Sketch timelines

[keithd]

Why don't sketches have timelines?

The only reason I've ever heard is that sketches are just collections of geometry with no sense of order. Hence, no timeline.

While that may be true, all the geometric elements in a sketch were drawn in some order. And replay of that order would accurately reproduce the sketch.

I bring it up because it seems to me that sketches would benefit from timelines in the same ways that 3D Design features do.

--

Keith

[TheCADWhisperer]

@keithd 

There is a bit of order with Undo/Redo until you close the file.

Can you post an example that illustrates where this functionality would be particularly useful?

[Drewpan]

Hi,

The overwhelming advice on the forum is that sketches should be simple and only

try to do one or two things. Sketches are the framework of a model that should be

created with all of the tools. An example of this is that I can create fillets on a sketch

but when I do it usually breaks the constraints and they have to be redone. It is much

easier to create a square extrusion and use the fillet tool. The end result is the same

and the sketch and design are much more robust.

If the above is the case then why do you need a timeline on a sketch? You are much

better off with a series of simple sketches with imported data such as Projections

than to have one complicated sketch that tries to do everything. A complex sketch is

frequently what is presented by a beginner who often does not realise what a sketch

is actually for. A series of small sketches DO show up in the timeline and really achieve

what you want.

If I am drawing anything relatively simple it should not matter what order I do it in.

What is the difference if I draw a square first and a circle second? If you looked at

that sketch six months down the track would you know or care?

I very much doubt that the devs would prioritise a timeline for sketches over some of

the other feature requests still in the pipeline.

Cheers

Andrew

[johnsonshiue]

Hi! Within a sketch, the dimension, constraint, and geometry solving isn't order-specific. There is no consideration of creation order when they are solved. As a result, having a timeline does not fit the sketch solve paradigm. I suspect you may consider grouping your sketches. Instead of creating a large sketch, you may want to create multiple sketches as layers. One can project from another so that they are associated.

Many thanks!

[jeff_strater]

A couple of comments from someone who was around Fusion from the beginning.

There are several reasons why a sketch does not have its own timeline.  Some are listed here (it is a reflection of the way sketch geometry + constraints + dimensions are evaluated - there is no order implied).  The other, related, reason is: No one else in the industry captures any sort of sketch "recipe".  Given that parametric design has been around for over 25 years, and no one else feels the need to do this, is an indication that maybe not many people see it as necessary.  You are not the first to ask for this in Fusion - there have been a handful of requests over the years, but not that many.

Would it be possible?  Yes, certainly.  Sketches are created in an order.  You draw one line, then the next, then add some dimensions, then go back and add more geometry, then maybe delete some geometry, edit some dimensions, etc.  We could certainly capture every one of those edit steps in a timeline-like recipe.  But, the main reason that a timeline or feature tree exists in a parametric CAD system is for editing.  You can edit an Extrude feature, and change its extents by typing in a new depth value.  I struggle a bit to see what "editing a sketch line" would look like from a UI perspective.  What would the edit dialog contain?  You edit lines by either dragging them or changing a dimension.  You edit dimensions by changing their value, not by rolling back to their time of creation and entering a different value.

Another reason for a timeline is to give you a way to reorder operations.  Again, given that there really is not implied order in a sketch solve, reordering operations should have no effect.

Finally, I think this is kind of a "Midas Touch"/"be careful what you ask for" situation.  Even a simple sketch often takes  hundreds of steps to create.  That's a lot of data to try to comprehend and manage.  I suspect it would be overwhelming, and people would tend not to use it.  

[keithd]

The overwhelming advice on the forum is that sketches should be simple and only

try to do one or two things.

Well, then, I disagree with the overwhelming advice. I use both simple and complex sketches. I don't disagree that there are times when just simple sketches is an okay strategy. But more often than not, fewer more complex sketches, possibly supported by a few minor simple sketches, is a better strategy. In my opinion. For me, the only-simple-sketches strategy has drawbacks.

First, I end up with a browser full of sketches all named SketchXX. Unless you are diligent, thoughtful, clever, and verbose in renaming them it can be difficult and time consuming to locate the sketch that contains the parameter or geometry you're wanting to modify. 

