Sloped Minifig Leg 3817b is wrong

[Willy Tschager]

Guys,

I know we have put a lot of effort into the new 3817b/3816b - Minifig Legs by obsoeting the old 3817/3816 but the new geometry with the sloped backside is clearly wrong:

   

Any ideas how to correct the part without the need to reverse the whole conversion?

w.

[Max Martin Richter]

Guys,

I know we have put a lot of effort into the new 3817b/3816b - Minifig Legs by obsoeting the old 3817/3816 but the new geometry with the sloped backside is clearly wrong:



Any ideas how to correct the part without the need to reverse the whole conversion?

w.

I have how to solve this. But was is the real issue, that the real part doesn't match the LDraw one? Actually it is the slope, that's easy to see.
Is it the pivot point which is slightly off?

[Magnus Forsberg]

This issue is bigger than the regular minifig leg.

It all started with an incorrect assumption from me. I claimed that the footprint of the minifig leg should be inline with the hipstud in 3815.
https://forums.ldraw.org/thread-18102-post-18119.html#pid18119

Nils made a new leg based on that assumption, 3817b.dat
Chris and I reworked most of the patterned parts, and they have been made official.
The only downside was a strange slope on the backside of the leg. Hm....

Then I found this image. I think it comes from the original minifig patent document.


  Technical_drawing_minifigure.png (Size: 117.09 KB / Downloads: 221)
  Technical_drawing_minifigure red line.png (Size: 121.41 KB / Downloads: 223)

Notice the offset of the centre line through the foot.

This would mean that we have to redesign the leg again. The foot has to be moved 1 ldu forward. The foot should be moved, and the front of the leg, but not the shape and surface of the hip.
   

The easy way out of this is to revert back to the old design, 3816.dat,  but....
What should we do about all the legs that have been derived from the leg with the incorrect sloping backside?

   

Faun leg, Pirate leg, Robot leg, Springed leg and the Prostethic leg, and the Dual moulded leg and the Long leg and the Medium leg. 
Philo talked about something being strange in the design of the prostetic leg. Can't find it in the reviews of the part, though.

Maybe now you have begun understanding why I have been so reluctant to talk about this issue.
There is a whole chain of parts designed based on the bad original assumption.

Is there a way out of this?

[Sylvain Sauvage]

Then I found this image. I think it comes from the original minifig patent document.

Notice the offset of the centre line through the foot.
[…]
Is there a way out of this?

(There’s several copies of that technical drawing and they all purport to be from one patent application or another (USA, Australia…).  AFAICT, they all show the same offset.)

That makes the minifig out-of-system.  So we can’t use them for telamons.

Another question is: What does that imply for the geometry of a sitting minifigure?


Also, for what it’s worth, LDD has them totally in-system (that is, if you place a 1x1 plate in the back antistud and one under the foot, the two form a perfect 90° angle).  But then LDD and accuracy….

[Willy Tschager]

That's messy, indeed. I compared 3817 with 3817b. Beside the sloped back the new leg itself is also way smaller than 3817. My question is:

If we move the foot also the flat front in 3817b:

4 4 -2 7.657 -4 -18.804 7.399 -4 -19.23 20 -4 -1.5 20 -4

has to be moved forward?

We have currently 67 patterns of 3817b. Way more than of 3817. I don't think that returing to the original 3817 is a good idea. If possible we should use whatever we can from 3817 b and cycle them to 3817c.

w.

[Gerald Lasser]

May be we should look at the Unity data, will try to find one

[Orion Pobursky]

Based on this discussion, I think the right answer is to rip off the band-aid, create a correct version, triple/quadruple check it to make sure we get it right this time, and update everything.

That's a lot a work but sometimes the hard answer is the right one.

[Willy Tschager]

Based on this discussion, I think the right answer is to rip off the band-aid, create a correct version, triple/quadruple check it to make sure we get it right this time, and update everything.

That's a lot a work but sometimes the hard answer is the right one.

