Wednesday 7 October 2020

How NOT to draft a women's sloper: Why the FreeSewing.org Breanna sloper for women does not work at all (unfortunately)

In the loquatious manner of old books, the full title of this blog post continues:

Containing therein an explanation of the fundamental importance of the difference between dart width and dart angle.
Or
What shenanigans Marmota gets into when she has too much free time on her hands and wants a quick solution to something (Spoiler alert: It was anything but quick.)

I have to say right ahead that this was written back in May so some of the info about how the FreeSewing.org site is set up here is dated. The problems with the pattern remain exactly the same, though.


I also have to say right ahead that the sum total of my knowledge of programming is a neat zero, so I have no idea how the actual code works and cannot unfortunately fix it.

I do, however, by now know a thing or two about drafting sewing patterns, and about measuring and altering them to make sure they fit - seeing as I'm exactly the sort of non-average person FreeSewing.org is targetting:


In their system, back in May, the closer you were to a standard size, the more your diagram looked like a perfect circle. Mine was a perfect splotch.


So I can at least offer an explanation of why it doesn't work. Why the way they take those detailed measurements (that for the most part do make sense to me) and turn them into this pattern... results in a pattern that does not work at all. Long story short: the very base of the women's sloper is flawed and does not seem to take into account some fundamental truths of sewing and pattern-making.

Namely how you deal with darts and angles.

I hope that this post will both work as a (very detailed) review for other existing / potential users to see exactly why that particular covetable pattern (custom slopers are covetable, right?) doesn't work right now, as well as hopefully help the programmers do things better in the future when it comes to women's patterns. Because I do love the idea of OpenSource patterns!

(And maybe it will also help explain some things about drafting your own slopers. I learned a lot from this myself.)


It all started for me with a desire to get some quick standard-size slopers to compare a historical pattern to - namely, I started re-evaluating the kacabajka pattern because I know for a fact my old grading method is Bad, so I wanted to start again by seeing where the original historical pattern lay in terms of measurements and proportions to be able to average it out for a basis for new grading.

One can obviously expect some faults in patterns drafted for the aforementioned non-average people, but surely standard sizes should be standard?



Nope.

That was me downloading all the standard sizes and nesting them together in Inkscape demostratively, once I realised things were looking off.

A) That's definitely not how grading works and what nested standard-size same-bust-cup-size patterns normally look like. (It would help immensely if they made their standard size measurement charts available...)

B) There's definitely something wrong going on with the angles at the side seam. As well as the shape of the armhole. And probably also the length of the back darts.

Only the L and XL (I think those were the sizes there at the outside edge of my nest) do somewhat resemble what you would expect a standard women's sloper to look like. The rest is decidedly off.

So clearly there's something off in the algorhythm. Code. Whatever. And it sent me off on this whole completely different and big tangent than I expected...

Okay, but maybe that's just because it's a different system? Maybe those are just unusual-shaped patterns that will still fit?

Nope.

Let's look at what FreeSewing.org should be good for: Let's look at the sloper customised for me.


That's even worse than I expected. And because I'm well aware of the sort of measurement comparisons to and adjustments from standard sizes that I normally have to do, this made for an excellent exercise in "spot the problems".

Here below is a comparison with what my basic sloper does in reality look like on the right. (Well, strictly speaking, it's one I have not tried out in fabric yet, but I did measure it thoroughly and adjusted it accordingly so at the very least it's a far better start than the FreeSewing one on the left. It's adjusted from another free standard size pattern I found online, which is identical to my basic bust and waist measurements in the one size they do offer for free, so it formed a neat basis for my spate of Inkscape messing around - starting with changing it from a B cup to my DD cup - I suspect it's still a faulty method, but it was a start.)


Soo... do we play "spot the differences" now? I didn't count them. :D In almost every aspect, the FreeSewing sloper is just... wrong.

The thing that was immediately obvious to me (even without the comparison, which I eventually made a lot later in the process) was that the armhole on the Breanna was way too low (especially in the back) and the side seam way too short for me. I wondered if perhaps the whole thing was too short (and wondered how that would have happened with all those detailed measurements), so I made use of the wonderful measuring tool in Inkscape and measured and measured.

