Sticker boxes

[Orion Pobursky]

We decided to standardize the sticker boxes here:
https://forums.ldraw.org/thread-27388.html
https://forums.ldraw.org/thread-27402.html

We never followed though with sticker box naming.

Let's discuss how this should be accomplished.

Note: Release of all new stickers is suspended until we come to a decision.

[Travis Cobbs]

We decided to standardize the sticker boxes here:
https://forums.ldraw.org/thread-27388.html
https://forums.ldraw.org/thread-27402.html

We never followed though with sticker box naming.

Let's discuss how this should be accomplished.

Note: Release of all new stickers is suspended until we come to a decision.

I'm not very comfortable with proposing and debating naming conventions. I'll can provide my opinion about suggestions, but I don't feel I should be making the proposals.

[Orion Pobursky]

I'm not very comfortable with proposing and debating naming conventions. I'll can provide my opinion about suggestions, but I don't feel I should be making the proposals.

Opinions are what I want. This is an open discussion thread.

[Magnus Forsberg]

I don't think we should see backsides as subfiles to the sticker. I see them as subfiles to the brick, which they are formed after.

When I make a formed sticker I know which brick it should be placed on. Searching after part subfiles, and finding the corresponding sticker backsides would make finding them easy.

To me, sticker backsides are subfiles, not primitives.
I see no reason to include a description in the filename. It will soon be too complicated, and might need a lot of exemptions.

Let's keep it simple, as suggested back here.
This "practice" is already in use, in a number of stickers made by Philo, Jens Brühl and Evert-Jan Boer. Stickers made for some Technic Panels.
64394
13710
87080
24119
11947

The only thing we need to figure out is how to handle differently shaped/formed/sized/positioned stickers.

[Orion Pobursky]

I like that suggestion for formed stickers.

What about flat stickers?

[Orion Pobursky]

For flat stickers I think:
sticker-back-1-9-x-1-9 is bit excessive but I can live with it.
add an optional qualifier like "r" for rounded corners

Formed I like:
<part number>bNN

Should we have a new folder in p?
Something like p/sticker
If so we'll need a new !LDRAW_ORG type as well. Something like Sticker_Back or Sticker_Box

Note, not all sticker backs need to be subfiled, only the one's that get a lot of reuse.

[Orion Pobursky]

I'd like to reiterate:
No new stickers will be released until we have a standard

I will write and implement a standard myself if I have to but I'd like to collaborate on a community driving solution.

[Magnus Forsberg]

For flat stickers I think:
sticker-back-1-9-x-1-9 is bit excessive but I can live with it.
add an optional qualifier like "r" for rounded corners

Should we have a new folder in p?

Why add all the hyphens?  Simplify it to: stickerback19x19r2.dat

stickerback[ZZ]x[XX](optional qualifier)

ZZ and XX is the dimension in studs, one decimal, but without the dot (can't have any dots/points in the filename). Same digits as the sticker.
(r2 = all corners of the sticker has the radius 2 ldu) Do we really need to define the radius of the cornes? Maybe they all have the same radius? Are they scalable?

Are there something else that need to be defined, in the optional qualifier?  
Don't we already have the "Minifig Torso sticker backside" in the library?  Is that a subfile?
Sticker Minifig Shield ...
Sticker Minifig Vest ...
Sticker for Minifig Torso ...

As I said, IMHO the backsides are subfiles. 
If we name them all "stickerback......dat" we can keep them in the s-folder, and they will be easy to find.

[Orion Pobursky]

Hyphen are there because what's the difference between 1.1 and 11 or 0.9 and 9 without a separator and the separator can't be "." or ",". 

I'm fine with the s folder and I'm fine with "stickerback".

[Travis Cobbs]

Hyphen are there because what's the difference between 1.1 and 11 or 0.9 and 9 without a separator and the separator can't be "." or ",". 

I'm fine with the s folder and I'm fine with "stickerback".

What about?

stickerback11-9x1-9

The hyphens are then only used as a decimal replacement.

[Magnus Forsberg]

Hyphen are there because what's the difference between 1.1 and 11 or 0.9 and 9 without a separator and the separator can't be "." or ",". 