+1

[Max Martin Richter]

I checked the Minifig by placing it between a brick construction.
As far as I can tell the head stud is indeed slightly off, when the figure is placed in a completely straight position. If you try to bring the figure inline with the brick construction the torso and head will be slighlty bent forward (Maybe one or two degrees). The head stud still matches with the brick system to the play of construction.
Unfortunately I can't find my caliper for now and will come up with same measures, when I found it.

[Max Martin Richter]

I checked the Minifig by placing it between a brick construction.
As far as I can tell the head stud is indeed slightly off, when the figure is placed in a completely straight position. If you try to bring the figure inline with the brick construction the torso and head will be slighlty bent forward (Maybe one or two degrees). The head stud still matches with the brick system to the play of construction.
Unfortunately I can't find my caliper for now and will come up with same measures, when I found it.

I did some more research on the legs.
Indeed the pivot point is centered in the round top area but its off compared to stud system while the hip is fully symmetrical.
   
   

Another fact: it is not possible to bring the hip in a straight position, while it is build next to a wall:
   

[Philippe Hurbain]

I did some more research on the legs.
Indeed the pivot point is centered in the round top area but its off compared to stud system while the hip is fully symmetrical.

Another fact: it is not possible to bring the hip in a straight position, while it is build next to a wall:

For the brave soul who will cope with this correction, here is a leg mesh extracted from 3D catalog, compared to 3816/3816b.
Bottom stud hole is not aligned with joint hole, measured offset at center of stud hole is o.7ldu. BUT stud hole is weird, longer in z direction by more than 1 ldu!

[Max Martin Richter]

For the brave soul who will cope with this correction, here is a leg mesh extracted from 3D catalog, compared to 3816/3816b.
Bottom stud hole is not aligned with joint hole, measured offset at center of stud hole is o.7ldu. BUT stud hole is weird, longer in z direction by more than 1 ldu!

I will compare with the real part. But as far as I can tell, the drawing in the initial post seems to be correct.
After thinking what 1K and 1P means I came to the conclusion that it is 1 brick (Klodser) and 1 plate (plade).

[Max Martin Richter]

Here is a first draft of the outside. Need to do the inside tomorrow (or in the evening).

[Magnus Forsberg]

Here is a first draft of the outside. Need to do the inside tomorrow (or in the evening).

I don't see the benefit of using that many primitives. It will "break" into open gaps in hi-res mode.
Keep the work load as low as possible by reusing as much as possible of the current leg.

   

If we move the foot 1 ldu forward there is still a difference of 0.25 ldu, between the 8.75 radius on the hip and the 9 ldu on the heel.
Slope or no slope? 
Move the heel to 8.75 but keep the bottom square centered?

[Max Martin Richter]

I don't see the benefit of using that many primitives. It will "break" into open gaps in hi-res mode.
Keep the work load as low as possible by reusing as much as possible of the current leg.



If we move the foot 1 ldu forward there is still a difference of 0.25 ldu, between the 8.75 radius on the hip and the 9 ldu on the heel.
Slope or no slope? 
Move the heel to 8.75 but keep the bottom square centered?

Let me add the inner structure. Afterwards I'll upload the finished part and then we can start the discussion what we need to change, if this is fine with you.
As we don't have any hips and legs in 2022-6, we have time to create the optimal solution.

[Max Martin Richter]

Finished the leg.

I think just moving sections of the old part will not fully work to solve the actual problems.

[Willy Tschager]

Finished the leg.

I think just moving sections of the old part will not fully work to solve the actual problems.

There is overlap with the hip:

2 24 1.25 8.3149 -3.4441 1.25 9.000193 .000108
2 24 1.25 6.364116 6.364079 1.25 8.314789 3.444143
2 24 1.25 8.314789 3.444143 1.25 9.000193 .000108
2 24 1.25 6.364116 6.364079 1.25 4.5 7.609468
2 24 1.25 8.315202 -3.444197 1.25 8 -3.91587

Also with the studs at the back using a:

1 47 0 8 10 0 0 1 1 0 0 0 1 0 3022.dat

w.