The center front length is more or less OK. The shoulder slope angle is OK (that's nice). But that's it. From there all the way to the right of the pattern the problems get only worse and worse.

So as I measured, I began to see where the problem was.

* * *

There was a sidestep here decumenting my thinking process, but, really, you don't need to know all the nitty gritty details. What you need to know is that the sloper pattern is derived from a base pattern, which is just a front piece, with one dart.

* * *

So I looked at the base, one-dart-only front pattern you can get if you go into the advanced pattern settings down below on the site, and all the problems became more obvious.

Let's highlight all my points of comparison somewhat so you know exactly what I'm talking about.


To begin with, I think the programmer (in this case, it appears to be the founder of the site, Joost de Cock) does not know enough about the significant role all this plays in pattern drafting: A) what angles at seams look like when you sew them up, B) what angles placed at the fold look like when you open them up, and C) how darts actually alter the shape of things and how dart rotation actually works...

Because, let's say I rotate that waist dart to get a side bust dart instead...


That's not what you want your waistline to look like. Even though you're nominally keeping the intended length of the waistline there, by introducing an angle significantly different from 180 degrees you're actually considerably shortening the arch length this waistline covers...

More than anything, this looks like my old patterns for Barbie, and it's common knowledge that Barbie does not look like a normal human being. (And then I did compare it to a pattern for Barbie, see below, and it's worse because that pattern for Barbie was hand-drafted by someone who knows from drafting.)

Besides, no one (who isn't going for a historical impression) wants their basic bodice pattern to end in a point at the centre front. The centre front of the waist seam should be at a right angle to the centre front seam / fold, period.


To get back to my comparison lines, please also look at the apex-to-side-seam line, which I assume to be roughly the bustline. (Ignore the two bustlines, please.) Now, what exactly this says took quite a lot of figuring out for me as well, so let's look at hard numbers.

Hard numbers are: my apex to waistline measurement is 19 or 20 cm. My side seam measurement (in a tightly fitting garment) tends to be 22 cm. (I may have a low-sitting bust.)

So while towards the centre of the sloper, as far as dart length, the measurements are about right, towards the right of the pattern it all falls apart.

The line from the bust to the waistline (AKA the dart legs) is longer than the side seam. And we've just established, in hard numbers, that on the real me it's the other way round.

There's no way the horizontal lines would really sit horizontally and the vertical lines would really sit vertically on my body. The angles in that tetragon between the dart and the side seam are all wrong. And so... while you're keeping the given measurements in one direction, you're considerably shortening others in the other direction.


When I measure the pattern in this direction, underarm to base of the dart, and compare it to the same measurement on myself...

... my actual measurement is twice as big. (Not exaggerating at all! Twice as big! 36 cm.)

You will see what this does in 3D later...


Still, with my stated apex-to-apex measurement of 20 cm and HPS-to-bust measurement of 31 cm, the placement of the bust point is about right. Even the bust point to waist seam length is still about right in the base - at least along the legs of the dart. So if you ignored the big dart for now and did this in a more rectangular-grid manner like Bilikis does and like other patterns on the site do (the vertical red line at the side in my red-marked picture above kind of illustrates that rectangular grid), some of the theory behind all this might be fairly sound. At a cursory glance it looks sounder than Bilikis' nearly symmetrical back and front... (I've been referring back to Bilikis a lot lately...)

Unfortunately, just like you want your waist seam at the centre front to be horizontal, you probably also want your vertical waist dart to be indeed more or less vertical, which is where Bilikis gets it right and this draft begins to get it all wrong.

Theoretically, without taking into account the waist seam and side seam angles, the garment would look something like this:


Theoretically.

That dart slant may be about right for historical clothes (mental note to self...), but A) that's not what we're after here; B) even if we were, the rest of it is still wrong.

So in 3D practice, nothing sits where it should.

(And yes, I know this is just the base and not the finished pattern, but if you get the base wrong it will all go downhill from there, so none of this is improved in the finished pattern!)

* * *


With printer, paper, scissors, Scotch tape, and a handy knitting needle, I could neatly demonstrate how bad this gets in 3D.

And when I say "bad"...

