Slave 1 photogrammetry tests

thorst

Well-Known Member
Hello,

I have been at the Star Wars Identities exhibition in Cologne some days ago, and was able to take some pictures of one of my favourite ships, the Slave 1. I took them with the intention to try photogrammetry, and I was able to find some free programs to to this, which seem to work pretty well. I am using VisualSFM with MeshLab and cmpmvs to produce point clouds and a textured mesh from the pictures. I was using exclusively the pictures I took myself, to prevent any misunderstandings, you know what I mean ;). Here are some screenshots of the mesh:





Now I am working with Rhino to draw a clean NURBS model for further steps. Aside from grave asymmetry, I found another strange thing:
For absolute scale, I tried to use the body of R2D2 which was on display behind the Slave 1 and which was partially reconstructed by the software. I found the number of 19.35" for the cylindrical part of the body, and scaled the mesh accordingly. Now when I measure the body of the Slave 1, I get a length of almost exactly 26" (from the tip of the "nose" to the end of the skirt), and a width of 24.5" over the wings. However, the Chronicles list it to be 69cm times 66cm, which is 27.2" times 26".

Does anybody know if the measurements given in the Chronicles book can be trusted this far? It would mean, that either the reconstruction of the R2-unit is wrong, or the measurement for it is wrong.

Any help would be appreciated!

Cheers,
Thorsten
 

vfxsup64

Well-Known Member
RPF PREMIUM MEMBER
Ahhh...one of my favorite subjects - photogrammetry.

When I was with the ANH Falcon a little over a year ago, I thought about how to deal with the issue of getting absolute scale references. Certainly donor kit parts are key but for a model on exhibit, inside a display case, the solution was right in front of me...measure the box it came in (or rather the display case surrounding the piece).

So, I very discretely palmed a small measuring tape and ran it along one of the long straight lengths of the display case and made sure a few of my overall picture samples included that side of the display case. Once I solved that edge in the overall photogram and 3D tracking solve, I had a large measurement data point to refer everything else to. A good compare to the inner donor kit parts - and a good way to minimize/average out measurement error.

Now, in lieu of possibly getting frowned upon by the local exhibit staff, one could have a friend stand next to the model, holding something of known measurement - a piece of notebook paper or a walking cane - something that is in view of the camera along with the model - some scale reference to relate everything else to!


(P.S. Make sure that the scale reference does not move between pictures - it must be a rigid part of the "monolithic" structure you are solving for...)



Regards,

Andre
 
Last edited:

thorst

Well-Known Member
Thank you all!

Andre, I thought about measuring the display case, too, but it was impossible as one of the staff members had his post directly at its side and surveillance was very strong (every flash was cared of). The other idea - including something with a known scale is essentially the same as what I tried with the R2-unit. Now I am unsure if the relative large distance from the Slave to R2 is a problem for the solve, or if the model of R2 is just not the one for which I got measurements.

The third option - using donor kit parts - is impossible for me at the moment as I have none of the larger parts. And I guess only the wings are solved well enough to uses them for scaling.

Regards,
Thorsten
 

thorst

Well-Known Member
You've taken good shots and it shows in the results :thumbsup
Thanks! I am satisfied for it being my first photogrammetry project, but the area left of the cockpit, behind the "shpulder" isn't well solved as it was impossible to get good shots there.
No flashlights were a real bummer, too, as you can see between the nose and the skirt.

Cheers,
Thorsten
 

DaddyfromNaboo

Master Member
RPF PREMIUM MEMBER
Hi Thorsten,

I plan to go tomorrow . Does the Slave sit in a cabinet on a rectangular column? I think I can quickly measure the column, why should anyone have anything against that, since I am not touching the model and there sure are maaaany kids who are going to lean against the exhibits ;)

Thanks! I am satisfied for it being my first photogrammetry project, but the area left of the cockpit, behind the "shpulder" isn't well solved as it was impossible to get good shots there.
No flashlights were a real bummer, too, as you can see between the nose and the skirt.

Cheers,
Thorsten
 

dsp5500

Well-Known Member
Place a rule along the cabinet and as long as it is in a few of the shots you can use it for scale in the solve.

