CremonaCad: A new tool for designing violins

[Andrew Argraves]

For the past six months I have been working on an application to help in the design of violin family instruments. After many iterations, it's now at a robust enough point that I consider this early version ready for release! 

https://cremonacad.aargraves.com/

CremonaCad constructs the contours of violin (+gamba) instruments using a series of intersecting arc segments. This is distinct from some of the other applications or design philosophies which rely heavily on cartesian (xy) measurements. Breaking down the instrument into arcs is not only a more historically accurate approach to design (based on what we know of cremonese making), but it also feels quite intuitive to go through the various arc lengths and change the shape of the instrument. Once the "recipe" steps have been completed, CremonaCad can export template and mould patters to PDF or SVG.  

CremonaCad comes with several templates from historical instruments - violins, violas, cellos and basses. For new instruments, images can be uploaded and scaled to assist in making a copy.

 

This is still a very early phase of the project, so there are no doubt bugs and oddities in the code, and I am very much looking forward to any community feedback to make the application better. CremonaCad is free and open source, and you can find the code here : https://github.com/TheGravyBaby/cremona-cad

The next step as I see it is to refine the current recipe system, and then begin developing out features for arching templates. After that I hope to have a method for F-hole design within the year. 

I must also say that a huge amount of the research for this project came from many fantastic members of this community, most notable David Beard and Kevin Kelly. I hope CremonaCad can become a useful tool for making of instruments as well as developing further the research they have inspired. 

Edited June 5 by Andrew Argraves

[Will Turner]

Nice!    I've browsed through the interface and find that it's actually a great start!

Question:  When you select to export the PDF, is that a Tabloid (11x17) format?   It would be nice to indicate, or even allow the operator, select the page size for the PDF.

 

[Dr. Mark]

It was only a matter of time.  Next:  StradCad!  Given CremonaCad output and input wood properties, StradCad will produce ideal arching and thickness maps for plates and sides.  Progress cannot be stopped.

[Pluto Vermin]

Easy link to it:

https://cremonacad.aargraves.com

[Andrew Argraves]

On 6/5/2026 at 11:03 PM, Will Turner said:

Nice!    I've browsed through the interface and find that it's actually a great start!

Question:  When you select to export the PDF, is that a Tabloid (11x17) format?   It would be nice to indicate, or even allow the operator, select the page size for the PDF.

 

Thanks a bunch! Currently the PDF size is dynamically calculated to fit the instrument in question, but I don't reference standard sizes. It shouldn't be that hard to add a code block the finds the best standard size for a given instrument, so I can definitely work on adding that for the next release. 

Edited June 8 by Andrew Argraves

[Andrew Argraves]

One issue I have been trying to think through is the system around viol necks. The current method is a simplification that produces an effective approximation. I currently take the standard violin upper bout, which consists of two side arcs (vesica has become a term) and a top joining arc, and then graft an arc with opposite curvature to the join arc. 

All the user has to provide here is a radius for the upper arc (V0) and a width for the neck. Of course to get such simplicity there are tradeoffs. 

1. This presumes that V0 always intercepts the neck cutoff with a vertical tangent. This is certainly not the case on most instruments, though I figured the user could always just cut the neck off a bit earlier if they end up making the pattern. 

2. It seems many viols and basses have a straight edge connecting the neck arc to the upper bout vesica arc. I of course do not have a straight edge in this system, and always join with an arc.

==

Part of why this method was attractive is that it still allowed me to construct the upper bout using two arc lengths and then "tack on" the additional viol neck arc. In the standard violin style upper bout, given a height and a bout width, we actually only need the radii of the two arcs to define how they intersect. Its a very simple and easily comprehended model. 

 

To construct the viol upper bout using the straight edge method, I see a more complex parameter set. Let V0 be the circle used to define the neck. 

Params : 
V0.radius
V0.startAngle
V0.endAngle
UpperVesica.r
UpperBoutWidth
TotalInstrumentHeight

Below is a crappy model of that parameter set. The straight edge L length can be calculated given that we know the radius of the upper vesica as well as the width of the upper bout. Overall I find this a bit clunky but perhaps it is what must be done. 