... I mean "it barely fits together".


This is my "actual" sloper:


This is the Breanna base, with back from the sloper since the base has none:


This is the full-blown Breanna sloper:


Notice, among the things not yet fully pointed out, how the deep back armhole utterly distorts the shape of the "garment".

Just for good measure, all of them from the front, with bonus Barbie pattern at the left:



* * *


I assume this draft is doing some sort of widening of the dart and bodice from the upper left origin point through bust point depending on bust-to-waist ratio, and therefore also rotating the direction of the side seam accordingly:


The problem is, of course, that this way I end up with that ridiculous centre front point, an increasing disappearance / distortion of my actual bust-to-waist length, a dart that is exceedingly big, and a side seam that is angled out despite the fact my waist measurement is 20 cm smaller than my bust measurement...

What you would really want to be doing is basically treating the side seam as another dart. Which is what the rectangular construction does.

 

* * *
 
So yes, as a direct result of all that, in the Breanna sloper I end up with a side seam that is a whopping 7 cm shorter than it should be:


And that's clearly a problem that, with the coding, carries over into the back pattern piece and gives you the extremely improbable back sloper shapes I ended up with. (With a waist dart so short it doesn't go anywhere near my shoulder blade level!)

* * *

Now obviously the theory behind the draft is that some of that huge dart space in the base pattern will be rotated into the secondary bust dart. Except that when your base pattern is already flawed, that won't fix it - and the coding of that situation does, in fact, only exacerbate it because the way it treats the rotating of darts is flawed (I'll get to that). So the whole waist seam / side seam angle situation becomes even worse. Moreover, the weird angled vertical waist dart shape is also still very much present...



By now it should be clear why the Breanna angles are all wrong. And you should see how there's a good chunk of side seam missing.

Now look at this comparison between the base pattern and the sloper...


Why is the waist dart even more slanted in the basic vertical position? Why is not even its axis vertical? That rather defies my powers of explanation, but when I project the axis further, I begin to suspect it's because somehow in this position they're assuming an axis that runs to the centre of the shoulder. Because princess seams I guess? Except that's not how darts and princess seams work in 3D, as we've seen in my paper models.

If we want to know why it gets even shorter in center front, it may help to look at what happens when we move the waist dart just a step from the "06:00" position to the "17:00" position:



But wait, it gets even better at "16:00"!


This is all somehow a problem with the code for rotating darts.


* * *

Speaking of rotating darts, then.

You know what's the funny thing about rotating darts?

It's rotating. You're not moving and redistributing the distance between the ends of the dart legs. You're moving and redistributing the angle. Angle as in "degrees of a circular section". That means that:
A) depending on where you move your darts, you may end up with significantly different dart width / length measurements, and still have a pattern that fits the same way the original did, because you're maintaining the way it conforms to the angles of the body underneath it;
B) you're not supposed to change the shape of the outline of the pattern! You're only dividing it into parts and rotating them.

And obviously that's not what's happening in the FreeSewing code. To sum up what I've observed (though I may be wrong): When redistributing darts into more darts, the code treats them as distance, which is wrong (A). When rotating them in the clock positions, it does seem to treat them as angles (I think?), but it fails to maintain the outline of the pattern piece (it messes with it even in the first case), which is also wrong (B).

Let's look at this picture again:

You may notice that my green (actual sloper) side seam is taking away the same amount of waist width by which the green waist dart is smaller than the black (Breanna) waist dart.

If you're redistributing darts in a parallel fashion along a line, you can treat them like that (although in that case you have to watch where their points are pointing - you don't want to introduce an "apex" where it should not be). This is basically how you e.g. divide a single waist dart into multiple waist darts to better conform to a particularly curvy body.

However, once you start rotating around a point, things leave the area of distance and enter angle zone.

Come to think of it, that may become even more obvious when I look at another pattern from the FreeSewing collection:


This is just a quick PrintScreen of the Cathrin corset (not my measurements, I was just messing around), with the base at the left and the actual pattern panels on the right. The Cathrin is an underbust, so if you would be dealing with the bustline you would be tapering the "darts" to a different point and different angle - but it should serve to illustrate the principle neatly. The waistline here remains vertical all throughout. So if you rotated the "darts" in the front, you would end up with a side seam angled in direct opposition to what the Breanna does for me and the smaller "standard" sizes.

