Modeling workflows and their advantages\disadvantages on the workplace .

Joao Sapiro
sublime tool
wrote this today since i usually have to work from files of other artists and sometimes i have to redo their whole work since their workflow involves alot of destructivity. Hope this image helps you see some benefits of a non destructive workflow:
USEFULL LINKS :
Zip of the models used here, if you test other methods avoid altering the geo , i know its not super clean but it smooths fine :
https://dl.dropboxusercontent.com/u/575737/help images/Objects_test.7z
quadchamfer vs max chamfer by Pedro Amorim :
https://dl.dropboxusercontent.com/u/93793/Screenshots/1922_px_px.mp4
Possible corner solution for quadchamfer by steffen Unger :
http://www.steffenunger.com/advices/ughyay.gif

Replies
-
Top notch good sir!
-
Thank you so much for the breakdowns of different workflows. As someone who's starting to get more invested in hard-surface modeling I often find myself worrying about whether the workflow I'm using is the best one, and it's helpful to get an understanding of what exactly makes certain workflows better in specific scenarios.
-
The micro chamfer method could be done as a modifier on top of the base edit poly. This would still allow for quick edits to the larger shape of the object, and still allow for loop controls over those chamfers. Although I suppose the same result could be had by adding an edit poly modifier on top of the quad chamfer one.
Really, in all the methods, the moral is that Max's modifier stack is your friend.
This is a great breakdown, by the way. Thanks -
Great post Joao!
-
Great breakdown, thanks for taking the time to write it!
-
Lot of thanks for sharing this man!
-
Really well put together and explained perfectly, especially with the creasing bit.
Big thank you for sharing this!
To add to the microchamfer bit - atleast in Max, you could do all your microchamfering work on an edit poly layer (rename this layer, if it helps too), and then do the rest of the sub-d work via modifiers ontop of that stack. From there, you could always delete that edit poly layer instead of collapsing, therefore maintaining that original base shape.
@purplekami Comes down to which is the most efficient/fastest for that particular mesh. More complex stuff, i'd use something like opensubdiv, whereas if I had something much simpler, like something quickly box modeled, you can just use double smooth to achieve a similar result with half the amount of steps required.purplekami said:Thank you so much for the breakdowns of different workflows. As someone who's starting to get more invested in hard-surface modeling I often find myself worrying about whether the workflow I'm using is the best one, and it's helpful to get an understanding of what exactly makes certain workflows better in specific scenarios.
-
uuuuugghhhh, so good. Anyone knows of a Maya version of this?
-
Klash-120 said:uuuuugghhhh, so good. Anyone knows of a Maya version of this?
Definitely saving this post. -
"To add to the microchamfer bit - atleast in Max, you could do all your microchamfering work on an edit poly layer (rename this layer, if it helps too), and then do the rest of the sub-d work via modifiers ontop of that stack. From there, you could always delete that edit poly layer instead of collapsing, therefore maintaining that original base shape."
That can work definately, the problem is that when you alter the baseshape then you have to redo the chanfers nontheless , even if you save the selection it is prone to get screwed over.
A simple modifier that toggles clean nice non destructive loops seems a faster solution for me since when i send it to clients if they want to alter stuff and they dont have quad chamfer , even removing the loops they get the baseshape.
"@purplekami Comes down to which is the most efficient/fastest for that particular mesh. More complex stuff, i'd use something like opensubdiv, whereas if I had something much simpler, like something quickly box modeled, you can just use double smooth to achieve a similar result with half the amount of steps required."
Yup, you can pretty much use whatever you feel is more confortable, but one thing i notice when i have models where they used different methods it can become quite messy and still makes editing more complicating since some pieces have opensubd and others doublesmooth etc. Id advise to keep things consistent and clean .
But whatever works, works ( i wanted to put a simple smiley but this forum makes that super hard )
Ill be sharing more shit like this in the future ! Meanwhile for good modeling practises id always advise to check the stickied how the fuck do i model this -
Tomiajayi said:Ha! no Maya version for you :P Max owns poly modeling !
Definitely saving this post.
Klash-120 - Creasing is the equivalent in Maya:
Everything you see done with the smoothing groups can be done with creasing inside Maya. As long as you dont delete the history of your subdivided model, you can always go back. Programs like Max and Zbrush can also rebuild levels of subdivided geo if the levels are lost. When using creasing, dont subdivide your models, use the preview smoothing feature (3). If it isnt smoothed enough, you can increase the number of iterations somewhere in the preferences or attributes of the model. yay -
Kanni3d said:Really well put together and explained perfectly, especially with the creasing bit.
Big thank you for sharing this!
To add to the microchamfer bit - atleast in Max, you could do all your microchamfering work on an edit poly layer (rename this layer, if it helps too), and then do the rest of the sub-d work via modifiers ontop of that stack. From there, you could always delete that edit poly layer instead of collapsing, therefore maintaining that original base shape.