Second, I use constraints instead of dimensions as much as possible, and I rely on User Parameters for most dimensions. But constraints do not apply between sketches unless you use projections, which are evil. Maybe not pure evil, but close. In my experience, easily the most fragile feature in Fusion. To be avoided except where absolutely necessary. They are easily and frequently lost, which is annoying at best but at worst can cascade into a real mess, leaving your timeline a shambles of yellow or red errored-out features that are a nightmare or impossible to unwind. With a single complex sketch, constraints work with fewer projections to go bad and you know exactly where to go to find the geometry or dimension you're looking for. With most things easy to locate in one or two sketches, it's okay to use a few fixed dimensions and dxxx references instead of having to make every dimension into a User Parameter (so you can find it later). And that's a plus considering that Fusion's Parameter UI isn't that great.

I'm not advocating for ridiculously over complex sketches. I'm just saying that the other extreme of a mass of simple ("only try to do one or two things") is not good advice. In my opinion.

And it just seems to me that for sketches with some modicum of complexity a timeline would be useful. In many of the same ways it is useful in the 3D Design space.

--

Keith

[keithd]

There are several reasons why a sketch does not have its own timeline.  Some are listed here (it is a reflection of the way sketch geometry + constraints + dimensions are evaluated - there is no order implied). 

There is no order implied by a 3D solid, either. A 3D body in Fusion has an order ONLY because it has a timeline. I see no inherent fundamental difference between the features in a 3D body or in a 2D drawing, other than the third dimension. I can't disagree that Fusion may evaluate them differently. But that's a Fusion design choice, not a basic fundamental difference from a mathematical or physics sense.

The other, related, reason is: No one else in the industry captures any sort of sketch "recipe".  Given that parametric design has been around for over 25 years, and no one else feels the need to do this, is an indication that maybe not many people see it as necessary. 

What? People rode horses for thousands of years and no one (apparently) felt the need for an automobile. I hope you really don't actually think that justifies it.

You are not the first to ask for this in Fusion - there have been a handful of requests over the years, but not that many.

Would it be possible?  Yes, certainly.  Sketches are created in an order.  You draw one line, then the next, then add some dimensions, then go back and add more geometry, then maybe delete some geometry, edit some dimensions, etc.  We could certainly capture every one of those edit steps in a timeline-like recipe.  But, the main reason that a timeline or feature tree exists in a parametric CAD system is for editing. 

Well, I spend at least as much time editing sketch features as I do 3D features. If timelines are useful for editing 3D features, why not 2D features?

You can edit an Extrude feature, and change its extents by typing in a new depth value. 

And you can edit a sketch Pattern feature, and change its pitch by typing in a new value. Same.

I struggle a bit to see what "editing a sketch line" would look like from a UI perspective.  What would the edit dialog contain?  You edit lines by either dragging them or changing a dimension.  You edit dimensions by changing their value, not by rolling back to their time of creation and entering a different value.

I think with a little work a sensible UI could be developed. Certain things, like lines maybe, might be different. But many features, like circles, patterns, constraints, etc., are not at all conceptually different than 3D features from an editing sense. Reordering might well be useful if the sketch had a timeline. And lots of timeline features, like rolling back, suppressing, etc., would be just as useful in a sketch as in a component. 

Another reason for a timeline is to give you a way to reorder operations.  Again, given that there really is not implied order in a sketch solve, reordering operations should have no effect.

Yea, but that is only because that's the way Fusion chose to solve a sketch. If a sketch had a timeline and order then reordering might well have an effect. Even if it doesn't, that doesn't mean other timeline features wouldn't be useful.

Finally, I think this is kind of a "Midas Touch"/"be careful what you ask for" situation.  Even a simple sketch often takes  hundreds of steps to create.  That's a lot of data to try to comprehend and manage.  I suspect it would be overwhelming, and people would tend not to use it.  

Apparently, lots of people choose direct editing over parametric design, too. Their choice. Of course you might have to be a little clever in UI and timeline design to manage the complexity. But the timelines for some of my 3D models are overwhelming. But I still don't resort to direct editing.