... you know what, that's easier to show than describe.

Front is on the left, back is on the right.



Imagine a side dart above where the side front piece dips down, and you will probably begin to see how you could build a whole bodice block above this...

So yes, it is probably a lot more accurate to draft the basic women's block on top of a rectangular base anchored at center-front-waist and center-back waist somewhat like this (just with some ease instead of negative ease), compared to what the site is currently doing... and guess what, that's where you normally draft your starting lines when drafting women's slopers.


* * *

... with all this theory I came up with for how the sloper code maybe works, I still have no idea why, when toggling for a bigger / smaller underarm dart, both the darts become bigger / smaller.

That's wrong on so many levels I'm not even sure where to begin.

*headdesk*

* * *

The worst thing about all that is that it is basically just Part 1: What Happens to a Longer-Torsoed Larger-Busted Otherwise-Pretty-Rectangular Lady. (I'm not rectangular below the waist, but that doesn't enter into the Breanna pattern.) When it comes to drafting patterns for myself in a more or less rectangular manner, I'm actually pretty straightforward.

What happened when I entered measurements for a decidedly short-torsoed, narrower-waisted sway-backy lady but with a non-negligible belly protrusion, and with wide bust span ( = the rough measurements I derived from Ludmila Tomková's kacabajka) was that I kept getting an error message... Now obviously those were speculated measurements, and I was lacking her actual neck circumference, shoulder width and measurements below the waist. But I did make some educated guesses based on my own measurements and their standard size estimates so by all rights all the measurements should at least roughly resemble a person who did, at one point, really exist...

... (I was short on other women to test it at in lockdown, okay?) ...

... I did even lengthen her vertical measurements to match the estimates better while maintaining the rest of the proportions, and lowered her bust span, and still got an error message...



... which was a pretty good indication that there was something rotten with the base code for women to begin with because aside from her very short torso, general historical smallness and 1840s-corset-influenced bust span, I think her proportions actually are a fairly common shape for adult women. She seems to have a pronounced belly compared to a small bust (in terms of cup size, not overall measurement), but that is, say, something I can see on my own mother...

I messed around with her measurements more and I widened her neck and changed her shoulder slope, which however of course results in a wrong shoulder slope for her. Only then did I finally get a pattern. Where the base pattern for the front once again demonstrates neatly how this manner of drafting the base indeed does not work:



You may wonder why I said that shoulder slope is wrong when it seems to match nicely. Well, that's because on the original kacabajka pattern, the front shoulder slope angle is significantly smaller than the back shoulder slope angle -  what we're looking at around the shoulders is an 1840s pattern with the shoulder seam pushed backwards, not a contemporary one where you want it sitting directly on the shoulder.

But because the Breanna drafts the front and back shoulder angle more or less identical (or maybe even the other way round from the kacabajka), you'd actually want far more of an angle there than there is. What you see here is the result of my lengthening the shoulder slope measurement (shoulder point to center back high hip, at the time of writing) to be slightly closer to what they estimate for a standard size. But of course that means this pattern reflects Ms Tomková's real shape even less.

Despite the fact that I estimate Ms Tomková's shoulder slope angle was more or less identical to mine or my sister's (decidedly existing, real, non-average people), it did not like her with it. Somehow, it seems that big difference between a sway back-y back and a bigger-belly overbust front length measurement was exactly what the code did not like and why it kept spitting out an error message.

Yet those are all features I see on a lot of decidedly existing, real ladies in the Curvy Sewing Community.

So it definitely needs to be reworked entirely.

