Giona's "Dream Bike Build" WIP

[Giona]

I think it's turbosmooth modifier ontop of unwrap what's causing problems. Apply small checker texture on the model and you should see distortions along the seams.

Ok, I'll try later. Should I add more loops in the geometry close to the welded areas to prevent deformations caused my turbosmooth?

How many hours of not riding your bike? :)

Too many!! ahah

Love Salsa bikes but the only Salsa thing I've ever owned were some red salsa bar grips (best grips ever)

loving the renders too Giona!

Thanks Peter!
If you're still into biking, consider a Salsa Fargo, with the proper tyres you can ride almost anywhere with it.

[romullus]

Ok, I'll try later. Should I add more loops in the geometry close to the welded areas to prevent deformations caused my turbosmooth?

Yes, additional loops would help to lessen artifacts, but never eliminate them completely. What you could try, is to replace TS with opensubdiv modifier and try differen UV interpolation methods, some of them are much better than the one that turbosmooth uses.

[Giona]

Little update about the frame.
Now is welded :)
I used Substance Painter to do it and I'm starting to love that software..

The unwrap is still not perfect, but I'm quite satisfied since I never unwrapped a complex object like this. On top of the unwrap now I used the OpenSubdiv modifier as suggested by Romullus.

In attachment you can see some screenshot of the uv problems.
Maybe I need to use more cuts to flatten the geometry better?

Some test renders. The shaders are not detailed, I'll improve them once the bike is finished.

[Juraj]

Are you sure you didn't complicate your life needlessly? I can't really zoom into the detail of the photo of the frame, but in my opinion the crown tube in reality doesn't feature any shaping where it connects to bottom frame tube.
I.E, this section didn't need to connect as single mesh and didn't need to be unwrapped together. The crown tube, and both bottom and top frame tube could have been detached meshes with simple tubular unwrap.

But I can be wrong, it's hard to tell from photo. But in render, it creates incorrect highlights. The highlights from bottom frame tube, extend onto crown tube, visible on first image.
Should it be like that?

[Giona]

Are you sure you didn't complicate your life needlessly? I can't really zoom into the detail of the photo of the frame, but in my opinion the crown tube in reality doesn't feature any shaping where it connects to bottom frame tube.
I.E, this section didn't need to connect as single mesh and didn't need to be unwrapped together. The crown tube, and both bottom and top frame tube could have been detached meshes with simple tubular unwrap.

But I can be wrong, it's hard to tell from photo. But in render, it creates incorrect highlights. The highlights from bottom frame tube, extend onto crown tube, visible on first image.
Should it be like that?

I Think you're right. To fix the highlights distorsion I have to modify the geometry in order to keep the head tube straight in the intersection with the other two tubes.
A bit of chamfer beetween the tubes must be modeled anyway, but the one on the head tube is a bit too much at the moment.

I'll post some update later.

Thanks for the comment!

[Juraj]

A bit of chamfer beetween the tubes must be modeled anyway

But why :-) ? Not nitpicking hah! Curious. I don't think there should be any chamfer between tubing themselves.

Would there be a chamfer in reality? Unless the welds are filled down, but here are full welds. There should be tiny chamfer between the welds and tubing, but that can be done by shader with CoronaEdge map (or more powerfully with mesh blended normals, but I don't know how that is achieved in 3dsMax...)

[Giona]

A bit of chamfer beetween the tubes must be modeled anyway

But why :-) ? Not nitpicking hah! Curious. I don't think there should be any chamfer between tubing themselves.

Would there be a chamfer in reality? Unless the welds are filled down, but here are full welds. There should be tiny chamfer between the welds and tubing, but that can be done by shader with CoronaEdge map (or more powerfully with mesh blended normals, but I don't know how that is achieved in 3dsMax...)

No problem, is better if we discuss about it, maybe there's a smarter way to do that effect :)

In attachment you can see a photo of the real weld. I'm trying to simulate that "chamfer".
I did a test without any chamfering, just cylinders intersected, and it works only if I use the corona round edges and as additional bump the corona normal generated by substance painter.
But I have to decrease the amount of the normal map a lot, otherwhise the effect looks wrong and the line in the middle appears again.

In the second photo is visible that the weld looks also with a bulge effect.

My first thought was to model the transition beetween the tubes to show that.

Let me know your thoughts!

[Juraj]

Ok now I see, this is just miscommunication :- ). I didn't consider welds (unless filed down) to be a chamfer in this situation.
What you have looks already amazing and works.