Edited June 8 by Andrew Argraves

[Andrew Argraves]

I think I have found a happy medium with the viol neck implementation. The join will still remain an arc, but by allowing arc to have a center unbounded by the Y axis, many more options emerge. 

An arc with a very long radius approximates to the straight edge, where as a smaller radius works with instruments that have a more curved join. In my testing this seems to work well with many different historical instruments, even if it is indeed an approximation.

 

More importantly, I've added controls for purfling and the fluting channel platform. Creating the fluting channel was a bit tricky as it required joining the upper and lower bouts and the C bout with a Biarc formula as to bypass the corners. But now that this channel exists, we can move onto developing the plate arching, which intersects the fluting channel rather than the outer edge of the instrument. 

Unlike the contour of the violin, which can quite nicely be parameterized and modeled by a computer, the arching is far more organic and variable. With the prospect of full simulating "actual" arching aside, there seem to be three categories of curves used for "ideal" arching: catenary curves, curated cycloids, and splines. In theory the other two curves can be approximated well by splines, but I think offering explicit controls for each type of
curve would be best. 

The real question in regards to making arching templates is how much of the arching can practically be inferred. Given a single vertical "long arch" as well as a single horizontal bout arch, can the remaining bout arches be inferred or do they need to be defined uniquely? Inferring them would essentially involve skewing the bout arch to fit within the height constraints of the long arch as well as the width constraints of the fluting channel.

I'll leave the above question to the community, but of course I am always happy to talk to myself on this thread like a madman xD
                                                                                                                                   

Edited June 22 by Andrew Argraves

[Will Turner]

On 6/22/2026 at 8:28 AM, Andrew Argraves said:

I'll leave the above question to the community, but of course I am always happy to talk to myself on this thread like a madman xD
                                                                                                                                   

I appreciate your hard work.   This is pretty cool.  As for the arching, there are a lot of scans of instruments that might provide some reference for a potential formula?   But I do see variations from instrument to instrument by the same maker, so I don't have a specific recommendation.  

Just a cursory list of some research I had seen.

R.M. Mottola, “Comparison of Arching Profiles of Golden Age Cremonese Violins and Some Mathematically Generated Curves” (quantitative comparison of longitudinal vs transverse arching fit, discusses how well simple curves match real plates) https://www.savartjournal.org/articles/12/about.html

Charline Dequincey, “Recording arching profiles: Ahead of the curve” (includes a method presented at VSA Violin Makers in Oberlin, June 2018). https://www.thestrad.com/lutherie/recording-arching-profiles-ahead-of-the-curve/13284.article

I'm sure there's probably more that may help to determine optimal "inferred" arching to go with the defined. 

 

[Marty Kasprzyk]

On 6/7/2026 at 11:08 PM, Andrew Argraves said:

One issue I have been trying to think through is the system around viol necks. The current method is a simplification that produces an effective approximation. I currently take the standard violin upper bout, which consists of two side arcs (vesica has become a term) and a top joining arc, and then graft an arc with opposite curvature to the join arc. 

All the user has to provide here is a radius for the upper arc (V0) and a width for the neck. Of course to get such simplicity there are tradeoffs. 

1. This presumes that V0 always intercepts the neck cutoff with a vertical tangent. This is certainly not the case on most instruments, though I figured the user could always just cut the neck off a bit earlier if they end up making the pattern. 

2. It seems many viols and basses have a straight edge connecting the neck arc to the upper bout vesica arc. I of course do not have a straight edge in this system, and always join with an arc.

==

Part of why this method was attractive is that it still allowed me to construct the upper bout using two arc lengths and then "tack on" the additional viol neck arc. In the standard violin style upper bout, given a height and a bout width, we actually only need the radii of the two arcs to define how they intersect. Its a very simple and easily comprehended model. 

