Tile 1x1 Round

[Orion Pobursky]

I'm sure I missed some sort of relevant discussion but why are the top edges rounded on the 1x1 round tile but nowhere else? It seems to me that these edges should be sharp just like every other part.

[Max Martin Richter]

I'm sure I missed some sort of relevant discussion but why are the top edges rounded on the 1x1 round tile but nowhere else? It seems to me that these edges should be sharp just like every other part.

The first version was uploaded with sharp edges. We decided to round the edges as these are much more rounded at this tile than on any other tile.

Just put a real 2 x 2 round tile and 1 x 1 round tile next to each other and compare.

[Travis Cobbs]

The first version was uploaded with sharp edges. We decided to round the edges as these are much more rounded at this tile than on any other tile.

Just put a real 2 x 2 round tile and 1 x 1 round tile next to each other and compare.

I'm going to have to say that you're wrong about that. To the the best of my ability to discern, the edges have exactly the same bevel on the 1x1 as on the 2x2 (along with a number of others):



Furthermore, even if they aren't exactly the same, they are certainly extremely close. Of course, the apparent size of the bevel is higher on the 1x1 tile due to the fact that you have to zoom in by a factor of 2 for it to have the same size top surface. But I don't think that should be a consideration in LDraw parts.

[Magnus Forsberg]

The first version was uploaded with sharp edges. We decided to round the edges as these are much more rounded at this tile than on any other tile.

Just put a real 2 x 2 round tile and 1 x 1 round tile next to each other and compare.

I'm sorry Max, but the first version was made with a larger radius, 1 ldu (left).
When I made the half-round tile, 24246, it was decided that it should have a transition from a sharp to a smoother, rounded edge.
At the same time the 1x1 round tile was changed to have a 0.5 ldu radius (right).
The chamfered version in the middle was rejected.

[Orion Pobursky]

I'm sorry Max, but the first version was made with a larger radius, 1 ldu (left).
When I made the half-round tile, 24246, it was decided that it should have a transition from a sharp to a smoother, rounded edge.
At the same time the 1x1 round tile was changed to have a 0.5 ldu radius (right).
The chamfered version in the middle was rejected.

For the more part, we don't round the edges of studs or other round, flat surfaces. I don't think we should here either.

[Travis Cobbs]

For the more part, we don't round the edges of studs or other round, flat surfaces. I don't think we should here either.

I agree. I don't see how this part is any different from any of the other round parts; as my photo hopefully shows, the size of the bevel is pretty much the same on all of them. The only difference is that when you zoom in to have it appear a given size on-screen, it's more obvious. However, I don't think LDraw parts should be designed based on how they will appear when they take up a set number of pixels, but instead based on the size of the real-life geometry. And in real life, this bevel is extremely small.

[Magnus Forsberg]

How is this beautification different from the recent changes added to 54200 and 85984?

[Steffen]

I also wondered about that change.

We should either beautify all parts similarly or none, but not just some IMHO.

[Steffen]

I agree with Orion and Travis.

In the photo, I can see no difference in "smoothness" between the 1x1 and 2x2 round.

I think currently we have an inconsistency in this matter in the library which we should resolve in this or the other way.

Either make all parts "smoother" that are,
or remove the "extra smoothness" from the few selected ones that currently received that overhaul.

[Philippe Hurbain]

How is this beautification different from the recent changes added to 54200 and 85984?

The bevel radius is much larger, 2.5ldu. Quite noticeable.

[Jaco van der Molen]

I agree with Orion and Travis.

remove the "extra smoothness" from the few selected ones that currently received that overhaul.

+1

IMHO the bevel of the current 1x1 round tile is too much.
Woul adjusting this affect the patterned 1x1 round tiles?

[Gerald Lasser]

my heart says "no" to the removal of the rounded edge, as it simply looks good, but I think the more pragmatic decision it do do it away.
It greatly reduces surfaces to render.

May be we need to cut one in half to compare the radii better.


Concerning the patterns, I would keep the patternable area with a 9 LDU radius as it is now.

[Philippe Hurbain]

+1

IMHO the bevel of the current 1x1 round tile is too much.
Woul adjusting this affect the patterned 1x1 round tiles?

No, we just need to adjust the subpart. This will nonetheless cost a ring, not really needed otherwise.
I also favor the simpler sharp edge.

[Magnus Forsberg]

No, we just need to adjust the subpart. This will nonetheless cost a ring, not really needed otherwise.
I also favor the simpler sharp edge.

This is my part/error. I'll do all the needed changes later today. This is not the first time I have to change it.

[Jaco van der Molen]

How is this beautification different from the recent changes added to 54200 and 85984?

A new reply to this old thread:

Some patterned 85984 have sharp edges, while some use the beautified version?