[Magnus Forsberg]

Looks like the entire leg should be moved 0.25 ldu, except the hip joint cylinder and the entire surface around it.

Code:

1 11 0 -12 0 1 0 0 0 1 0 0 0 1 3815b.dat
1 16 -.25 0 0 1 0 0 0 1 0 0 0 1 mfleg.dat

1 47 10 8 8.75 0 0 1 1 0 0 0 1 0 3023.dat
1 47 10 28 -1.25 1 0 0 0 1 0 0 0 1 3024.dat

[Max Martin Richter]

Looks like the entire leg should be moved 0.25 ldu, except the hip joint cylinder and the entire surface around it.

Code:

1 11 0 -12 0 1 0 0 0 1 0 0 0 1 3815b.dat
1 16 -.25 0 0 1 0 0 0 1 0 0 0 1 mfleg.dat

1 47 10 8 8.75 0 0 1 1 0 0 0 1 0 3023.dat
1 47 10 28 -1.25 1 0 0 0 1 0 0 0 1 3024.dat

Yes, I discussed this with Willy yesterday. But I forgot my caliper to check, where the problem comes from.
I’ll check this tomorrow, when I’m back home.

PS: I removed several useless round prims. I have no idea, why I added these in the first iteration. Some of them were totally useless. Chrd+tndis fully covered around by quads and triangles. :-D

[Max Martin Richter]

Looks like the entire leg should be moved 0.25 ldu, except the hip joint cylinder and the entire surface around it.

Code:

1 11 0 -12 0 1 0 0 0 1 0 0 0 1 3815b.dat
1 16 -.25 0 0 1 0 0 0 1 0 0 0 1 mfleg.dat

1 47 10 8 8.75 0 0 1 1 0 0 0 1 0 3023.dat
1 47 10 28 -1.25 1 0 0 0 1 0 0 0 1 3024.dat

I got access to my caliper, it's easier than I thought.
The underside, I modeled is 0.25 LDU off.

The inner side is just 1.1 mm, the outside is 1.3 mm, I modeled 1.2 mm. I'll fix this...

[Max Martin Richter]

Here it is...

[N. W. Perry]

Once the geometry is worked out, I'd like to help in the changeover of patterned versions, if I can!

[Orion Pobursky]

Once the geometry is worked out, I'd like to help in the changeover of patterned versions, if I can!

Thanks for volunteering. We're waiting on a couple of independent measurement reviews. I'd like to get the base leg revision on the tracker and certified. Then we can work on pattern revision.

[N. W. Perry]

Thanks for volunteering. We're waiting on a couple of independent measurement reviews. I'd like to get the base leg revision on the tracker and certified. Then we can work on pattern revision.

No hurry. I'm still trying to write my buffer exchange script for LDCad. 

[Willy Tschager]

Here it is...

Looks extremly accurate to me.

* May I ask to eliminate the T-junk on the inside:

2 4 4 28 4.75 4 24 4.75
2 4 4 24 4.75 4 18 4.75
2 4 4 18 4.75 4 12 4.75
2 4 4 12 4.75 4 5.5 4.75

* I'm totally unhappy with this long streched tringle:

3 16 1.25 28 8.75 1.25 8.75 6.187125 1.25 6.187125 6.187125

For the rest it would be a go from my side. Well done Max!

w.

[Max Martin Richter]

Looks extremly accurate to me.

* May I ask to eliminate the T-junk on the inside:

2 4 4 28 4.75 4 24 4.75
2 4 4 24 4.75 4 18 4.75
2 4 4 18 4.75 4 12 4.75
2 4 4 12 4.75 4 5.5 4.75

* I'm totally unhappy with this long streched tringle:

3 16 1.25 28 8.75 1.25 8.75 6.187125 1.25 6.187125 6.187125

For the rest it would be a go from my side. Well done Max!

w.

I'll fix these things in the evening and upload it as c version. 
Question from my side: Should I take the history from the b version and add another line or should there be no history for c as it is a completely new file?