Andre already mentioned the same above.
 

thorst

Well-Known Member
Hi DaddyfromNaboo,

if you can do this, that would be great! Thank you for the offer! To be able to use the information, I would need some (at least around 10) pictures which show the Slave 1 as well as the scale reference you use from different angles, such that I can include it in my solve.

The Slave 1 model sits in the corner of the first display case of the exhibition. I think it isn't square, the longer side faces to the entrance and shows the Boba Fett costume and some Stormtrooper helmets. The Slave 1 sits at the left side as seen from the entrance. Behind the Slave is R2. Directly besides the Slave, a staff member was standing all the time when I was there.
I guess measuring the smaller side of the cabin would be the easiest - if the staff doesn't interrupt. Be prepared that the lighting is horrible - I used mostly ISO 6400 and about 1/30 sec exposure time at 1/fstop=4.8. It is important for the reconstruction that the scale reference is actually visible (not too dark), so the distance between the poles of the cabinet would be the best I guess.

Sorry for the long description, but I am still excited by all of this :D

Thorsten
 

DaddyfromNaboo

Master Member
RPF PREMIUM MEMBER
HI Thorsten,

no problem. I am not sure if I´ll take the ol´ camera with me tomorrow, but I´ll take the measurements if possible. So see you tomorrow in this thread ;)
 

thorst

Well-Known Member
Here is a small update:

I managed to get the first iteration of the basic parts drawn as NURBS surfaces. It showed again how much asymmetry this model has, and also how it was improvised in some areas.

Now I managed to get it back into Meshlab to do checks with the original photos. Here are some examples where you can see that I am not too far:







(the original pictures in these comparisons are not corrected for lens distortion, so errors in the corners are not unexpected)

Now I see some problems that I need to address, but overall I am happy with the result, and still so excited about the whole photogrammetry process!

Cheers,
Thorsten
 

Commander Max

Sr Member
I hope you don't mind a suggestion.

How about getting a measurement of one(or more) of the kit parts on the model. In relation to the rest of the model that should give an approximate size. Another is the lip of the platform it's sitting on(edge of the mirror, to the edge of the platform).

This technique reminds me of something that's going to be in our smart phones very soon, a 3d scanner. Today you can get one for less than $400 with decent resolution. A whole lot of people are not going to be very happy the day 3d scanners become common place.
 

thorst

Well-Known Member
Thank you, Commander Max!
I have already got some measurements of wing parts for this purpose from another member here, and within the reading error (approx. 0.3 mm), it agrees exactly with the reconstructed model scaled with the body height of R2-D2. I just didn't post this yet as I wanted to show the final iteration of the surfaces then, too, which I don't have finished yet (I am on iteration #10 now, it's almost there).

Of course, I would be happy for each new measurement I can get! I only can't use anything on the underside as the reconstrction is too bad there.

(Don't want to discuss politics here, so I don't go deeper into the second part here, sorry ;-) )

Cheers,
Thortsen
 

E3kehoe

Member
I'm finding this really interesting and also it's Slave 1 sooo yay!!! But I don't really understand what's going on and why. Sorry. Great work but what is the ultimate goal of doing this kind of work. I just don't understand.
 

thorst

Well-Known Member
E3kehoe, a CAD model can be the base for a "real" model. I am currently researching the possibilities to get plugs for vacuforming milled, based on the computer model. I will never ever be able to afford all donor kits, but if I really start building this, I'd like it to be as close to the original as I can do it, and the basic shape is a thing that I want to have perfect (within my abilities).
 

E3kehoe

Member
Alright that's kind of what I thought was going on. That makes my really excited to watch how all this goes down. :) I would love to have a studio scale Slave 1 more than anything. I just resorted to modifying a toy to get as close to one as I could ,with good results but still not an actual studio scale. I still kinda find it strange that of all the Star Wars ships that no one has a studio scale kit available to purchase..
 

thorst

Well-Known Member
I have never built a studio scale model before, so don't put too much hope into me! :wacko
But I have always been a fan of the Slave 1, and as I saw it live in Cologne I knew I need at least to try it.

And it is also a test for photogrammetry for me. It is an exciting technology that is just becomming advanced enough for dummies like me.

Cheers,
Thorsten
 
Top