I'm fine with the s folder and I'm fine with "stickerback".

What if we decided to always include 1 decimal?
11 would be written 110
0.9 would b 09
9 would be 90
90 would be 900

[Rene Rechthaler]

what about "stickerbackWWWxHHH"?
WWW (width) and HHH (heigth) always 3 decimals, the last one the tenths, fill with leading/following zeros -> 0.1 (001) 1.0 (010) 10 (100)
-> would improve sorting
for "special" sticker backs for specialized parts (technic panels) use the part number, not the size
for example s\6155286bs01.dat -> stickerback15458.dat (used exclusively on this part)

should the parts on the tracker now stay or get edited? unofficial or official ones too?

[Orion Pobursky]

Once we decide on a standard then we can decide how we will implement. 

I like Rene's suggestions a lot. 

I was going to propose something similar for irregular stickers. 

What about circular?

Do we want to stick to an "r" at the end for rounded corners?

[Chris Böhnke]

I am with that, Rene's suggestion sounds good. Adding an "r" for radius as well.

Regarding the previous question, the tiny radius found on primarily newer (>2012) stickers seems to be identical (or at least we can assume them to be equal. Scalable would be nice, but I don't know how that could be done (probably it's not even necessary?).

Some very large radii (>5 LDU?) is something I would consider a deliberate design element of an "irregular" shape.

Round (perfect circle) stickers would be for the matching 2x2, 3x3 etc. tiles only?

"stickerbackround-X" with X for the radius?

Round stickers from 1970-2010 (?) have a noticably larger radius (or smaller offset) than the modern ones. Those should also only exist as 2x2 round tiles, so simply add a suffix "large"/"L" or something?

[Orion Pobursky]

Ok here's a draft:

Sticker backing boxes

All stickers will have a backing box (the sides and bottom surface of the sticker) as a separate subfile located in the parts/s directory

Sticker Naming

Flat Stickers

Quadrilateral:
stickerback<WWW>x<HHH><optional qualifier>.dat

Where:
<WWW> and <HHH> are a 3 digit numbers representing the width (for WWW) and height (for HHH) of the sticker. The value of this
number is LDU/20 rounded to the nearest tenth. A leading zeros will be added for a number less than 10. Example: if the width of the
sticker is 143 ldu, WWW is 072.
<optional qualifier>
 - "r" for modern stickers with 1 LDU rounded corners

Circular:
<to be decided>

Irregular shape:
stickerback<short description|part number>.dat

Where:
<short description|part number>
If the sticker is intended to fit a specific part's shape, use that part's part number. Otherwise, use a short description (e.g. Star)

Formed Stickers

<part number>bNNNN.dat

Where
<part number> is the number of the part to which the sticker is formed
NNNN is a 4 digit number with leading zeros starting at 0001

[Philippe Hurbain]

Are quadrilateral backs really useful, as a single edgeless box does the job???

---

[Magnus Forsberg]

Are quadrilateral backs really useful, as a single edgeless box does the job???

---

Should we rename the box5-12? And the other std sticker backside primitives?

[Chris Böhnke]

Are quadrilateral backs really useful, as a single edgeless box does the job???

---

I was thinking the same about the circular backs.

Is assuring a more uniform sticker size (for tiles, bricks and such) an argument here?

Typically stickers have a 1 LDU offset from the base part border...

Not entirely sure on this, just an idea

[Gerald Lasser]

I was thinking the same about the circular backs.

Is assuring a more uniform sticker size (for tiles, bricks and such) an argument here?

Typically stickers have a 1 LDU offset from the base part border...

Not entirely sure on this, just an idea

I do not see the big benefit for a real square back, for me it really start with the rounded corners.

I am definately also leaning towards a "standard" Stickersize, e.g. for the 2x2 Tile to have a sticker 1.8x1.8 and not 1.85, 1.9, whatever.

Also for the curved ones. we should here be a bit more strickter to always use the same size imho, so a sticker for the curved 1x2 slope has always the same size, I think this is also what we can see from the stickersheets. This should make the design of the backsides easier

[Orion Pobursky]

I do not see the big benefit for a real square back, for me it really start with the rounded corners.

