Most Common Parts that remain missing from LDraw

[tom alphin]

Happy 2025!

As some folks here know, I launched a LEGO Parts Guide in 2024 which includes detailed analytics to determine the 'most common' parts based on # pieces produced across all sets, # sets using the part, # colors produced in, # of years produced.  The default sort is the "Overall Rank" which is a 'hybrid' ranking metric that factors in both #pieces & #colors.

I wanted to start a new list to capture very common LEGO parts which remain missing from LDraw.  In particular, I'd love to see the roughly 100 parts (out of the top 2000 parts of all time) which aren't in LDraw yet!

• #1064 - Large Figure Foot, Type 2 (Part 90661)
  
• #1221 - Rubber Band Small 13mm (Square Cross Section) (Part 71509)
  
• #1231 - Belville, Equipment, Bow Small (Part 30112c)
  
• #1278 - Helmet, Sports [Plain] (Part 93560)
  
• #1349 - Large Figure Head Connector Block 3 x 4 x 1 2/3 (Part 32553)
  
• #1388 - 2×2 Clikits Star (Part 45463) - Feelings about Clikits aside, this was a very common part despite only being used in sets from 2003-2005
  
• #1462 - String Cord Thin 50cm (Part 56823c50) - How should I think about String within LDraw?  Is it assumed that people just use LSynth?
  
• #1478 - Duplo Bed 3 x 5 x 1 2/3 (Part 4895) - I don't think the other DUPLO beds (part 4886) is in LDraw either.
  
• #1484 - Clikits Icon, Star 2 x 2 Large with Pin (Part 45462)
  
• #1520 - Duplo Drawer 2 x 2 with Semicircle Cutout (Part 6471)
  
• #1529 - Rubber Band Medium (Square Cross Section) Thin Cut (Part 70012)
  
• #1551 - Clikits Icon, Star 2 x 2 Small, with Hole (Part 46286)
  
• #1555 - Duplo Suitcase (Part 6427) OR #2203 Duplo Suitcase / Bag (Part 20302)
  
• #1608 - Clikits Icon, Heart 2 x 2 Large with Pin (Part 45449)
  

I stopped at #1500 for now, but there are a lot more you can find in the programmatically-generated list at https://brickarchitect.com/parts/not-in-ldraw.php

---Tom Alphin


P.S. Interestingly, I am observing about the same number of missing LDraw Parts when scoping my list to sets from the past 5 years (113 parts without an image) as when I scope to sets released across all years (87 parts without an image). 

It's super interesting to see how LEGO parts fit into a typical long-tail distribution with the vast majority of LEGO pieces being produced representing just a small number of the most common moulds.  Maybe I can do an article with some tables showing this long-tail distribution and how it correlates to parts which have / don't have 3d models in LDraw as well.

[Jeff Jones]

maybe i could author 90661

[tom alphin]

maybe i could author 90661

That would be great :-) 

I look forward to not seeing the 'no image' icon in the future!

[Rene Rechthaler]

that list already has some pictures (LDraw renders), which comes from the different naming between sites...
rebrickable is quite good for that because it has links to the other pages including LDraw.
Some parts also have different variants, which also get lost sometimes.
Other parts like the rubber bands arent in LDraw yet because they mostly get modeled according to their use.
Other Themes like Scala/Belville, Duplo, Clikits and Bionicle arent popular and hard to model.
Some arent just linked, for example, these two are the same part but not linked
https://brickarchitect.com/parts/7b
https://library.ldraw.org/parts/8912
same here (but idk which variant matches):
https://brickarchitect.com/parts/3324
https://library.ldraw.org/parts/5966

[tom alphin]

Some arent just linked, for example, these two are the same part but not linked
https://brickarchitect.com/parts/7b
https://library.ldraw.org/parts/8912
same here (but idk which variant matches):
https://brickarchitect.com/parts/3324
https://library.ldraw.org/parts/5966

I have manually linked many of these parts.  Those are good examples that I have missed, thanks.  This is why I'm going to continue adding parts that I have confirmed are missing to the OP in this thread.

[Gerald Lasser]

Happy 2025!