 

To construct the viol upper bout using the straight edge method, I see a more complex parameter set. Let V0 be the circle used to define the neck. 

Params : 
V0.radius
V0.startAngle
V0.endAngle
UpperVesica.r
UpperBoutWidth
TotalInstrumentHeight

Below is a crappy model of that parameter set. The straight edge L length can be calculated given that we know the radius of the upper vesica as well as the width of the upper bout. Overall I find this a bit clunky but perhaps it is what must be done. 

I used to think the graceful and natural looking shapes of old instruments were originally just the natural shapes of bent wood ribs.

But I found bent wood ribs were hard to photograph without shadows so I switched to using narrow strips of plastic.

Attached is an example of how a bent spline easily matched a Stradivari's sloped shoulder viola d'amore pattern.

[David Beard]

Very interesting and inevitable work.  Kudos.

 

However, the historical human practice appears not to have used an idealized a prior approach to design.

Instead, it appears they applied their arcs and geometry directly and interactively during the course of making.   This is how their unique hand working of design ideas ends up being free and asymmetric, but well structured at the same time.

[Marty Kasprzyk]

23 hours ago, David Beard said:

Very interesting and inevitable work.  Kudos.

 

However, the historical human practice appears not to have used an idealized a prior approach to design.

Instead, it appears they applied their arcs and geometry directly and interactively during the course of making.   This is how their unique hand working of design ideas ends up being free and asymmetric, but well structured at the same time.

Don't you find it odd that no Stradivari construction drawings exist?  Many of his paper patterns were saved but they all appear to be just tracings.

[David Beard]

12 minutes ago, Marty Kasprzyk said:

Don't you find it odd that no Stradivari construction drawings exist?  Many of his paper patterns were saved but they all appear to be just tracings.

No.  That's my point.  These geometry methods appear to have been directly and interactively as they built each instrument.

There are occasional drawing to help resolve a special question. So you have partial paper soundhole layout diagrams as he considered how to deal with his CV and TV viola designs for the Tuscan set.  And, you have a similar diagram as he decide how to do the Soundhole layout in the early Form B cellos.  But these are exceptions, and only partial diagrams.

 

I believe the constant work pattern was to use dividers and ratios to work the geometry recipes directly as the makers worked the wood.

This is why their design follow off of and embrace the vagaries that develop in the wood, rather than correct them back to some prior worked out design.

That paradigm of a worked out design first, then working to plan -- that paradigm belongs more to later eras.

[Torbjörn Zethelius]

On 7/11/2026 at 2:59 AM, Marty Kasprzyk said:

On 7/10/2026 at 3:42 AM, David Beard said:

Don't you find it odd that no Stradivari construction drawings exist?  Many of his paper patterns were saved but they all appear to be just tracings.

A lot of stuff has gone missing from the Stradivari collection. For example, there's the story about Giacomo Stradivari burning a 'bible' that contained the 1704 varnish recipe. What else was in that bible, and was it really a bible? The design method was a well kept secret, among many other things related to violin making IMO. 

[Marty Kasprzyk]

On 7/9/2026 at 9:42 PM, David Beard said:

Very interesting and inevitable work.  Kudos.

 

However, the historical human practice appears not to have used an idealized a prior approach to design.

Instead, it appears they applied their arcs and geometry directly and interactively during the course of making.   This is how their unique hand working of design ideas ends up being free and asymmetric, but well structured at the same time.

The bow's shape in the attached cave painting can be closely duplicated with three circle arcs which proves that the cavemen about 50 thousand years ago were already using compasses to design their bows.

[David Beard]

6 hours ago, Marty Kasprzyk said:

The bow's shape in the attached cave painting can be closely duplicated with three circle arcs which proves that the cavemen about 50 thousand years ago were already using compasses to design their bows.

Right you are Marty!!

Many curve families can be forced to confirm to other curve families.

Almost all freehand can be traced by a host of curve methods to 'as close as you like'.   That's almost the basis of integral calculus.