I am definately also leaning towards a "standard" Stickersize, e.g. for the 2x2 Tile to have a sticker 1.8x1.8 and not 1.85, 1.9, whatever.

Also for the curved ones. we should here be a bit more strickter to always use the same size imho, so a sticker for the curved 1x2 slope has always the same size, I think this is also what we can see from the stickersheets. This should make the design of the backsides easier

I agree with all of this. I do think, for consistency and size standardization, that we should have square cornered subfiles, even if it's just a reference to box5-12. I addition, I chose quadrilateral instead of rectangular to lump in the handful of trapezoid, parallelogram, and rhombus shaped stickers.

[Jeff Jones]

this may be already answered
but for parts with a couple of stickers (ex 98604, 44791)
should they get prims or just subfiles. 
what about parts with one sticker (ex 64343, 53394)
should they be inlined or subfiled for the left/right versions?

[Orion Pobursky]

New Draft (added circle):

Sticker backing boxes

All stickers will have a backing box (the sides and bottom surface of the sticker) as a separate subfile located in the /parts/s directory

Sticker Naming

Flat Stickers

Quadrilateral:
stickerback<WWW>x<HHH><optional qualifier>.dat

Where:
<WWW> and <HHH> are a 3 digit numbers representing the width (for WWW) and height (for HHH) of the sticker. The value of this
number is LDU/20 rounded to the nearest tenth. A leading zeros will be added for a number less than 10. Example: if the width of the
sticker is 143 ldu, WWW is 072.
<optional qualifier>
 - "r" for modern stickers with 1 LDU rounded corners

Circular:
stickerback<RRR>circle
Where:
<RRR> follows the same rules as above for WWW and HHH but for the radius of the circle.

Irregular shape:
stickerback<short description|part number>.dat

Where:
<short description|part number>
If the sticker is intended to fit a specific part's shape, use that part's part number. Otherwise, use a short description (e.g. Star)

Formed Stickers

<part number>bNNNN.dat

Where
<part number> is the number of the part to which the sticker is formed
NNNN is a 4 digit number with leading zeros starting at 0001

[Travis Cobbs]

Seems ok to me.

[Gerald Lasser]

one addition, to add readability, insert a "-" after "stickerback", like we have it for e.g. stugs


stickerback-019x019
stickerback-019x019r

[Chris Böhnke]

Seems ok to me.

I agree.

[Magnus Forsberg]

How about all the sticker primitives we already have? Some shapes without edgelines. Resizable.

1-4cylc3, 4-4cylc3 and 48/4-4cylc3
box3-12, box3u12, box4-12 and box5-12

box5-12 is used in ~1200 parts
The circular prims are used to create oval shaped stickers.

Many are used in regular parts and subparts, not in a sticker.

I think it would be wrong to not use/have/allow a resizable standard backside.

[Orion Pobursky]

I think it would be wrong to not use/have/allow a resizable standard backside.

I don't disagree with this sentiment but I think it will lead right back we are right now. 

I think standardization of sticker sizes is a Good Thing ™. Everyone's measurement methods will be slightly different. All stickers for a 2 x 2 Tile, 2 x 4 Brick, or Slope Brick Curved  2 x  2 x  0.667 should all be the same size. We could have a reference doc that lists what dimension to use for certain brick sizes, but I think this is a cumbersome solution. The easiest way to ensure this is to have a standard sized backing files and make the author explain if they choose to do things differently.

[Magnus Forsberg]

I think standardization of sticker sizes is a Good Thing ™. 

Standardized sizes and using a resizable primitive, are not contradictions.

[Orion Pobursky]

Standardized sizes and using a resizable primitive, are not contradictions.

Here's what I don't want, I measure 1.85, you measure 1.95, and someone else measure 1.9 for a sticker for 3068b. Now we have 3 different sizes for should be a consistent 1.9

Also, scaling doesn't work for rounded corner stickers. So we have a subpart for everything but circles and right-angle quads which isn't consistent either.

I'll concede that maybe I don't understand your point.