@purplekami Comes down to which is the most efficient/fastest for that particular mesh. More complex stuff, i'd use something like opensubdiv, whereas if I had something much simpler, like something quickly box modeled, you can just use double smooth to achieve a similar result with half the amount of steps required.
Both methods got their pros and cons, and both methods have their same fair share of artifacts. that said, neither open subdiv nor doublesmooth offer the same amount of control over the edge quality that quadchamfer does. As the edgewidth always directky correlates with the density of your surrounding mesh. Which might be okay in some cases, but in many is just not up to par with quadchamfer.
@edomi: creasing would suffer from the same issues like opensubdiv or doublesmooth, only very indirect control over the edge quality -
Neox said:
-
Edit: Nevermind. was trying to get the same results with the built in Max modifier, but it's a POS compared the one Marius made.
-
nice comparison between the methods.
you can change sub-d level with Page-Up and Page-Down keys :]ENODMI said:If it isnt smoothed enough, you can increase the number of iterations somewhere in the preferences or attributes of the model. yay
-
Tzur_H said:you can change sub-d level with Page-Up and Page-Down keys :]
-
I'm a huge fan of crease sets for the speed they allow.
-
Actually double smoothing in Max is exactly what creasing does in Maya.
In Max:
Put one TurboSmooth with 2 subdivisions with "Separate | Smoothing Groups" on and then another Turbosmooth without the Smoothing Groups option.
The equivalent In Maya:
Select all the hard edges have them have a crease value of 2 (always use a Crease Set) and the subdivision level to 2+1=3. This can be done in the Crease Set Editor by selecting the mesh in the viewport and clicking on Display -> Optimal Subdivision Level. Press 3 and "voila", Smooth Preview!
This means you can easily translate unsmoothed high res geometries between Max and Maya by using Hard Edges/Smoothing Grops. When exporting from one software, make sure you tick "Smoothing Groups" ON and make sure Triangulate or Keep edge orientation is OFF in the Geometry section of the FBX export settings. Then when importing in the other software, make sure you again have Smoothing Groups ON in the Geometry section of the FBX improt options. After that you just need the respective method above to re-add the smooth in the target application.
PS: If you are using FBX version 2016/2017 in both packages, then creases should be translated automatically. If versions don't match, use my method above.
-
the only problem i see with creases as i have written :
- not every program support them equally and sending crease info between them can be a pain , lots of fiddling that is time spent dealing with it.
- if you want to alter the base geo/remove/add shapes you will have to redo the creases , even if you have scripts to save the edges you still have to manually select them instead of smoothing groups for example that are set by face and super fast to work from , i dont have to worry about altering shit.
- sending collapsed crease sets incase your client package doesnt support them/client wants regular sudb modeling can be a pain for them to edit on their end as it can be super dense meshes.
-
ValN84 said:Actually double smoothing in Max is exactly what creasing does in Maya.
In Max:
Put one TurboSmooth with 2 subdivisions with "Separate | Smoothing Groups" on and then another Turbosmooth without the Smoothing Groups option.
The equivalent In Maya:
Select all the hard edges have them have a crease value of 2 (always use a Crease Set) and the subdivision level to 2+1=3. This can be done in the Crease Set Editor by selecting the mesh in the viewport and clicking on Display -> Optimal Subdivision Level. Press 3 and "voila", Smooth Preview!
This means you can easily translate unsmoothed high res geometries between Max and Maya by using Hard Edges/Smoothing Grops. When exporting from one software, make sure you tick "Smoothing Groups" ON and make sure Triangulate or Keep edge orientation is OFF in the Geometry section of the FBX export settings. Then when importing in the other software, make sure you again have Smoothing Groups ON in the Geometry section of the FBX improt options. After that you just need the respective method above to re-add the smooth in the target application.
PS: If you are using FBX version 2016/2017 in both packages, then creases should be translated automatically. If versions don't match, use my method above.
My workflow at the moment when working with wierd hardsurface shapes is to make a proxy model in zbrush of the hardsurface shapes then build around this in maya to get an approximate lowpoly then slowly subdivide those lowpoly models and add supporting edges to get the highpoly. -
"My workflow at the moment when working with wierd hardsurface shapes is to make a proxy model in zbrush of the hardsurface shapes then build around this in maya to get an approximate lowpoly then slowly subdivide those lowpoly models and add supporting edges to get the highpoly."
so you use two apps , to get an approximation of the shape you want. That must be so exausting ! imagine having to do a mecha with lots of pieces with curved surfaces , i would have dhiarreah just by thinking on the work with that workflow
I recomend checking the how u model the shapes sticky thread, lots of great pearls of wisdom from Perna in regards to workflow.
The objective of this image is to weight the pros and cons of various methods of modeling with client needs in mind and if your coworkers need to edit stuff
Feel free to share others etc, the more info the better ! -
Joao Sapiro said:
so you use two apps , to get an approximation of the shape you want. That must be so exausting ! imagine having to do a mecha with lots of pieces with curved surfaces , i would have dhiarreah just by thinking on the work with that workflow
Great stuff mate. Are you using Miauu's QC exclusively? Pain in the arse that Adesk omitted some of its best features when they stuck the guts of it into Max. Marius is still selling QC for $40 so it must have been part of the deal that he keep the best bits.
