T1 and T2 Endoskeleton Research Thread

[Too Much Garlic]

ghostman showed me that I had trimmed too much of the bottom spine piece, so I went back and redid it and only separated the two pieces, then trimmed the second piece to where I thought it should be trimmed to... I may trim more... to match the remnants of lines going down the sides and the back of the piece...

Then I started shape deforming the pieces. This is just a WIP to see if I could do it, but it turned out so well. I can see where I should do things differently next time. On the bottom piece I pulled out the top right corner at the front (left in pictures), but I should have pulled out both top and bottom instead. I'll do that next time.

Then I took the second spine. And that same right forward corner I pulled it up so it got more in line with how high it was on the left side.

Then I took the third spine piece and tried to pull up the bottom a little, so it didn't extend as far down as it did.

And here's a picture showing the difference between the split parts on the left, and then the shape deformed parts on the right. It's not perfect and would prefer a scan off a real SWS spine. But since that is probably not gonna happen unless I get rich and just buy one... which... is highly unlikely.

The top surface of the 1st spine and the bottom surface of the 2nd spine hasn't been filled yet.

 

[Too Much Garlic]

Viewing these pictures show me that I should pull out both right corners of spine 1 and 2. Seems they were pushed in.

 

[Too Much Garlic]

Had to take a break...

Tomorrow I will start working on closing the open parts of the models. Then I will tweak the shapes some more.

I kinda wonder if the DeltaDesignRus method used on the skull could work here getting the warp out of the scans and get the accurate shape back by fixing them indiividually?

 

[DeltaDesignRus]

Had to take a break...

Tomorrow I will start working on closing the open parts of the models. Then I will tweak the shapes some more.

I kinda wonder if the DeltaDesignRus method used on the skull could work here getting the warp out of the scans and get the accurate shape back by fixing them indiividually?

The method will work in one case only if there are photos of the original parts from all possible angles, so that there are at least 50 photos. It's a bit easier with the skull; it has many corners, seams, and holes where markers can be placed. But with the vertebrae, there are fewer such parts, and in the film, the vertebrae are not fully visible.

 

[Too Much Garlic]

Yeah, I kinda reckoned it could be tricky.

 

[ghostman]

Further polish and clean.
The areas where details have been spoiled will be carved out later.

View attachment 1788263

Wrap the scan parts in a symmetric topology.
By averaging the left and right coordinates, the model can be deformed into a symmetrical model.
However, distortion is not removed.

 

[DeltaDesignRus]

Wrap the scan parts in a symmetric topology.
By averaging the left and right coordinates, the model can be deformed into a symmetrical model.
However, distortion is not removed.

View attachment 1788969

Do I understand correctly that you create a low-poly mesh on top of the scan, and then when you deform the low-poly, the scan also changes shape?

 

[Too Much Garlic]

Man... I wish I could do that with my Vader scans, so I can make completely symmetrical Vader armor.

 

[DeltaDesignRus]

Man... I wish I could do that with my Vader scans, so I can make completely symmetrical Vader armor.

The method of creating symmetry that ghostman uses looks very interesting and should be perfect for Vader. However, with the Terminator, it's a bit more complicated, and it's a matter of taste. Personally, I think that if the original was asymmetrical and crooked, then if the goal is to create an exact copy, then all parts should remain crooked and asymmetrical.

 

[DeltaDesignRus]

I'm more interested in whether it's possible to use the ghostman method to change the shape of the scan (high-poly mesh) by moving points on the low-poly model. If yes, how is this done in ZBrush?

 

[Too Much Garlic]

Well, ghostman has his own project going where he wants a symmetrical endoskeleton. So that's his own thing, but I love that he's showing it off here, while we are also working on making the most accurate endoskeleton to the production look.

And yes. Working in low poly to fix something in high poly seems so much more easier.

 

[DeltaDesignRus]

I really like how the skull looks here, but I don't fully understand why it looks so cool here. Is it because of the differences in shape, or lighting? Also, in these shots, the skull is in the top left corner, and the lens optics (distortion) stretched the shape of the head, making it look more fierce and aggressive. I hope when I finish converting my M1 to T1, it will at least somewhat resemble these photos. Additionally, besides symmetry, I like the crooked teeth of the original terminator. Asymmetry and crooked teeth make the terminator skull even more grotesque, making it look even more menacing.