[Massimo Maso]

Hello,
for me, in general, the idea is good and tends to simplify the work considerably because you don't have to invent a new shape for the sticker every time, but I'm afraid there are still problems standardizing it too much.

The advantage is above all to reduce the volume of the library because there are a lot of duplicates. (Many of them I made myself like the back of the stickers for 15068 or 11477).



Some food for thought:

- If we want to standardize the dimensions and tie them to a part, we must leave ourselves the possibility of using the same sticker box on different parts. For example, the sticker box used on the top face of 2431b, is also used for the vertical face of 3010 because it does not cover it completely.

- Scaling the boxes as Orion rightly indicates is not possible once you use the connecting radii because they would be deformed.
For formed stickers there is the problem of fold lines that also prevent scaling in the direction orthogonal to the fold.

- Even if it is a bit off topic, I also recommend that viewers consider replacing primitives in faces on which a sticker is attached, such as 93606c.
The small thickness of 0.25 is in some cases not able to cover the greater roundness of the surface on which the sticker is attached and it pops out ruining the effect of the sticker.





Note: selfishly, for the work I have done so far, the standard I have kept is different from the one proposed: 2LDU distance from the edge and 1.5LDU radius for the edges. In my opinion, these dimensions work very well. Better than 1LDU border and 1LDU radius that I have seen proposed by Chris Böhnke.

[Orion Pobursky]

- If we want to standardize the dimensions and tie them to a part, we must leave ourselves the possibility of using the same sticker box on different parts. For example, the sticker box used on the top face of 2431b, is also used for the vertical face of 3010 because it does not cover it completely.

Reuse is fine and expected. What I'm trying to avoid is the creation of duplicates.

Note: selfishly, for the work I have done so far, the standard I have kept is different from the one proposed: 2LDU distance from the edge and 1.5LDU radius for the edges. In my opinion, these dimensions work very well. Better than 1LDU border and 1LDU radius that I have seen proposed by Chris Böhnke.

This is also fine and it's easy to "fix" the parts that don't conform.

[Massimo Maso]

Reuse is fine and expected. What I'm trying to avoid is the creation of duplicates.

In one of the first proposals it seemed that the stickerback should take the name of the part on which it was to be attached. if it's just an indication of size then we're good to go.

[Massimo Maso]

This is also fine and it's easy to "fix" the parts that don't conform.

For flat stickers I agree that it is a fairly simple operation.
Apart from the areas of rounded corners, everything is linear.

For formed stickers it is another story.
You have to review all the intersections with the folds: as long as they are straight lines nothing changes, but if there are arcs or cut figures it is to be reviewed from scratch.

Some stickers, fixed on different pieces, are aligned so that the decoration is continuous from one piece to another.
If you climb them, at least one check is to be given.




However, it is not that we all have to do them again immediately. When the time comes, we will do it a little at a time.

[Orion Pobursky]

Let's put this to bed:

- Do we want 1 or 1.5 LDU for the standard corner radius?
- Do we want 1 or 2 LDU for the standard distance from edge?
- Are we ok with having the non-rounded corner rectangular and circular/oval stickers use resizable prims instead of a stickerback file?

[Philippe Hurbain]

Let's put this to bed:

- Do we want 1 or 1.5 LDU for the standard corner radius?
- Do we want 1 or 2 LDU for the standard distance from edge?
- Are we ok with having the non-rounded corner rectangular and circular/oval stickers use resizable prims instead of a stickerback file?

2 / 1 / yes for me.

[Orion Pobursky]

I'll note that if we want 1 ldu rounded corner and 2 ldu border this is will cause Massimo, out most prolific sticker author, a lot of rework. I think this should be taken into consideration.

[Philippe Hurbain]

I'll note that if we want 1 ldu rounded corner and 2 ldu border this is will cause Massimo, out most prolific sticker author, a lot of rework. I think this should be taken into consideration.

Imho 1ldu corner is too small to be worth the trouble, barely visible so we could as well user a simple resizable box.

[Gerald Lasser]

I'll note that if we want 1 ldu rounded corner and 2 ldu border this is will cause Massimo, out most prolific sticker author, a lot of rework. I think this should be taken into consideration.

