Silk: a NURBS workbench
Forum rules
Be nice to others! Respect the FreeCAD code of conduct!
Be nice to others! Respect the FreeCAD code of conduct!
Re: Silk: a NURBS workbench
what was the point of all this?
Well, we don't have to use ugly curves. and we can edit this model.
let's change the center 3L sketch. pull the zig-zag line up. the model updates on the drop portion of a sketching drag-drop. this is fairly close to realtime...and actually WAY better than realtime. realtime would lock up our computers forever. just do small drag-drops. play around .
immediately better .
but the zebra still shows the same 'quality'. lines meet, but connect sharply. .
let show what what a severely non tangent model looks like (by breaking the parallel constraint of the back blue node sketch): .
here you can see a clearly non G1 zebra: the lines do not even connect
Well, we don't have to use ugly curves. and we can edit this model.
let's change the center 3L sketch. pull the zig-zag line up. the model updates on the drop portion of a sketching drag-drop. this is fairly close to realtime...and actually WAY better than realtime. realtime would lock up our computers forever. just do small drag-drops. play around .
immediately better .
but the zebra still shows the same 'quality'. lines meet, but connect sharply. .
let show what what a severely non tangent model looks like (by breaking the parallel constraint of the back blue node sketch): .
here you can see a clearly non G1 zebra: the lines do not even connect
Last edited by emills2 on Wed Jun 07, 2023 4:09 am, edited 2 times in total.
Re: Silk: a NURBS workbench
now let's try to get something 'nice' out of this model.
i made the node length ratio 1:1 instead of 3:2, and made the 3L sketches have vaguely similar shapes. now you can see the grid is fairly evenly spaced. it's nothing amazing, but it may fulfill a job. .
and still, the zebra reveals the low quality there are ways to use these rules, as part of more advanced setups, which are actually, very useful. but even those setups still rely on blends for the majority of the model. so you might as well start with blends, because you'll need them...if you stick with Silk.
why would you stick with Silk after reading this? well, remember the end game here:
a model with no G1 rules at all, just blends. .
a model that relies mostly on G1 rule for matching the torus perfectly, and a single blend on the inside. [edit: first i matched the torus perfectly, and then, i intentionally slightly broke a G1 rule as a design choice, ending with a result i wanted, and an acceptable error]
This is ONE Silk surface of type 6X4, mirrored twice
i made the node length ratio 1:1 instead of 3:2, and made the 3L sketches have vaguely similar shapes. now you can see the grid is fairly evenly spaced. it's nothing amazing, but it may fulfill a job. .
and still, the zebra reveals the low quality there are ways to use these rules, as part of more advanced setups, which are actually, very useful. but even those setups still rely on blends for the majority of the model. so you might as well start with blends, because you'll need them...if you stick with Silk.
why would you stick with Silk after reading this? well, remember the end game here:
a model with no G1 rules at all, just blends. .
a model that relies mostly on G1 rule for matching the torus perfectly, and a single blend on the inside. [edit: first i matched the torus perfectly, and then, i intentionally slightly broke a G1 rule as a design choice, ending with a result i wanted, and an acceptable error]
This is ONE Silk surface of type 6X4, mirrored twice
Last edited by emills2 on Wed Jun 07, 2023 5:42 pm, edited 2 times in total.
Re: Silk: a NURBS workbench
here i use my best tricks to get G3 directly on the 4X4 surfaces, so that i get better than just mirror good across the mirror plane. this is a fairly complex setup, and although it can do more than mirrors, it is currently limited to the halfway point of a surface or curve.
some flaws here in the back, which can be fixed with Surface WB Filling Surface tool
and since i know my 'half' 4X4 surface is almost as good as a single surface (which would actually be perfectly smooth), i feel confident to work the blend on half the model, thereby doing half the blending work
.some flaws here in the back, which can be fixed with Surface WB Filling Surface tool
Re: Silk: a NURBS workbench
a long long time ago, when i was still trying to make the surfaces 'line up good directly', i got up to this point, following G1 rules: .michacassola wrote: ↑Tue Jun 06, 2023 12:40 pm way to get two 4x4 surfaces next to each other that are already smooth...
aligning just 2 surfaces, you might think it's not too bad. but then you build surfaces on multiple sides, like above, and now the G1 seams cross each other!! now the flaws look like crosses and really jump out.
i even figured out rules to do G2 manually! but now the second surface has no design freedom at all. completely impossible to bridge two surfaces this way. you start from one surface, and the G2 'bridge' just goes off wherever it wants. you can 'touch' another surface, but now you can't even do G1 on the other side of the 'bridge' and that's when i said, 'this is going nowhere'. i have G1 but it looks like crap. i don't have freedom to design my curves exactly the way i want (because of the 2 G1 rules), so i sacrifice the middle of my curves for G1....and this is what i get???