As some folks here know, I launched a LEGO Parts Guide in 2024 which includes detailed analytics to determine the 'most common' parts based on # pieces produced across all sets, # sets using the part, # colors produced in, # of years produced.  The default sort is the "Overall Rank" which is a 'hybrid' ranking metric that factors in both #pieces & #colors.

I wanted to start a new list to capture very common LEGO parts which remain missing from LDraw.  In particular, I'd love to see the roughly 100 parts (out of the top 2000 parts of all time) which aren't in LDraw yet!

• #1064 - Large Figure Foot, Type 2 (Part 90661)
  
• #1221 - Rubber Band Small 13mm (Square Cross Section) (Part 71509)
  
• #1231 - Belville, Equipment, Bow Small (Part 30112c)
  
• #1278 - Helmet, Sports [Plain] (Part 93560)
  
• #1349 - Large Figure Head Connector Block 3 x 4 x 1 2/3 (Part 32553)
  
• #1388 - 2×2 Clikits Star (Part 45463) - Feelings about Clikits aside, this was a very common part despite only being used in sets from 2003-2005
  
• #1462 - String Cord Thin 50cm (Part 56823c50) - How should I think about String within LDraw?  Is it assumed that people just use LSynth?
  
• #1478 - Duplo Bed 3 x 5 x 1 2/3 (Part 4895) - I don't think the other DUPLO beds (part 4886) is in LDraw either.
  
• #1484 - Clikits Icon, Star 2 x 2 Large with Pin (Part 45462)
  
• #1520 - Duplo Drawer 2 x 2 with Semicircle Cutout (Part 6471)
  
• #1529 - Rubber Band Medium (Square Cross Section) Thin Cut (Part 70012)
  
• #1551 - Clikits Icon, Star 2 x 2 Small, with Hole (Part 46286)
  
• #1555 - Duplo Suitcase (Part 6427) OR #2203 Duplo Suitcase / Bag (Part 20302)
  
• #1608 - Clikits Icon, Heart 2 x 2 Large with Pin (Part 45449)
  

I stopped at #1500 for now, but there are a lot more you can find in the programmatically-generated list at https://brickarchitect.com/parts/not-in-ldraw.php

---Tom Alphin


P.S. Interestingly, I am observing about the same number of missing LDraw Parts when scoping my list to sets from the past 5 years (113 parts without an image) as when I scope to sets released across all years (87 parts without an image). 

It's super interesting to see how LEGO parts fit into a typical long-tail distribution with the vast majority of LEGO pieces being produced representing just a small number of the most common moulds.  Maybe I can do an article with some tables showing this long-tail distribution and how it correlates to parts which have / don't have 3d models in LDraw as well.

I have seen that a few parts are not pictured although the part is available

Could you upload this to google sheets to be able to better comment on it. We can always refer to if from here or have a detailed discussion on some parts.

It would also be easier to add the raw-source or add an effort estimation to do the part. Older Duplo stuff seems pretty fine to be done from pictures.

Thanks

PS: I made a google Sheet here
https://docs.google.com/spreadsheets/d/1...id=0#gid=0

[Rene Rechthaler]

Added category "electric/flexible", changed "Diplo"(typo) into "Minifig part", sorted some parts into categories...
any way to auto-assign some categories according to the name? (Duplo/Scala)(Large Figure -> Constraction)

[tom alphin]

PS: I made a google Sheet here
https://docs.google.com/spreadsheets/d/1...id=0#gid=0

That's fantastic and seems like a much easier place to take notes of progress.

How did you assign the 'area' categories?  Was this a manual process, or did you derive them from the part names?

[Rene Rechthaler]

That's fantastic and seems like a much easier place to take notes of progress.

How did you assign the 'area' categories?  Was this a manual process, or did you derive them from the part names?

mostly manually so far...

I wonder if this way works: filter the name (for example "Scala") then assign the "area" to all visible...
(didnt want to try)

[Gerald Lasser]

That's fantastic and seems like a much easier place to take notes of progress.

How did you assign the 'area' categories?  Was this a manual process, or did you derive them from the part names?

yes, I did it manually as I was walking through them.

The basic stuff I could do with a "Left=" query, that would be quicker to single out Duplo, Constraction and Scala/Belville.