Also, the thing about Pedro's example is that it leaves a 6-valance pole that subdivides terribly. It's almost better to leave 'quad intersection' disabled and split the N-gon down the centre, creating 2 quads with a diagonal edge back to the corner vert with an edit-poly mod under the TS. Although re-routing topo kind of defeats the purpose of the workflow.
-
Is marius quad chamfer really worth it if we have quad chamfer in max? what would be the main differences?
-
Vincent3d said:Is marius quad chamfer really worth it if we have quad chamfer in max? what would be the main differences?
https://dl.dropboxusercontent.com/u/93793/Screenshots/1922_px_px.mp4
( by pedro amorim ) -
Vincent3d said:Is marius quad chamfer really worth it if we have quad chamfer in max? what would be the main differences?
Quad intersection
Edge weight interaction
The Edit Chamfer controls.
-
Hello,
first, thanks for sharing.
Second, so you are using smoothin groups to feed the quadchamfer, right?
Third, I'm wondering if it's possible to do the same but in Blender.
-
yes you can drive the bevel modifier in blender with edge weights - its pretty solid.
-
kio said:yes you can drive the bevel modifier in blender with edge weights - its pretty solid.
-
Joao Sapiro said:so you use two apps , to get an approximation of the shape you want. That must be so exausting ! imagine having to do a mecha with lots of pieces with curved surfaces , i would have dhiarreah just by thinking on the work with that workflow
I just find it easier to go straight to sculpt and rough out all the forms and then retopo in maya than starting directly in maya with nothing and then finding Ive made the forms not match the concept enough, if you know what I mean? I make mostly lowpoly stuff at work so Ive just been using this workflow at home really, thats why Im interested in this thread to see if theres any easier way to quickly get a highpoly in maya -
Ged said:would love to see a visual tutorial of this workflow. I tried creases a while back in maya but I think the maya history stack got messy and it ended up being a pain in the ass.
My workflow at the moment when working with wierd hardsurface shapes is to make a proxy model in zbrush of the hardsurface shapes then build around this in maya to get an approximate lowpoly then slowly subdivide those lowpoly models and add supporting edges to get the highpoly.https://www.youtube.com/watch?v=QFckVSyvVy4
-
Hey rexus thanks but I was refering to the post before mine which I quoted explaining a maya subd workflow with creasing, that video is good but not maya.
-
I did some tests based off of the shapes Joao posted, to see what kind of results I'd get. I like to rely on the Double Turbosmooth method a lot, but as mentioned above by Joao it can get really heavy and dense. Not only is the double turbosmooth method increasing the density by a factor of about 4,000%, it also has pinching along the curved edge and gives non even edges around the model.
Some pics of the tests:
-
Joshflighter said:
-
you could solve that pinch by triangulating this part tho. but it would only work in this case, on curved surfaces you can usually not get away with this
-
I assume you are "struggling" with boxmodelling in this case? as you change a shape (a cube) into another shape (say one armor plate). I always hated that approach as well, because it is so indirect and strange to me.
Personally in that case i would use silhouette modelling, focussing on the shapes first, then on the topology.
i usually only use sculpting for the blockout wgen it is really hard for me to wrap my head around a shape, in your example they are all pretty simple to follow -
Neox said:
-
Ged - Nothing wrong with your workflow. working this way can really help nail the look of a concept or piece that is very organic or stylized. A lot of work that is modeled straight forward tends to lose most of those artistic rhythms, and it can become a very tedious game of pushing and pulling all the relative shapes until you find it. As far as silhouette modeling, youll want to bring in some orthographic images into your viewport to model over and use as a ref. You usually start with a geo plane, cutting and extruding to match the shapes. Its honestly not any easier than box modeling to me, it just takes practice to get good and efficient at it. It was mentioned before, but head over to here: http://polycount.com/discussion/56014/how-the-f-do-i-model-this-reply-for-help-with-specific-shapes/p137
Theres a ton of examples on modeling.
-
ugh, hate creasing....it's so evil
but I mostly use this method, it involves manual cleanup and planning but is effective.
Possible corner solution for quadchamfer by steffen Unger :
http://www.steffenunger.com/advices/ughyay.gif
-
ENODMI said:Ged - Nothing wrong with your workflow. working this way can really help nail the look of a concept or piece that is very organic or stylized....
-
-
maybe you should try mesh fusion and pFusion of modo 10.2
-
irvingao said:maybe you should try mesh fusion and pFusion of modo 10.2
z
-
Yup.
The problem is that you need a whole studio to use modo and addapt its workflow for modo, because if you need to send it to maya or max you get polysoup like @musashidan stated.
The purpose of this post is to show that using smart ways of work in wich artists can easily edit each others work simply helps speed up production tremendously , thats why i mentioned quadchamfer since its the biggest time saver i have on our workflow