Sharp:
85984.dat
85984p70.dat
85984pc1.dat
85984pn1.dat
85984d50.dat
85984p40252-3.dat
85984p40252-1.dat
85984ps1.dat

Beautified:
85984d02.dat
85984d01.dat
85984pz0.dat

Is there a reason or do the sharp-edged versions need to be updated?

[Jaco van der Molen]

I'm sure I missed some sort of relevant discussion but why are the top edges rounded on the 1x1 round tile but nowhere else? It seems to me that these edges should be sharp just like every other part.

The real parts do have this bevel, but my issue is the way these parts render in LDCad and LPub3D.

Perhaps we over-beautified them? Sure, they are not sharp, but do we really need that extra smoothness?
I guess only if you do a super hi-res photo-realistic render, but otherwise, it is not really needed IMHO.
I have a feeling normal straight tiles have a bit of a bevel and we do not have that in the LDraw parts either.

Currently, I am making a rather big model that uses some of these round tiles.
IMHO they render awful in LPpub3D using the native renderer.
See this example, especially when using white parts on a white page and the way I set up rendering with black outlines for white parts and white outlines for black parts:

Native renderer in LPub3D
   

As you can see in the image the white 1x1 round tile does not really stand out.
So does the 1x1 tile with rounded end.
Where I am not exactly sure if this is typical and due to the native renderer of LPub3D.

LDView does a bit better job, but still "missing" some edges:

   

And, if I may say so, the render in LDCad is not that good too.


  Naamloos-1.png (Size: 135.34 KB / Downloads: 45)

I especially dislike the somewhat strange shadow effect on the 1x1 tile with rounded end.

I know better to complain about parts because I know how much work and effort goes into making these, but can we do something about it?

Does anyone share my opinion that the renders are not that good?

Can we achieve a rendering more like LEGO does draw these parts, like for example on page 81 of the instructions of 31065-1: Park Street Townhouse (see an excerpt attached).

[Orion Pobursky]

The fix is on the PT. All the patterns should auto update when the underlying subpart updates. 

I do agree about the rendering. However, some of the lighting artifacts you're seeing are just quirks of the rendering engine.

[Jaco van der Molen]

my heart says "no" to the removal of the rounded edge, as it simply looks good, but I think the more pragmatic decision it do do it away.
It greatly reduces surfaces to render.

May be we need to cut one in half to compare the radii better.


Concerning the patterns, I would keep the patternable area with a 9 LDU radius as it is now.

It only looks good if rendered big and hi-res.

Can't we just have 2 versions?
One with straight edge for modelling and instructions (LPub3D native or LDView render).
One with the bevel to render hi-res and photo-realistic.

Or is that overdoing things?

P.S. I have yet to test how Studio renders these parts.

[Jaco van der Molen]

The fix is on the PT. All the patterns should auto update when the underlying subpart updates. 

I do agree about the rendering. However, some of the lighting artifacts you're seeing are just quirks of the rendering engine.

Thanks. I'll try the unofficial parts.

[Jaco van der Molen]

P.S. I have yet to test how Studio renders these parts.

Studio doesn't do a better job either in modeling and step editor.
Instructions look better and very close to what I want in LPub3D
Again, especially when using white parts.

   
   
     

[Jaco van der Molen]

Thanks. I'll try the unofficial parts.

Better! Only now I seem to miss the vertical edge?

[Orion Pobursky]

Better! Only now I seem to miss the vertical edge?

Since it’s different for both rendering engines, I’d say that this is a engine problem and not a problem with the part itself.

[Jaco van der Molen]

Since it’s different for both rendering engines, I’d say that this is a engine problem and not a problem with the part itself.

Yes this is probably due to the native renderer from LPub3D.

[Philippe Hurbain]

Yes this is probably due to the native renderer from LPub3D.

Yes it is. Native renderer, based on LeoCad, doesn't render condlines.

[Roland Melkert]

And, if I may say so, the render in LDCad is not that good too.

This is a real nasty problem, first some background:

LDraw data has no normals so the renderer has to 'invent' them. LDCad does this based on type 2 lines. It is averaging all normals at positions not divided by a type 2 line.

The problem with this part is it has no type 2 lines at the top corners (as seen on your image) so it averages all normals on that position causing a ~45 degree normal at the corner, resulting in the shaded effect (it is trying to make it round).

I'm aware of this problem for awhile, but still not sure on how to solve it with just the LDraw data.

I might resort to a new shadow library meta/hint for this.

I'm also open to suggestions on how to improve this.

[Magnus Forsberg]

As the author of both these parts I must say that this is a non-issue. The subfiles have been changed and the rounded "edge" is removed.
Please download and use the unofficial subfiles if you want better renditions.

But from a technical point of view it might still be interesting.