Listen, this wasn't intended as a big feature ask, or anything like that. There are a thousand improvements and features I'd like to see in Fusion before timeline's in sketches. It was more a curiosity. It just seemed odd to me. I click on a component to open it and it has a timeline I can play with. I click on a sketch and, and, and, nothing. None of the features (or similar help) I get from the 3D timeline. It just seems like a difference without a reason, to me.

I was really just curious if anyone had a compelling reason. I can easily accept an argument that it would require a massive redesign to Fusion and precious developer time would be better spent on, say, an Electronics library UI that was actually usable. But so far I've not heard any convincing reason that sketches shouldn't have timelines.

--

Keith

[Drewpan]

Hi,

One of the main reasons that sketches should be simple is that a sketch is the foundation

of a model but it is the weakest link in the fusion tools. A simple example.

Fully constrained sketch.

Cut out the unwanted corner.

Notice the error in the lower right corner - unconstrained sketch.

In this case I only have to add a single dimension to fix it but I have seen sketches 

totally destroyed constraint wise from a similar operation.

End result.

Take our original sketch.

Now use the tools instead.

You have ended up with exactly the same model but not only is the workflow

captured in the timeline, it is also a much more robust model and unlikely to break

down the track as the foundation is more solid. I don't know why fusion works

like this but it does.

Every time you add geometry to a sketch you make it more complex. There is nothing

bad about this but the tool operations are much more stable than sketches. Sure you

cannot do everything with the tools, but you should not use complex sketches instead

of the tools just because you can. The Sketch tool is one of many. A screwdriver can

be used to do up screws and open paint cans. It can also be used instead of a chisel in

some circumstances but should it? Use the right tool for the right job. It is the same

in fusion.

I agree that using the simple sketch strategy can create many sketches and they can be

hard to track but remember that they are captured in the timeline and you don't have to

name EVERY one. If it takes six to ten sketches to create a feature then you could name

the first one and leave the rest as SketchXX. Another consideration is that some workflows

dump ALL of the sketches in the Top component in the browser tree, consider activating the

component you are working on and keeping the relevent sketches within that component.

All of the sketches related to that component are all in the same place - much easier to

locate a single sketch.

I disagree that projections are Evil and this is probably why you are struggling with sketches.

You should only be projecting the minimum from another sketch into a new one and any point

or line projected is already constrained. This makes your job easier not harder. Yes it is easy

to break a projection by making a change to the original sketch, but you should know what the

change was and it should be easy to fix it on the FIRST yellow broken error. Once you fix the

first one, usually all or most of those errors disappear. Parameters are carried over into new

sketches, the projected geometry is already constrained and it should make no difference to

your workflow.

My advice to you is that if the Design in Place workflow is not for you then don't use projections.

Stick to the Design at Origin method and use Joints to create your model.

If you are not a very experienced user, use simple sketches to do one or two things is excellent

advice, especially for beginners. I think you might find that most of the Gurus in the forum use

the advice they give out, including this. Some of the Gurus have decades of experience and they

ALWAYS use this advice themselves on their own models. There is probably a good reason for this.

I am not saying that this is the ONLY method the Gurus use. What is simple for some is complex

for others. The advice is to stop people from trying to create a Rocket Engine in a single sketch

and wondering why they are having problems.

Cheers

Andrew

[TheCADWhisperer]

---

@keithd wrote:

I was really just curious if anyone had a compelling reason. 

---

@keithd 

See my first reply.

Can you post an example that illustrates where this would have been useful?

[keithd]

Hi,

One of the main reasons that sketches should be simple is that a sketch is the foundation

of a model but it is the weakest link in the fusion tools.

But you'd expect it to be "weaker" in some sense. It need only deal with 2D geometry, not 3D features.

A simple example.

Fully constrained sketch.

Cut out the unwanted corner.

Notice the error in the lower right corner - unconstrained sketch.

In this case I only have to add a single dimension to fix it but I have seen sketches 

totally destroyed constraint wise from a similar operation.

End result.

Take our original sketch.

Now use the tools instead.

You have ended up with exactly the same model but not only is the workflow

captured in the timeline, it is also a much more robust model and unlikely to break

down the track as the foundation is more solid.