I understood Philos reply as 1,5 Radius, and 1 LDU from the edge of the part

[Gerald Lasser]

Let's put this to bed:

- Do we want 1 or 1.5 LDU for the standard corner radius?
- Do we want 1 or 2 LDU for the standard distance from edge?
- Are we ok with having the non-rounded corner rectangular and circular/oval stickers use resizable prims instead of a stickerback file?

- 1.5 LDU radius
- 2 from the edge of the part (many stickers are 1.8 x 1.8) 
- yes

[Philippe Hurbain]

(many stickers are 1.8 x 1.8) 

You have a point...

[Orion Pobursky]

Based on the discussion here, the release restriction on square cornered, non-standard corner rounded (> the default rounding of modern stickers), round, and irregular shaped sticker back is lifted.

[Magnus Forsberg]

Let's put this to bed:

- Do we want 1 or 1.5 LDU for the standard corner radius?
- Do we want 1 or 2 LDU for the standard distance from edge?
- Are we ok with having the non-rounded corner rectangular and circular/oval stickers use resizable prims instead of a stickerback file?

I want to avoid a massive rework of all the stickers.
A standard is perhaps good, but I think it is wrong to force the use of something that feels wrong to the author.
Corner radius and distance from edge may vary over time. Older sticker tend to be bigger, reaching all the way to the edge of the part.

All we really need to do is to define a way to number and describe the backsides that are reused often.
Some parts are often stickered, and need a common sticker backside. That's all.

We should choose a standard as close to the problem we are trying to solve.

[Orion Pobursky]

New Draft (reduced requirements to quad stickers with standard rounded corners and formed stickers):

Sticker backing boxes

The backing box is defined as the sides and bottom surface of the sticker

For most stickers, the surface of the sticker is intended to be applied to an entire face
of a single part. In this case, the size of the sticker should be that the edge of the sticker is 2 ldu
from the edge of the part face. If actual, measured sticker size is +/- 1 ldu from this standard then
the sticker should be modeled to the standard. Exceptions to this rule will be handled on a case by case basis.

Flat Stickers

Quadrilateral:

The backing boxes for older, sharp cornered stickers should be modeled
with an appropriately scaled box5-12.

Newer stickers have rounded instead of sharp cut corners. The radius of standard rounded corners will be 1.5 LDU.
All stickers with standard rounded corners shall use a subfile located in the parts/s/ directory with the following
naming convention:

stickerback<WWW>x<HHH>.dat

Where:
<WWW> and <HHH> are a 3 digit numbers representing the width (for WWW) and height (for HHH) of the sticker. The value of this
number is LDU/20 rounded to the nearest tenth. A leading zeros will be added for a number less than 10. Example: if the width of the
sticker is 143 ldu, WWW is 072.

Circular:

Circular and oval shaped stickers should use an appropriately scaled 4-4cylc3

Irregular shape:

Irregular shapes should use quads and triangles with conditional lines where appropriate.

Formed Stickers

<part number>bNN.dat

Where
<part number> is the number of the part to which the sticker is formed
NN is a 2 digit number with leading zeros starting at 01

[Orion Pobursky]

All we really need to do is to define a way to number and describe the backsides that are reused often.
Some parts are often stickered, and need a common sticker backside. That's all.

I'm not suggesting we go back and rework every non-conforming sticker.  I do think defining a standard for all new stickers is useful. If and when LEGO changes the way stickers are manufactured, we'll change the spec.

[Orion Pobursky]

Based on an off-forum suggestion, I tweaked the above draft for formed stickers to bNN vice bNNN

[Orion Pobursky]

Thank you all for the discussion.  

I will be creating a final draft and moving this discussion to the library admin team for final approval and discussion of implementation.

[Magnus Forsberg]

Thank you all for the discussion.  

I will be creating a final draft and moving this discussion to the library admin team for final approval and discussion of implementation.

I have some questions here that might need an answer:
https://library.ldraw.org/parts/46396

[Orion Pobursky]

Ok. Well. We haven't finalized the spec yet and it'll have to wait until I'm back from vacation.