 

[ghostman]

Do I understand correctly that you create a low-poly mesh on top of the scan, and then when you deform the low-poly, the scan also changes shape?

It is only shown in low poly for illustration purposes, but in reality it wraps more than 3 million polygons.

 

[ghostman]

T-800 is not symmetrical but distorted because it is handmade.
If it were a real robot, it would not be distorted.
My project is not to reproduce a prop, but to make an endoskeleton without distortion.


View attachment 1737933View attachment 1737934

My purpose is not to reproduce props.
I have written about that before.
I'm not saying that the main goal is to remove distortion, even to the point of perfect symmetry.
To some extent, leave some areas unsymmetrical.

 

[laellee]

Just jumping in as I follow this thread, and the above discussion is in my wheelhouse you could say. I'll try to keep this short, but just my two cents.

I have dealt with a lot of original and production-used prop scans over the years, and the quality of those scans is all over the place. An Iphone scan is not going to compete with a $20k commercial scanner, but even those high-end scans can leave a lot to be desired depending on the user, the settings, and the software the scanner data is being run through. These scans invariably end up needing a ton of cleanup and correction to better represent the scanned item, and (in my experience) the majority of that work is done manually in the cad software and takes a significant amount of time and skill.
As was described above, another option is to take an original scan and retopologize a 'new' model over the original, creating a fresh low-poly model directly over the original that can then be enhanced to any level. I can duplicate every asymmetric detail, every dent and scratch, all of it, down to levels of detail beyond what the eye can see, as long as those details were recorded in the initial scan; for me it comes down to how many hours I want to pour into the process. This method does also make it much easier to make 'perfect' symmetrical replicas as well, but I use it predominantly to make (for lack of a better word) 'digital casts' of those original pieces with all the warts and goofs of the original piece; with enough elbow grease and patience, those fresh models can equal and in some ways even rival (think shrinkage) traditional silicone/whatever casts.

Just my two cents, sorry if it's a bit of a derail but I monkey with this stuff quite a bit

 

[Too Much Garlic]

Love all the insight in how to work and fix 3D models and scans.

 

[gizmo]

Would love to see some of your work Laellee. Sounds fantastic.

Ben

 

[DeltaDesignRus]

My purpose is not to reproduce props.
I have written about that before.
I'm not saying that the main goal is to remove distortion, even to the point of perfect symmetry.
To some extent, leave some areas unsymmetrical.

I'm not saying you're doing anything wrong, so everything's fine. But I can't help but notice that the details of the robot body really should be more symmetrical, whereas the head doesn't necessarily have to be. Since the Terminator has the appearance of a specific person, and that person almost certainly doesn't have perfectly symmetrical facial features, their skull isn't symmetrical either. This means that to replicate the shape of a person's face, Skynet would have had to create slightly asymmetrical skulls for each appearance. Even Arnold has a very asymmetrical face, remember how crooked his glasses were in T2. I also like the idea that all the differences in Terminator skulls are just features of their construction under the biological shell (the human appearance they were mimicking), which explains why Terminator skulls in the universe differ from each other.


Moreover, I'm very interested in observing your work, and I would like to see its progress more often. You have a very pleasant mesh topology. I myself have been wanting to start retopologizing my skull for a long time (when I finish its more accurate version), ensuring that its topology is correct so that the surface is smooth without dents when smoothed. I wonder, based on which skull your skull is made, and what are its temple details?

 

[Too Much Garlic]

I agree that I feel they made skulls for each model to match the face. Which is why I do not like the Genisys skull - it does NOT match Arnold's face.

Also why I don't like the TX endoskeleton design from T3. Feels like it's from an entirely different movie - much more alien. Would have preferred they made it match Kristanna Loken's look more. But on top of that, then there was the issue with how they used and portrayed that character that just felt too cheesy and dumb, instead of intense and scary. They could have made her genuinely frightening.

 

[DeltaDesignRus]

Does anyone know if this vinyl kit from Thailand is a copy of the M1, or did they purchase the original molds (which have reached the end of their lifespan) and started producing kits themselves?