. When there is no blueprint, and when you don't want to create model by eye, I think it may come in handy.

http://insight3d.sourceforge.net/
OMG, that is sweet! i didn't even know that exists. I will def be messing around with that, nice tom!

[adamk145]

Yup there is even an Autodesk program which you can get along with other licenses automatically : Autodesk - Autodesk ImageModeler
It even allows for direct modeling over the pictures, which is pretty helpful when recreating stuff.
Back on topic, those are some great blueprints Tom ! I wish someone could rip NFS:HP's Sesto Elemento model but unfortunately it's only available on PS3/X360 ...

[tom120934]

Indeed, insight3d isn't the only software to do this job. This one is Free (Libre) and available for free, so I chose it. One drawback : no support available. The author doesn't reply (neither e-mail nor support request through SourceForge, where the project is hosted)

About modeling within ImageModeler / insight3d, I would not recommend that because they probably do not offer the same features as our usual 3D software. That's the reason why I wanted to export cameras from insight3d to Blender, to be able to model directly on top of pictures within Blender.

I've been asked by Prazeres how to import into Blender cameras and points. You can export the camera info by 2 different ways : "Export cameras (.txt)" or "Export RealViz project (.rzml, .rzi)". Unfortunately I didn't succeed, because the exported data for cameras (location, rotation) doesn't match the imported model. Opening the .rzi file into ImageModeler didn't work either.

For points, I do export the model as VRML to a text file. Into Blender, use the menu File > Import > "X3D & VRML 97" (the last entry in Blender 2.49b) Do not use the File > Import > "VRML 1.0", it won't work.

[helicase]

check this video i made: YouTube - Autodesk Photo Scene Editor Car Example

[VitaminD]

photo scene editor (project photofly) is not suitable for this kinda calibrations
it needs much more photos
if you have video camera and can shoot an object from different sides then you can try it

[Prazeres]

nice work!

can you explainhow to export the ponts and camera to blender because i tried and i didn't get it. i need to know how to export to blender so lately i can export it to maya

[tom120934]

helicase, VitaminD : I don't know about Photofly, but I guess my own solution need much more work. Well, it doesn't matter to me, as I don't seek texture reconstruction from the photographies.

Prazeres : have you seen my post #15 in this page ?

[Prazeres]

lol sorry i didn't read that one