Side note: the upper inner structure has a flat area in its back to imprint L for left and R for right there. I haven't modelled this detail as there is no benefit for the file nor there is any connecting or something else affected by not having this detail.

Max

[N. W. Perry]

I'll fix these things in the evening and upload it as c version. 
Question from my side: Should I take the history from the b version and add another line or should there be no history for c as it is a completely new file?

Side note: the upper inner structure has a flat area in its back to imprint L for left and R for right there. I haven't modelled this detail as there is no benefit for the file nor there is any connecting or something else affected by not having this detail.

Max

Seems to be a misplaced primitive floating around:
1 16 13.75 .05 1.25 0 1 0 8.75 0 0 0 0 -8.75 1-16chrd.dat

[Max Martin Richter]

Seems to be a misplaced primitive floating around:
1 16 13.75 .05 1.25 0 1 0 8.75 0 0 0 0 -8.75 1-16chrd.dat

Jops, I already fixed this in the uploaded file on PT.

[Magnus Forsberg]

Jops, I already fixed this in the uploaded file on PT.

No, that 1-16chrd is missing, and needed. See my review.

[Max Martin Richter]

No, that 1-16chrd is missing, and needed. See my review.

Mmh, yes, I removed it before the upload as it was misplaced (and forgot to check, where it comes from). I'll fix it now.

I would suggest to move the further discussion to the PT.

[Philippe Hurbain]

I'll fix these things in the evening and upload it as c version. 

I made an experiment to see how difficult it is to adapt an existing pattern with "b" leg geometry to corrected "c" geometry.
See https://www.ldraw.org/cgi-bin/ptdetail.c...6cpc6a.dat (and related)

I chose this one as it was a good case with front/side/toe pattern.
The most tricky part is the adaptation of front curved subpart, shortened at bottom. Adaptation made using LDPE only.

Credit/history scheme follows the same pattern used when going from initial to "b" geometry.

[Magnus Forsberg]

I made an experiment to see how difficult it is to adapt an existing pattern with "b" leg geometry to corrected "c" geometry.

I chose this one as it was a good case with front/side/toe pattern.
The most tricky part is the adaptation of front curved subpart, shortened at bottom. Adaptation made using LDPE only.

What tools did you use? 
I remember using "Merge to Nearest Face (Direction)" and "Merge to Nearest Edge" a lot.

[Philippe Hurbain]

What tools did you use? 
I remember using "Merge to Nearest Face (Direction)" and "Merge to Nearest Edge" a lot.

That's more or less what I did too. Leg straight part and toes moved with "set xyz" with proper z value. Whole pattern translated by 0.25 ldu in x direction. I also had to shrink a bit vertically the bottom golden ring to adapt to the modified geometry.

[Magnus Forsberg]

Faun leg, Pirate leg, Robot leg, Springed leg and the Prostethic leg, and the Dual moulded leg and the Long leg and the Medium leg. 

Back to the leg issue.

What do we do with all the legs derived from the bad design?

• Faun leg, 24083:  Should stay as it's made, since it's never used in a mixed assembly.  Edit: It's now corrected.
  

• Pirate leg, 2532:  Should it too have the foot moved forward 1.25 ldu?  IMO yes. I think it's wrong.  Edit: No need to change, as described by Philo. All shortcuts are updated, using the c-version leg.
  

• Robot leg, 10177:   Need to be changed, leg is ok, but the foot need to match the regular leg.  Edit: It is now corrected.
  

• Springed leg, 43220:  Not affected. Should stay as is.
  

• Prostethic leg, 80690:  Need to be changed, to have the hole inline with the right leg.  Edit: It's now corrected.
  

• Dual moulded leg, 20460:  Need to be changed, to have a flat non-sloping back side.  Edit: It's now corrected.
  

• Long leg, 87775:  Need to be changed, to have a flat non-sloping back side.   Edit: It's now corrected.
  

• Medium leg, u9441:  Need to be changed, to have a flat non-sloping back side.   Edit:  It's now corrected.
  

The only big issue is the pirate leg. How should it be changed to match the new position of the foot?
Or is that difference explained in another way?

