Ingmar Franz (duerer)

54 replies · posted

Backhoe Contribution - duerer - Wheels / Rear Axle

This thread is dedicated to the wheels and the rear axle of the backhoe project.

  • A reshaped part of the rear axle:

    And with subsurf turned on:


  • Current state of the gearbox:

    The following part is really tricky since I get many 5-poles (file here):

    Furthermore, the three holes aren't really circular as they should allthough I've used the "Circle" tool in the "Loop Tools " addon.


    • When this happens for me I just add in a circle and size it to the same rough dimensions and manually swap it with the non-uniform circle.

    • Thank you for the tip, blanchsb. I think the vertices adjacent to the circular holes are pulling too much due to the subsurf modifications despite a second circle around each hole as holding edge.

    • Not sure about all of the 5 sides poles. When you have that many circles close together you can work the number of circle sides to your advantage. Using a 16 sided circle versus an 8. I would say keep them to divisions of 4 in most cases. But I have seen 6 and 12 sided circles work sometimes too. You may just need 16 or more sides to get better edge flow with fewer poles possible or at least get those poles not next to each other (creating more geometry in between them).

      Also here is another easy fix. You can remove this one:

    • Thanks blanchsb , this really helps 😀 !

  • Some retopology done today:

  • Rear axle and wheel hub finally connected:

    • I have to do some retopology on the green attachment ring in the image above in order to avoid heavy distortions, but I'm ending up with six pentagons (green) and two areas with complicated topology (red):


      Does anybody know a clever solution? File is here. I escpecially want to avoid 6-poles at all and 5-poles on sharp edges. 


      PS: At least the pentagons don't cause any shading issues, but I want to avoid them nevertheless.


    • I'll take a look duerer 


    • So one comment I have right off the bat: Don't try fixing things with sub-div ON in edit mode. It's bets to clean up the raw mesh data. Hopefully you have been doing that. I just noticed you have it ON in edit mode. That can really throw people off big time. If the raw mesh is clean the sub-div will definitely reflect it.

    • Thanks blanchsb  for your help. I've found a solution. First I managed to turn the six pentagons into four quads (yellow) and two tris (green) with some edge spinning and vertex merging:

      Then, I managed to create one tri and the rest only quads in each of the red areas above so that I could connect the two new tris and the two tris created before with a loop cut that turned the four tris in total into four quads (green in the image below):

      The final all-quad-mesh can be found here.

    • The retopologized attachment ring in place:

    • Those axles and housings and hubs are the most beautifully designed pieces, but  a pita to model and they are so hidden, why is that?

      Or this gorgeous piece:

      It's even hard to get the shape right, but then the retopology...

      I haven't touched those rings yet, but your topology looks solid duerer !


    • Well crap I just recorded an amazing video of myself stumbling all over that model haha. It is still uploading. You get 16 minutes of bliss listening to my calming voice.

      Another thing is you've got 160 vertices that are duplicated on this model. I hit M + B and poof you're mesh data stayed the same but your vertex count went down by like 160 or so.

      I'll let you know when it is uploaded to youtube. If you are content after watching it and want me to delete it, just let me know.
      My advice may not be the best advice but this is my thought process working through assumed constraints on pieces I felt you didn't want altered while solving the poles issues.

    • Ok duerer here my wonderful video ha!




    • blanchsb a wonderful video indeed! Thank you Shawn!

      We should all do more videos like this...

      I will try to make one when I get to this part... But, I think you are creating a lot of unnecessary geometry: when you've got this:

      I mean this part:

      simply Edge Slide:

      and Join:

      I know I will end up with a completely different topology and now I am extra stimulated to get to this part, because I would like to see how I am going to solve this;)

    • Oh yeah look at that. haha. Yup that does it great. Thanks spikeyxxx

      I know the tricks but I don't always see them when I am in the trenches. That is great to point out. 

    • I finally got my motion capture studio set up with my new Kinect (Fathers day gift I got myself haha) So now you can see my pretty face. I still struggle not sounding monotone but a few 1000 more hours of videoing myself should solve that.