* * *

Final notes to the developers of the FreeSewing.org site: As I said, I have no idea how the code works so I have no idea how exactly to deal with all this on a programming level. But you may want to:
A) Somehow rework the code to maintain the outline of the basic pattern when rotating darts.
B) Include the side seam measurement in your set of measurements somehow. It's a very helpful measurement to be aware of and stick to when you're drafting for female curves. A similar sort of thing could probably be covered by a HPS-to-bottom-of-shoulder-blade measurement (that's what the back dart is there for), because that's obviously also the level where the Breanna back piece fails miserably right now.
C) Considering Ms Tomková and the many women (or even men, come to think of it) with belly protrusions and / or sway backs, including the natural waist front (and front hips, and front "seat") measurement (i.e. vertical measurements only in the front) would also be very helpful for capturing the real shape of the body.
D) Speaking of curves: A thing I noticed in patterns other than the Breanna. I do carry more of my hip mass in the bones on the sides so the rectangular style of drafting that area would probably work for me just fine for the most part; but probably the absolute majority of curvy women tends to carry it in other types of tissue in other areas than the side seam, so in your rectangular-drafted patterns for women, especially the more detailed ones, you may generally want to look into ways to redistribute the width differential more evenly, like some sort of toggling dart width and placement that would change the side seam angle maybe? Something a bit more precise than just your current contour / waist ease settings somehow... (This is one area where you can absolutely do that through width!)
E) While you are at it, could you maybe please include all those pretty armscye and sleeve cap adjustment functions with the Breanna? I know for a fact certain styles of lower but more curved sleeve caps set into higher armscyes fit me much better than the super-basic eased tall narrow one. :-)


Final note to regular readers: This was written in mid-May of 2020 (with revisions for readability in October of that year), and the Breanna was only introduced in February 2020. I'm holding out hopes that if you read this at a much later date, things will have changed... :-)

3 comments:

  1. Dear Hana,

    I happened upon this snarky evisceration of the Breanna block and got curious about why somebody would go through all the trouble of writing such a lengthy review.

    I thought perhaps salty takedowns was your jam, and your blog some sort of altar on which you'd ritually sacrifice the work of other people, to loud cheers of your followers.

    I must admit I felt a pang of disappointment when I perused the front page and noticed this doesn't seem to be the case. But that only deepened my intrigue.

    I am now cautiously leaning towards the theory that you genuinely care about good fitting clothes, and that you are merely trying to help. It's perhaps a bit of a naive interpretation on today's internet, but you do explicitly address 'the developers' which for all intends and purposes is me.

    So, allow me to start by saying that you are not wrong. Breanna has serious shortcomings. I myself am painfully aware of it, and I have toyed with the idea of retracting the pattern altogether. But there's a subset of people who get good results with it, and it seems to be popular with people making doll clothes too.

    I have since developed another block, called Bella. It's also not perfect, although I think it's a bit more robust. If nothing else, it's get less flexibility which makes it simpler to develop.

    To be frank — and I doubt you'd take offense to that after reading the above — I find all of this rather frustrating.
    I don't pretend to know how to draft for women. But I do get questions all the time about adding more womenswear to the freesewing.org, and that starts with a well-designed block.

    I have long waited for somebody to step forward to take this on themselves. Alas, I've been waiting in vain.
    Since that the fastest way to get a question answered on the internet is to answer it yourself incorrectly, I set out to create these blocks.

    When I see a post like this, it breaks my heart a little because you clearly put a lot of thought/work into this. And I can't help but wonder why you didn't you reach out to us/me and help us improve things?

    Then again, it's not too late. So I hereby formally invite you to work together to design the womenswear block that you think/know to be the best one out there.
    I gladly offer my services to code/implement your design/vision, so that the entire sewing community can benefit from it.

    You can reach me at joost at freesewing.org, we also have an active community on Discord: https://discord.freesewing.org/

    I hope to hear from you soon.

    love
    joost

    ReplyDelete
    Replies
    1. I tried to reach you at the time, before I wrote this, and could not find a simple way to do it, so that was actually one of the reasons this post happened in the first place. That, and long lockdown - I think if I was writing it today, I would hold back on the snark a bit. And unfortunately snarkiness seems to be a bit of a default for me in writing this sort of thing when it isn't necessarily the tone that would come across in speaking...
      I'm not on Discord, and entering new online waters is a bit of a terrifying experience I have not had much mental space for since you switched from whatever hard to enter space you were using before that. But I'll certainly be happy to help whichever way I can when I finally find the mental space for it.

      Delete
    2. P.S. So, basically, sorry for the snark and thanks for reaching out anyway!

      Delete