I sorted it right now by"used until" and an estimate is that around 420 should be avaialble from the BI-App

[tom alphin]

I sorted it right now by "used until" and an estimate is that around 420 should be avaialble from the BI-App

What is the rule of thumb for parts in the BI app?  anything newer than a certain date is usually there?

Further, what is the process of extracting part from BI and making it work within LDraw?

[Gerald Lasser]

What is the rule of thumb for parts in the BI app?  anything newer than a certain date is usually there?

Further, what is the process of extracting part from BI and making it work within LDraw?

You could say from 2018 most parts should be in there. Digital Bricks goes back a bit further, but not for all themes.

Every part is a Unity Asset and available in the App after you select the set to build.
Use an Asset Browser to look for the mesh, export the mesh as obj. And then use obj2dat to convert it.

[tom alphin]

Use an Asset Browser to look for the mesh, export the mesh as obj. And then use obj2dat to convert it.

In all seriousness, couldn't the LDraw project automate that process and add thousands of (provisional) parts to the database overnight?  It's really remarkable that The LEGO Group is giving us the meshes for almost every new part!

Automating that process could benefit the LDraw community greatly, allowing the LDraw community to use it's limited resources to focus on fixing bugs and creating 3d models for vintage parts.  It might require evolving the 'good enough' criteria for inclusion in the parts database, or introducing a middle tier of 'provisional' parts that are 'unofficial' but installed automatically as part of app installer, or by making it easier to download unofficial/provisional parts within the LDview tool or even as part of the installer.

It's likely this has been discussed in the past, but maybe with a track record of TLG giving us good quality meshes we should rethink the strategy going forward?

---Tom

[Cam's Bricks]

In all seriousness, couldn't the LDraw project automate that process and add thousands of (provisional) parts to the database overnight?  It's really remarkable that The LEGO Group is giving us the meshes for almost every new part!

Automating that process could benefit the LDraw community greatly, allowing the LDraw community to use it's limited resources to focus on fixing bugs and creating 3d models for vintage parts.  It might require evolving the 'good enough' criteria for inclusion in the parts database, or introducing a middle tier of 'provisional' parts that are 'unofficial' but installed automatically as part of app installer, or by making it easier to download unofficial/provisional parts within the LDview tool or even as part of the installer.

It's likely this has been discussed in the past, but maybe with a track record of TLG giving us good quality meshes we should rethink the strategy going forward?

---Tom

I am not a part author officially but a big part of authoring parts is making them fit into the system of Ldraw. Like are there similat parts? How are they rotated? How are the scaled? Does this connect well with others? Some of that can be automated but most of it cannot be. Plus there are sub files and areas made for decals and other decor that have to be manually split from the mains files and organized. 

Most of the "easy" parts have been done just leaving the heavy parts new and old.

[Orion Pobursky]

I'd like to echo Cam's sentiment.

While exporting raw files is easy, prepping for the library is non-trivial. I don't mind having raw files posted here on the forums, but I'm not a fan of some sort of "official" channel for unprocessed files, especially on the library site. That said, digital-bricks serves as a decent repository for raw files and some enterprising user could probably set up a site for Builder app exports.

[Gerald Lasser]

The only pieces fitting "well" (without touching them) are hair pieces, their origin is fine, however they will have intersections with the head geometry.

Most other parts don't share the origin where the LDraw origin would be, and as they are exported right now, are missing plain studs and understuds, those need to be added manually, respectively a method needs to be derieved to determine where they are.

I share Orion's view of a DB like Website as repositoryto get the raw parts.

Another interesting thing, also rare now, but I observe it more often, is the number of surfaces is increasing. An extreme example is the Lunar rover tyre, which had 210k lines, coming to serveral MB in size.

So, while a mass conversion into a repository is possible, the results need to be curated in terms of orientation and origin, before the files are disseminated IMHO.

There was a case recently where Stud.io used a "raw" part (28192) for a long and then took over the LDraw part with the usual origins etc. This angered a lot of people... (mildly spoken)

[Jeff Jones]

The only pieces fitting "well" (without touching them) are hair pieces, their origin is fine, however they will have intersections with the head geometry.