    • Thanks, blanchsb for all your efforts and the video 😀👍 !!! A calming voice is definitively what I need now the most.👂😀!

    • I've decided to model the backside of the attachment ring as the negative form of the front. Furthermore, I've found some 6-poles and some overlapping vertices where there should be connections. Finally, I used "Bridge Edge Loop" to close the gap between front and back at the ring's inner rim:


      File is here.

    • spikeyxxx, you're absolutely right about the beauty of these pieces! They're for me not only technical pieces, they're artworks🏆!

    • From this image, the attachment ring looks completely solid:

      But on this drawing, one could thinks that there're holes where you can see through:


      Or are these just cavities that are hidden from the other side?

      Spikeys first video snapshot from above suggests (especially the deepening at the image bottom) that there could be holes but I'm not completely sure:

      Maybe, the design has changed since in Spikeys screenshot the front ring seems to be one piece with the disc below. Or is this just an illusion caused by the paint layer?

      But from this section of the front axle, it seems again a continuous structure (red circle):




      And the same on the rear axle:


      So, I'll keep my model as it is with some artistic freedom.


    • blanchsb your video is really helpfull and looks as if you were already part of the CGCookie crew for years. I'm wondering when they  will offer you a permanent job in their team 😀.

  • And here the newest cookie from the retopology kitchen:

    File is here.

    And the part in its context:

  • The rear axle end housing is more complicated than I thought. Here a quick update on this work in progress:

    • It looks like a modeling cornucopia! 

    • It sometimes looks to me like a new version of "Battleship Galactica" 😉😁.

    • I finally managed to get an all-quad-retopo of my rear axle end housing but it still causes a really bad shading especially in the area where the green bolt holder is connected to the red main body:


      File is here.

    • I think that a "Topology Basecamp" here at CGCookie with training of how to clean up problematic topology (N-Gons, poles, triangles etc.)  would be a good idea. I'm thinking at something what William Vaughan does in the two volumes of his "Topology Workbook". But the examples described there are mostly with flat planes and not with closed surfaces like my rear axle end housing above.

    • By the way: Can anybody explain this in my eyes arbitrary color bleeding in the areas of the beveled edges?


      PS: I just found out that a complementary color bleeding effect with swapped colors can be achieved by calculating the normals inside.

    • @ duerer I had a quick look at your file (don't want to look at it too much, because I am at the moment also struggling with the housing and I do not want to get 'prejudiced'/torn into doing it in a certain way...) and the problems (both the shading and the random color bleeding) are caused by your use of the Bevel Modifier...

      If you apply it and look at the geometry it creates (set the distance a bit higher so you can actually see what's happening), you''ll see the mess: concave n-gons and poles to give you nightmares...No mesh like that could survive a Subsurf!

      I'd never use Bevel Weight with a delicate mesh like this and make sure I'd have full control over the topology, but of course you are free to use any method you like, but at least use more than 1 segment on the bevel (best 2, just like when you use edge loops to control the sharpness of an edge, you'll put one on each side...)

      It's a difficult part. I will post my file when I've finished the housing in my thread (with just the housing), so you can see how I did it. Maybe that will be helpful in some way.

    • Thanks, spikeyxxx for your tips. The "Bevel Modifier" is certainly not easy to use since you can easily mess up things with the wrong settings and on such a complicated part like this, it's obviously better to use holding edge loops. Changing the "Inner Miter" type from "Sharp" to "Arc" makes things even worse. But it was worth a try.  The axle end housing is by far the most difficult part that I've modeled on the rear axle.
      Another difficulty for me is to contain additional geometry as local as possible. But when it fits in one part, it often does not / no longer fit in another part. Furthermore, keeping geometry as local as possible can generate a relatively dense geometry which will be visible as sharp edges in smooth shading like scars. It's a really sensible balance that you have to keep and just a little step too far to the left or to the right can produce a really ugly result. Nonetheless, this part is an excellent opportunity for training your skills in terms of creating a clean topology and my top candidate for a future "CGCookie Topology Basecamp".