Bit of digressing: Reality vs approach I would have tried:

- Reality: Tubes are just connected by barely connecting, weld is placed on top of the connection. Between tubes and welds is "chamfer", the tiny sub-milimeter connection of weld to the tube.

- I would model the same big "chamfer" you did, but only temporary as help. I would detach it, this will become the weld.
- I would revert the tubing back to just tubes with simple cylinder unwrap.
- Detached "chamfer" would become displaced geometry (in geo by subdiv or render time disp, doesn't matter)
- CoronaEdge map would be used to connect the displaced weld to the tubes, this is the last micro connection so the weld isn't just sitting/intersecting on top of tube. This effect is less than 1 milimeter.
  (Although blended normals on chamfer would be more accurate, as the chamfer is often changing in size to almost microscopic)

After writing this down, I realize that while this will make unwrapping easier...it might make much more work to blend normals. So my suggestion is kind of shit.

Your solution on other hand is pretty smart and effective, and would much nicely transition to Unreal for example. And you would have to use your technique for hydroformed aluminium frames where tubes are preshaped around connections as well often.

[Giona]

Ok now I see, this is just miscommunication :- ). I didn't consider welds (unless filed down) to be a chamfer in this situation.
What you have looks already amazing and works.

Bit of digressing: Reality vs approach I would have tried:

- Reality: Tubes are just connected by barely connecting, weld is placed on top of the connection. Between tubes and welds is "chamfer", the tiny sub-milimeter connection of weld to the tube.

- I would model the same big "chamfer" you did, but only temporary as help. I would detach it, this will become the weld.
- I would revert the tubing back to just tubes with simple cylinder unwrap.
- Detached "chamfer" would become displaced geometry (in geo by subdiv or render time disp, doesn't matter)
- CoronaEdge map would be used to connect the displaced weld to the tubes, this is the last micro connection so the weld isn't just sitting/intersecting on top of tube. This effect is less than 1 milimeter.
  (Although blended normals on chamfer would be more accurate, as the chamfer is often changing in size to almost microscopic)

After writing this down, I realize that while this will make unwrapping easier...it might make much more work to blend normals. So my suggestion is kind of shit.

Your solution on other hand is pretty smart and effective, and would much nicely transition to Unreal for example. And you would have to use your technique for hydroformed aluminium frames where tubes are preshaped around connections as well often.

I will try your suggestion to see if it will improve the visual. It's an approach similar to the one I tried in the beginning with the script that makes geometry around the intersected tubes.
So practically I should use the "detached chamfer" with a displacement map that replicate the brush texture I used in Substance Painter. Then the CoronaEdge map will take care of the micro connection.

When you talk about blended normals on chamfer, what are you referring to?

[Juraj]

Like this, but I don't know how it's done in 3dsMax. Perhaps with script like this: http://www.scriptspot.com/3ds-max/scripts/average-copy-border-vertex-normals

(The picture shows two separate meshes, there is no geometry change or chamfer, connection,etc..)

[Giona]

Like this, but I don't know how it's done in 3dsMax. Perhaps with script like this: http://www.scriptspot.com/3ds-max/scripts/average-copy-border-vertex-normals

(The picture shows two separate meshes, there is no geometry change or chamfer, connection,etc..)

That's interesting, but seems a bit too much work.

Today I'm going to model the wheels. Starting from the disk brake attachment on the hubs.
On the disk I added the dirt left by the brake pads on the outer ring.
To do that I made a custom texture in substance designer. I attached the screenshot of the graph if you're interested.

[Naxos]

Great job, indeed !

[Giona]

The wheels are almost finished.
For the tyres I used a displacement map generated in Subtstance Designer with the correct patter of the Vittoria Mezcal tyre. I'll add later more details, logos and name on the tanwall side.

My background is architectural visualization, so I'd like to know if is a correct approach to use a displacement map for tyres or if should be better to model them. I hope not, it would be a nightmare for me :)

[Giona]

After a few days of stop, I'm back modeling the last components of the bike.

Since I'm not an expert in modeling, I'd like to know if there's a way to optimize and speed up the process of modeling the cassette:

Each cog has a different amount of teeth, so I'm starting from a cylinder with the same amount of segments as teeth, in order to model the first one and then clone it.
The problem is that to add some final details, I need the entire cog, and it can be really slow for the bigger cogs.
I need to practice to see if I can add all the details I want into one tooth and then clone it.

Any suggestions? How would you approach this kind of modeling?

Thanks!

[romullus]

I would try to do this as procedurally as i can. Something like this.