[Orion Pobursky]

What do we do with all the legs derived from the bad design?

From my standpoint, the solution that accurately represents the real world part is the correct solution.

[Willy Tschager]

Faun leg, 24083:  Should stay as it's made, since it's never used in a mixed assembly.

   

IMO it has to be corrected, as the stud cavity has been modeled with 3817b as reference and is off. I have to check the rest. I would prefer that we tackle them one by one.

w.

[Max Martin Richter]

IMO it has to be corrected, as the stud cavity has been modeled with 3817b as reference and is off. I have to check the rest. I would prefer that we tackle them one by one.

w.

+1

[Magnus Forsberg]

IMO it has to be corrected, as the stud cavity has been modeled with 3817b as reference and is off.

"Should" was the wrong word. It could be unchanged, as I said, it is never used in a mixed assembly.
But if I compare my version to the ldd reference, the foot is indeed moved. Should be corrected.

[Philippe Hurbain]

The only big issue is the pirate leg. How should it be changed to match the new position of the foot?

I checked LDraw model of 2532 pirate leg against 2532 model from Unity microgame, they match perfectly so pirate leg is fully symmetrical in Z direction. This makes sense to use it for left or right leg! (Though it seems to be used only for right leg in existing minifigs).

Or is that difference explained in another way?

LEGO solves this by modeling the stud hole as a rectangle (screenshot showing LDraw 2532 + leg model from catalog app)!!! (4.99mm x 5.45mm)
But IRL I measure a perfect square of 4.8mm. So I guess that a pirate is very slightly rotated (or flexibility everywhere does the trick!)

Code:

1 6 -0.01940531 13 84.37732055 0.99805947 0 0.06226795 0 1 0 -0.06226795 0 0.99805947 2532.dat
1 47 0 41 85 0 0 1 0 1 0 -1 0 0 3021.dat
1 1 -0.01940531 1 84.37732055 0.99805947 0 0.06226795 0 1 0 -0.06226795 0 0.99805947 3815b.dat
1 1 -0.01940531 13 84.37732055 -0.99805947 0 -0.06226795 0 1 0 0.06226795 0 -0.99805947 3816c.dat

[Willy Tschager]

Robot leg, 10177:   Need to be changed, leg is ok, but the foot need to match the regular leg.

   

Confirmed. Magnus, since you authored the original could you please adjust it?

w.

[Magnus Forsberg]

Magnus, since you authored the original could you please adjust it?

Yes, I'll do that one.

[Magnus Forsberg]

Medium leg, u9441:  Need to be changed, to have a flat non-sloping back side.

Working om the still unofficial medium leg.

[Magnus Forsberg]

• Faun leg, 24083:  Should stay as it's made, since it's never used in a mixed assembly.  Edit: It's now corrected.
  

• Pirate leg, 2532:  Should it too have the foot moved forward 1.25 ldu?  IMO yes. I think it's wrong.  Edit: No need to change, as described by Philo. All shortcuts are updated, using the c-version leg.
  

• Robot leg, 10177:   Need to be changed, leg is ok, but the foot need to match the regular leg.  Edit: It is now corrected.
  

• Springed leg, 43220:  Not affected. Should stay as is.
  

• Prostethic leg, 80690:  Need to be changed, to have the hole inline with the right leg.  Edit: It's now corrected.
  

• Dual moulded leg, 20460:  Need to be changed, to have a flat non-sloping back side.  Edit: It's now corrected.
  

• Long leg, 87775:  Need to be changed, to have a flat non-sloping back side.   Edit: It's now corrected.
  

• Medium leg, u9441:  Need to be changed, to have a flat non-sloping back side.   Edit:  It's now corrected.
  

All the legs, and their shortcuts, are now reworked. Please help and give them a review.
We should remove the legs with bad geomerty as soon as possible.

[Orion Pobursky]

Note: the obsoleted parts do not need to be reviewed, I will fast track those. The new, corrected parts do need reviews and I'd like them ready for the next release which I am targeting for the weekend of 6/17 but will delay if needed.