    • duerer  the Mesh Modeling Bootcamp by jlampel really helped me wrap my head around topology better. I went from bad topology to not as bad to decent. I still make mistakes a lot but doing that learning flow really helped. The other tidbit from theluthier that I am terrible at is: get the main outside edges all sorted then sort the detailed extrusions and give them good holding edges that take all of the push and pull from a Sub-D mod and then work your way connecting the two and figuring out the n-gons and poles on fairly flat areas. Hopefully that makes sense. Watching his live stream working out the rear digger arm and details was very helpful. I want to go back and watch that a few more times because I still forget to do that.

      Now granted you're and Spikey's job is an extreme situation where very little geometry isn't going to be a challenge. I almost wonder if modeling things on a flat plane and then wrapping them into circles and joining them would maybe be a clever way of working with cylindrical objects??? Or maybe I am just rambling? spikeyxxx has really good ideas when it comes to this stuff. I want to see more WIP pictures Spikey!

    • Thanks, blanchsb, for your hints! The challenge with the axles is that there'are a lot of curved and straight surfaces intersecting and flowing into each other. Adding a new part on curved  surfaces so that it doesn't distort your curvature especially in that relatively small space is very tricky and frustrating. As soon as it fits in one place it often causes problems in another place.

    • Okay, here's a new version of my rear axle end housing without "Bevel Modifier" and with holding edges instead: 

      But the shading issues due to bad topology in the red main body and the green bolt holder remain: Either I get pinching because of too dense mesh parts or  too big parts on curved surfaces cause problems. I'm somehow turning in a circle where, when I solve an issue another one  arises. File is here.

    • Hmmmm...... do you have the original booleans where it is more smooth? One idea would to implement the shrink wrap approach on the troublesome areas and move the base mesh. Kent covered this for adrian2301  during the 2nd or 3rd live stream. I think that this will probably could help solve that. 


      Or on a more manual but more controlled approach you could reference the booleans OR duplicate the red part and close it up and snap the green section to that smooth duplicate perhaps? Treat it like the Zerk fitting that Kent also made. He was using booleans for that workflow and then snapped the retopology to the simple boolean shapes (livestream 3: timestamp 2:57:00)

    • blanchsb, thanks for the tips! I've already removed all booleans and replaced them by manually modeled mesh so that I have the best control (at least theoretically if I were more skilled on that  😉 ).  The "Shrinkwrap" method is certainly worth a try allthough I fear that it's a little bit more setup work on this complicated mesh 🤪.

    • duerer I think blanchsb 's suggestion of a Shrinkwrap is definitely worth the trouble here.

      Your surface is a bit wobbly now:

      and:

      Those are the ones that immediately caught my eye, but there are probably more areas like this and trying to manually correct them will be a nightmare and the results will never be great...

    • You can append an older submission to get those booleans back if they are the same size perhaps?

    • Here's an old version of the main body of my rear axle end housing with booleans (but only those for the bolt holes):

      https://www.dropbox.com/s/bhpk4e9cvewafno/RearAxleEndHousing_OldBooleanVersion_4Juli2020.blend?dl=0


      I've also cut out and magnified a drawing of this part from an earlier post in this thread which seems to show some kind of contour lines that can maybe give a hint on how to do a proper topology in order to match the exact shape:



    • Yeah it looks like you struck gold those are nice hintful lines.

  • So, here an update on my rear axle:


  • One new part modeled (called "yoke"?) that can be seen from outside the rear axle end housing under certain angles:

    File is here.



  • Some alignments done. The most important is this one:

  • And that's the newest version of my rear axle:

    • Connecting the different parts of the gearbox especially certainly will bring a lot of fun 🤪.

    • Am I reading that correctly? 10 million faces? Jeez you are a beast Ingmar!

      I think it is really coming together quite nicely my friend!

    • Yes, and Kent definitively will like that almost 10 million faces 😉😁! The new "yoke" part has already 2640 faces. And there're two of them. Some parts are still too much subdivided where the "Subsurf Modifier" alone could do a good job (for example in the wheel hub and rim). But since this rear axle is really complicated, I don't know how low I can push this.

    • Off topic: Did they switch off here in the forum the redirection directly to a new reply? I have to search for "hours" as a workaround.