It was time to start blending. keep my curves exactly the way i want, especially the middle portion, and cut off small pieces near the seams.
as soon as i had a test prototype in hands, the design power just multiplied by one million. no more lining up sketches, no more unwanted dimensions in the sketch to match ratios....just match endpoints and the basic sketches/polys/grids/surfaces are done. some people even like to design directly with the blend curves and surface (see AndyK's earlier post in this thread) which is fine as long as there aren't too many surfaces to join (my steering wheel grip is an example of this as well).
see here
viewtopic.php?p=516039#p516039
and here
viewtopic.php?p=675315#p675315
i rarely use these direct tools personally. the manually built poly6, grid64, and grid66 (and their curves and surfaces) exist as direct tools, but i use automatic versions in my blended models. i only design with poly4, but it's fun to see what people manage to come up with in terms of workflows.
Last edited by emills2 on Wed Jun 07, 2023 7:59 pm, edited 4 times in total.
Re: Silk: a NURBS workbench
Youtube user Rafael3D (https://www.youtube.com/@Rafa3D - spanish language) has just shared a model with me and allowed me to post it here.
this is a case of G1 modeling that takes advantage of symmetry to get great results. it's a remix of my steering wheel grip that he remade completely without arcs. all simple 3L sketches, followed with standard CAD tools.
Thanks Rafa!
(his channel covers many different CAD programs, many FreeCAD workbenches, and he is a very accomplished designer - excellent resource for Spanish speakers)
(his channel covers many different CAD programs, many FreeCAD workbenches, and he is a very accomplished designer - excellent resource for Spanish speakers)
As luck would have it, he also designed a shoe in FreeCAD (Curves WB?) a while back https://www.youtube.com/watch?v=u50sKJH7-aM. Micha, I don't know if you speak spanish, but this video might have some hints, even just visually to see how he structured his model for Curves. I hope you give Silk a serious shot long term, but if there is an existing setup that works in Curves...get the job done now and get paid!
- michacassola
- Posts: 23
- Joined: Fri Feb 17, 2023 10:17 am
- Contact:
Re: Silk: a NURBS workbench
Hi, unfortunately the video does not have any hints about the actual surfacing and I do not understand Spanish. He only shows some curves, that in no way can be making up the shoe form shell surfaces on their own when sweeping or surfacing in another way. He never clicks on the folders he has the sketches, curves and surfaces in. I thought that that was intentional and gave him a thumbs down on Youtube (bevor ever starting writing here) as he just talks and turns the shoe around with and without Zebra...
I think I will have to try to figure out a better Curves WB way, as I need more direct control over the outcome than with Silk. But it was very much worth the work as the blending will have good use for me on other things. Thank you very much again! Your insights and your Workbench are very good!
I think I will have to try to figure out a better Curves WB way, as I need more direct control over the outcome than with Silk. But it was very much worth the work as the blending will have good use for me on other things. Thank you very much again! Your insights and your Workbench are very good!
Re: Silk: a NURBS workbench
Switch on subtitles auto-translation to English (Settings -> Subtitles->Auto-Translation). Ignore FreeCAD translated to free cut
Anyway, he has to content behind a pay wall, so I can't tell if there are more details for the shoe in his other videos about Curves.
Anyway, he has to content behind a pay wall, so I can't tell if there are more details for the shoe in his other videos about Curves.
- michacassola
- Posts: 23
- Joined: Fri Feb 17, 2023 10:17 am
- Contact:
Re: Silk: a NURBS workbench
Thanks @PrzemoF, if nothing else he proves it is possible with the Curves WB.
I will have to keep trying.
I will have to keep trying.
Re: Silk: a NURBS workbench
Yes, I was aware that this video was not a detailed tutorial, and only pointed it out to add data for you to decide between Silk, Curves, or something else entirely. The fact that he at least showed his curves (so we can count them, etc), and specified the model was from Curves WB is the bare minimum that i considered to be useful information.michacassola wrote: ↑Thu Jun 08, 2023 11:48 am Thanks @PrzemoF, if nothing else he proves it is possible with the Curves WB.
I will have to keep trying.
His detailed tutorials are very complete, and can consist of multiple episodes of one hour each (I don't know if there exists a full detail tutorial for this shoe model, possibly not). If you don't understand Spanish, auto-translate will likely lead to frustration, because it will be stunningly difficult to determine the technical terms/functions/buttons/parameters.
I think there is room for all different kinds of useful content/videos at different levels. Exposing overall capacity of a tool/system. Going over the main workflow simply by naming tools/functions used. And at the lowest level...the nitty gritty of every single button click and every number typed in.
When we actually like the tool and overall result, we often wish there was a direct flow into the next level down. Unfortunately, that next level does not always exist.
I am certainly guilty of this as well. Silk has a few ultra detailed tutorials....some pictures of complicated models...some long posts on this thread about certain technical aspects....it is very much a "draw the rest of the owl" situation
in the end, what i have settled on is to do my best to give detailed answers to users who actually ask questions, and try to do more than my best for users who actually spend time using the tools. I am very aware that this leaves an enormous gap, and i cannot blame people who might think "if there's no documentation for a functionality, then that functionality doe not exist".
[edit: removed rambling joke]