However, that isn't what you see in Cremona type violin work.

Instead, we see shapes whose uniquely cleanest and simplest expression is as joined line and circle arc segments, connected and sized and positioned and proportioned in consistent traditional ways.

You're just very stubborn about opposing the idea.

[Andreas Preuss]

10 hours ago, Torbjörn Zethelius said:

A lot of stuff has gone missing from the Stradivari collection. For example, there's the story about Giacomo Stradivari burning a 'bible' that contained the 1704 varnish recipe. What else was in that bible, and was it really a bible? The design method was a well kept secret, among many other things related to violin making IMO. 

A sane Catholic would certainly not burn a bible. So we might guess what this ‘bible’ was.
 

The secret of Antonio Stradivari was to me something the Family had no idea what it was. Because everyone around them talked about a secret varnish receipe they just jumped on that making up stories for what reason ever. 
 

Arch constructing was more likely a recipe of working procedures which wasn’t written down and was just one of the many ‘workshop rules’ from sharpening tools to cleaning the dust. When it comes to the Stradivari workshop we have in my opinion the tendency to ‘overload’ our interpretations. In the end it was most likely just a workshop driven by a man with high expectations against himself fueled by the success those expectations brought him. 

[Marty Kasprzyk]

4 hours ago, David Beard said:

Right you are Marty!!

Many curve families can be forced to confirm to other curve families.

Almost all freehand can be traced by a host of curve methods to 'as close as you like'.   That's almost the basis of integral calculus.

However, that isn't what you see in Cremona type violin work.

Instead, we see shapes whose uniquely cleanest and simplest expression is as joined line and circle arc segments, connected and sized and positioned and proportioned in consistent traditional ways.

You're just very stubborn about opposing the idea.

We have a famous old violin and we want to make a new one which duplicates its outline shape.

So we carefully make a drawing with circle arcs to closely match its outline shape.

Then we use this drawing to make our new instrument.

 

Wouldn it be a lot easier for a violin maker just trace the original violin and skip all the geometric and compass circle drawing steps if he wanted to make an accurate paper pattern?

 

[David Beard]

27 minutes ago, Marty Kasprzyk said:

We have a famous old violin and we want to make a new one which duplicates its outline shape.

So we carefully make a drawing with circle arcs to closely match its outline shape.

Then we use this drawing to make our new instrument.

 

Wouldn it be a lot easier for a violin maker just trace the original violin and skip all the geometric and compass circle drawing steps if he wanted to make an accurate paper pattern?

 

Not really.

We use our dividers and the geometry recipes as we work.

 

If you make to a drawing ahead of time, you will not get classical results.

[David Beard]

54 minutes ago, Marty Kasprzyk said:

Wouldn it be a lot easier for a violin maker just trace the original violin and skip all the geometric and compass circle drawing steps if he wanted to make an accurate paper pattern?

Yes!!

If you want to do the entirely not classical thing of making a copy, you might as well, in fact should, do the easy thing and trace.

 

If you instead want to make in a classical way, then you just simply never would attempt to copy.

Even if you wanted to make a new classical violin following the design choices of a famous instrument, let's say the Harrison, then if you actually work in a classical way, you will fail to get a good copy.

The interaction of classical method with the actual vagaries of your making a new instrument will introduce differences of reasonable significance.

Look at all the flowers on an abundant plant.  They all follow essentially identical DNA recipes, but no two end up truly alike.  The subtly unique process of each flower's development has impact.  Everyone of the flowers ends up meaningfully and clearly distinct.

This divergence in development even when repeating identical recipe is essential to the nature and character of classical making.

Further, we don't see the progress of Cremona making including any efforts to control such vagaries away.  Rather, throughout the period, they continued to use loose physical methods that allow ample vagaries to develop.  They embrace this aspect of their traditions rather than trying to improve or control it away.

There is no copying.

If you want to use Vuillaume's patented artificial flower process, then you can make an artificial plant with all identical leaves and all identical flowers.