My post wasn't regarding the use of sketch tools vs 3D tools. I agree completely that the latter are superior and should be used instead of sketch tools where possible. My post was about why sketches don't have timelines. Your answer, as I understood it, was that if you use only simple sketches you don't need sketch timelines. With which I strongly disagree, regardless of the fact the 3D tools are stronger than sketch tools. (I might point out, however, that if sketches had timelines then your point is a bit weaker. But they don't, so I don't disagree with your point at all.)

I don't know why fusion works like this but it does.

Every time you add geometry to a sketch you make it more complex. There is nothing

bad about this but the tool operations are much more stable than sketches. Sure you

cannot do everything with the tools, but you should not use complex sketches instead

of the tools just because you can.

The question is the strategy of using a few number of more complex sketches vs a larger number of simpler sketches. 3D tool superiority applies in either case and so is irrelevant to the question. I am not, nor have I ever, argued for using sketch tools over 3D tools. Assuming we need to create a given set of operations with sketches, is it better to put the geometry in a few number of more complex sketches or a larger number of simpler sketches. I argue the latter is usually more problematic than the former. And that sketches with timelines would facilitate the former. I think your advocacy of the only-simply-sketches strategy, and your strong support of 3D tools over sketch tools, is largely a way to get around the shortcomings of the sketch tools, including the absence of a timeline. I think the more-complex-sketch strategy is fundamentally better, and would be even more so with sketch timelines. Primarily because of the fragility of projections but also because it's simpler to define and find geom

The Sketch tool is one of many. A screwdriver can

be used to do up screws and open paint cans. It can also be used instead of a chisel in

some circumstances but should it? Use the right tool for the right job. It is the same

in fusion.

I agree, but it's not to the point.

I agree that using the simple sketch strategy can create many sketches and they can be

hard to track but remember that they are captured in the timeline and you don't have to

name EVERY one. If it takes six to ten sketches to create a feature then you could name

the first one and leave the rest as SketchXX.

I've tried every imaginable strategy to manage a large number of sketches. In a large timeline, every strategy I've tried fails. Multiple simple sketches inevitably end up scattered across the timeline despite being related or having a logical grouping. And finding specific geometry to modify later on is not easy. The only mitigating strategy I've found is the use of User Parameters to the extreme, to avoid having to find geometry in earlier sketches when possible. But that isn't foolproof as its often the geometry itself, not dimensions, that need edited.

Another consideration is that some workflows

dump ALL of the sketches in the Top component in the browser tree, consider activating the

component you are working on and keeping the relevent sketches within that component.

I absolutely NEVER allow any sketches at the top level. Never ever. All my sketches are local to the activated component. But even single compo

All of the sketches related to that component are all in the same place - much easier to

locate a single sketch.

I disagree that projections are Evil and this is probably why you are struggling with sketches.

Who said I was struggling with sketches? I am not. I just think they would be better, and my preferred more-complex-sketches strategy would benefit from sketch timelines. 

You should only be projecting the minimum from another sketch into a new one and any point

or line projected is already constrained.

Bingo! Exactly why I prefer containing geometry in fewer sketches. Because it dramatically reduces dependence on projections. I've seen random YouTube "tutorials" advocate for avoiding projections at all cost. I find that impossible, or at least very difficult to achieve. So I use them. But judiciously, because I know that if something goes wrong and errors start popping up in my timeline, the vast majority of times it will be due to a lost projection.

This makes your job easier not harder. Yes it is easy

to break a projection by making a change to the original sketch, but you should know what the

change was and it should be easy to fix it on the FIRST yellow broken error. Once you fix the

first one, usually all or most of those errors disappear.

Not at all my experience.

Parameters are carried over into new

sketches, the projected geometry is already constrained and it should make no difference to

your workflow.

My advice to you is that if the Design in Place workflow is not for you then don't use projections.

Stick to the Design at Origin method and use Joints to create your model.

Projections from design in-place components are problematic. Projections from jointed components are simply a disaster waiting to happen. And the wait won't be long. But, in my opinion, designing with joints is a superior methodology in general. It is often a huge Parameter mess to avoid projections from jointed components. But I avoid it like the plague, as the consequences are almost always too ugly to bear.

If you are not a very experienced user,

I am experienced. With Fusion. And other CAD systems.

use simple sketches to do one or two things is excellent advice, especially for beginners.

Maybe for beginners. Not for me, because I've suffered the consequences enough.