Most other parts don't share the origin where the LDraw origin would be, and as they are exported right now, are missing plain studs and understuds, those need to be added manually, respectively a method needs to be derieved to determine where they are.

I share Orion's view of a DB like Website as repositoryto get the raw parts.

Another interesting thing, also rare now, but I observe it more often, is the number of surfaces is increasing. An extreme example is the Lunar rover tyre, which had 210k lines, coming to serveral MB in size.

So, while a mass conversion into a repository is possible, the results need to be curated in terms of orientation and origin, before the files are disseminated IMHO.

There was a case recently where Stud.io used a "raw" part (28192) for a long and then took over the LDraw part with the usual origins etc. This angered a lot of people... (mildly spoken)

i once opened 14419 from bi in assetstudio in the asset hierarchy i saw references to studs and antistuds that had the same names as the unity microgame version
i unfortunately didnt dig deep enough to find the location and rotation of it
here is the file if you want to find it

  BrickGeometryAsset_14419_F.zip (Size: 118.39 KB / Downloads: 1)

[tom alphin]

This is a very interesting discussion, and I want to thank everyone for sharing their insights.  It does seem like the key concerns that impact a lot of parts are the issues of origin (I think this means the location of 0,0,0 center position) and the issue of orientation.  

I wouldn't be surprised if a predictable transformation gives a more acceptable orientation and origin, but I do now understand why this is a bit more manual than I would have hoped. Perhaps a simple online tool where we could crowdsource the right origin and orientation might reduce friction for full integration into the collection at a later date?

---Tom

[Chris Böhnke]

This is a very interesting discussion, and I want to thank everyone for sharing their insights.  It does seem like the key concerns that impact a lot of parts are the issues of origin (I think this means the location of 0,0,0 center position) and the issue of orientation.  

I wouldn't be surprised if a predictable transformation gives a more acceptable orientation and origin, but I do now understand why this is a bit more manual than I would have hoped.  Perhaps a simple online tool where we could crowdsource the right origin and orientation might reduce friction for full integration into the collection at a later date?

---Tom

Whilst this should "mostly" not be an issue, I think orientation is more complicated than that. Orientation in LDraw is typically based on the concept of "part category" which can in some cases create a bit of disconsens.

Lego themselves don't help much here, as the category system they publish in things like LDD is an utter mess. Example?

Part 4266 ("Technic Hub Ø30.4 x 20") is listed as "Brick, Round" in LDD, despite clearly being a wheel.
Why?
4266's matching tyre got discontinued around 1994, however the hub was kept in use - mainly as motor housing for propeller planes up until 2004. LDDs database was likely made after that date, so they listed it as a round part (though it's not a brick either by any definition...)

Part 6797 (Technic Engine Air Scoop  4 x  4 x  1.333) is actually a Scala part, where it was used as a bin with an entirely different orientation.
Why?
Only few AFOLs around 1997-2003 were aware of the exact contents of Scala sets due to lack of interest. So when the part surprisingly appeared as a very visible decoration in a larger Technic set in 2001, many people assumed it was made for this Technic set.

Part catgories are often a very subjective thing and orientation is to a degree linked to it, just keep that in mind.

[tom alphin]

Whilst this should "mostly" not be an issue, I think orientation is more complicated than that. Orientation in LDraw is typically based on the concept of "part category" which can in some cases create a bit of disconsens. 

...

Part catgories are often a very subjective thing and orientation is to a degree linked to it, just keep that in mind.

I am very familiar with the issue of orientation, since I have developed my own standards for the best orientation to use when rendering the high-contrast label images.  For example, I prefer minifigures to face to the right (a 90° counterclockwise rotation from the LDraw default orientation).

Inverted slopes are also tricky, since there is often only one orientation where you can see all of the relevant facets of the part.


I think an automated tool would guess parse the part shape and size, guess on the best orientation and origin, and it would be trivial within the automated tool to specify the correct offset and orientation.  A simple web app would probably be best, with a small number of users who have read/write access to change the offset values.  In rare cases where rotation and offset doesn't result in a viable part, just make it easy to flag the part for manual editing and block automated output.