If you want to grow actual flowers, then you're stuck with actually growing a real plant and respecting the fact that design recipe and growth process are going to interact somewhat uniquely, every time you try.  Your flowers will be real, and deeply related, but not perfect copies of each other.

[Marty Kasprzyk]

Do violin players or their audiences care about this?

[Dorisch]

6 hours ago, Marty Kasprzyk said:

Do violin players or their audiences care about this?

The audience surely not, they won’t be able to distinguish it at all, but players often seem decepted by these kind of things. 

I can’t tell how many players I know that seek either a Strad or Guarneri pattern violin because they associate specific characteristics with them. 
 

 

[Riccardo964]

On 6/5/2026 at 4:28 PM, Andrew Argraves said:

CremonaCad

That's fantastic. I will certainly would love to be involved in this, as a former software engineer myself. I will "look around" (the git page) and see some opportunities for engaging.

Thanks for sharing this! Not only the news, but the tool itself. I commend your commitment to the violin craft and to the software craft as well.

Cheerios,

Riccardo964

[Davide Sora]

I don't want to sound too nostalgic or anti-technology, but relying solely on drawing software misses a precious opportunity to train your hand and eye. If you're comfortable with these programs, there are certainly very useful aspects to using them, especially for saving time. But picking up a pencil and trying to draw the curves we'll later need to create (manually) in our work remains, for me, a very useful and indispensable skill, definitely worth learning. Unless you need to program a CNC machine. Food for thought.

[Dr. Mark]

10 hours ago, David Beard said:

If you want to do the entirely not classical thing of making a copy, you might as well, in fact should, do the easy thing and trace.

If you instead want to make in a classical way, then you just simply never would attempt to copy.

Mr. Kasprzyk appears to be arguing, through demonstration and other means, that speculative associations yield no more than speculative conclusions.   For a busy workshop environment, efficient methods are rather compelling.

[David Beard]

20 minutes ago, Dr. Mark said:

Mr. Kasprzyk appears to be arguing, through demonstration and other means, that speculative associations yield no more than speculative conclusions.   For a busy workshop environment, efficient methods are rather compelling.

My work is just about answering the question:  'what did they do?'

My motives have nothing to do with what busy shops should do today.

Also, while I've committed to sharing my results publicly, I have no concern whether others find my results interesting or care to use them in making.

I want to make entirely within old methods.  That is my personal goal.  I'm getting very near to being able to do that now.  That is my motive.  That is what I care about.

I will share my results, all of them, at least once, publicly.  This is one of the places I do that.   And, I will engage in reasonable discussion about the results.  Again, here is a good place for that.

But, I don't care who agrees, disagrees, ignores, or adopts these results as work methods.

 

After I get to the place of making by these methods instead of researching them, then after that I have one further hope.  I'd eventually like to see a small community of makers making within these recovered traditions and resuming the old evolution/dialog around 'what are the best preferred choices working within these methods'.  

I believe that dialog is what took them from the results of Andrea Amati to the results of Strad and then Del Gesu.

My research results show the work of all those generations as within the same recipe methods which consistent present limited choice options at every juncture.  The results see the difference in their results as differences in their preferences in settling those many recipe options present in the tradition.

We make much of Strad's 'secret' because we hear that he got very mad at his son leaving for a couple years after the mother's death, and going to Naples, like sharing methods with the Gaglianios there.

My research does seem to indicate very good secret about design methods or choices among the Cremona community makers.

Perhaps finishing/varnish was an exception, perhaps not.   Certainly with design, they all seemed to know what the others were doing, and to make similar explorations of preferences in the options at basically the same time.

So, I suspect their 'secrecy' was more like that in a guild.  There were likely rules about who and how and when you shared info.  Strad's son likely didn't commit the sin of sharing info, but of sharing too much, with the wrong people, at the wrong time, in the wrong way.

It's further interesting that that son returned within about 3 years, and was ever after a continuing presence in the family work shop for the rest of Strad's life.

Anyway, cheers.