I think you might find that most of the Gurus in the forum use

the advice they give out, including this. Some of the Gurus have decades of experience and they

ALWAYS use this advice themselves on their own models. There is probably a good reason for this.

ALWAYS? Really? Well, maybe one of their reasons is to sidestep the shortcomings of the sketch tools, like lack of a timeline. I have and still do use both only-simple and more-complex sketch strategies, depending on the situation. Most often something in between. My experience is that I have fewer problems when I have fewer sketches, even if that means one or more sketches become complex. I try to find the right balance, knowing the problems I'm likely to create downstream with either strategy. Improved sketching, including a timeline, would help minimize the number of sketches and, thus, the number of projections. Which would be a win, in my opinion. Your Gurus can have whatever opinion they like. 

I am not saying that this is the ONLY method the Gurus use. What is simple for some is complex

for others. The advice is to stop people from trying to create a Rocket Engine in a single sketch

and wondering why they are having problems.

Gee, that's good advice. I should have thought of it.

Cheers

Andrew

--

Keith

[keithd]

Can you post an example that illustrates where this would have been useful?

Well, let's try it this way. I assume you agree there are examples where timelines are useful in the 3D Design space? If so, I assume you can list those. Okay, there you have my list of examples in answer to your question.

My point being that the uses would be no different, conceptually, in the 2D geometry space than in the 3D space. Sure, details are different, circles vs cylinders, for example, but also some are the same, geometry patterns vs feature patterns, for example. If you argue that geometry has no order, then on the same grounds I will argue that 3D solids have no order, either. 3D solids have order only because Fusion replays and rebuilds the model in a timeline order. There is no inherent order to a 3D solid. The same ordering methodology could be applied to sketch geometry, in which case a timeline would produce the same conceptual benefits. Like editing, rollback, suppression, attributes (e.g., construction lines), pick your favorite example. 

Or maybe you're suggesting that timelines in general aren't useful? In which case you must be an advocate of direct editing. So let's just imagine that Fusion were direct editing only. If a user then suggested adding timelines, would you ask them for examples where they would be useful? If your mindset was in direct editing, then you might. Analogously, if your mindset is in unordered 2D drawings (like MacDraw), then you might ask for examples where a timeline would be useful.

I'm taking the mindset of imagining ordered 2D drawings and am asking why timelines would not have the same (or similar) uses as in the 3D space?

--

Keith

[TheCADWhisperer]

Why are you using Fusion? (or any CAD)

I am not interested in a wall of text.

Show me an example in CAD.

[TheCADWhisperer]

@keithd 

To be clear - I am not here to argue that your suggestion doesn't have merit.

I am here to see an example that supports your suggestion.

Over more than 20 years I have gotten Autodesk to incorporate several of my ideas - but I always started with examples and then was invited to conversations with developers.

[jeff_strater · Accepted solution]

"I was really just curious if anyone had a compelling reason."

I think I gave the actual, historical reasons why the development team did not choose to implement a sketch timeline.  Whether they are "compelling" or not, obviously, are in the eye of the beholder, and you clearly think they are not.  That's fine.  We can disagree on that point.  But, those are the reasons.

[keithd]

"I was really just curious if anyone had a compelling reason."

I think I gave the actual, historical reasons why the development team did not choose to implement a sketch timeline.  Whether they are "compelling" or not, obviously, are in the eye of the beholder, and you clearly think they are not.  That's fine.  We can disagree on that point.  But, those are the reasons.

By compelling, I meant a compelling technical explanation of why it wasn't included, considering that it is included in the 3D space. I didn't get that.

But I completely accept the historical reasons you gave as the explanation.

And please don't misunderstand. I am not advocating for this feature. As I said earlier, there are a great many features I'd rather see Autodesk spend their development time on. Including many to the sketch environment. My original post was merely a question: "Why don't sketches have timelines?" I still think it's odd, considering that timelines are such an important feature to Fusion 3D. And I often run into situations where the absence is striking. I mean, just the fact that 2D works differently than 3D seems, well, odd.

But it is often the case that things are what they are for reasons other than technical merit. Which, judging from responses to my question, might be the case here.

I accept your response as the answer to my question. Thank you.

--

Keith