Chaos Corona Forum
Chaos Corona for 3ds Max => [Max] I need help! => Topic started by: build on 2017-04-25, 09:34:04
-
Hi.
With the new Version of Corona there seems to be a different in the handling of edges of geometries. Before I could apply the "round edges modifier" to separated objects with the same material. As long as they where connected the edges where never shown. But now this behavior changed. Even when I attach different geometries to one object the former edges still render as outer edges. Inserting new edges to an object still works okay.
Thought there may be a new setting inside the round edges modifier but still looks like the old one. Maybe I am missing something?
Tim
-
Hi, I don't think that the behavior has changed between Corona 1.6 and 1.5.2. Please see my examples below.
If I don't understand the issue correctly, please explain how to test it better.
You may also need to adjust the settings inside the rounded edges map itself, the ui is the same in 1.6 and 1.5.2 - basically you can tell this map to affect only the same object (other objects intersecting with it will not create rounded corners effect), same material only (objects with other materials will not create the rounded corners effect, but other objects with the same material will), or use exclude list to specify the objects. Screenshot attached.
-
I rebuild your scene and that works fine. The different between 1.6 and 1.5.2 shows if you put two geometries next to each other. This behavior changed. Now you have an additional edge there which just wasn't in 1.5.2.
Tim
-
I rebuild your scene and that works fine. The different between 1.6 and 1.5.2 shows if you put two geometries next to each other. This behavior changed. Now you have an additional edge there which just wasn't in 1.5.2.
Tim
(http://i.imgur.com/8KjDN9c.jpg)
The objects are touching and both have faces on the inside, so it makes sense that it would have edges rounded there. If you rounded the edges with geometry, it would do the same as your seeing. If you don't want an edge down the middle you would remove the faces that are touching.
-
Sure. I agree that this is logical. But before corona was not just logical, it was smart. It recognized that there are 2 objects sharing the same material, touching each other and therefor didn't add an edge there. Maybe this was a bug in all version before 1.6. But a useful one, in my opinion.
-
Maybe this was a bug in all version before 1.6. But a useful one, in my opinion.
I agree with Build. Also old scenes now render differently which is a problem if you need to go back to the old project and make some changes.
Is it maybe possible to make thick box or something so users can choose between different type of round edges behaviour ?
-
I agree. Deleting the inner faces at junctions like this is not intuitive, and will add considerable time to our modeling process. It breaks our old scenes, and will add to the work we need to do when a design changes.
This doesn't help me do anything I couldn't do before, requires changes to our workflow, and breaks our old scenes. Sounds like a bug to me.
-
I agree that the previous behaviour was a lot more "it just works". Cleaning up CAD data is time consuming as it is and this is a step in the wrong direction.
-
Well, I do not agree. I find this behavior more logical and "straight forward" doing as expected. It might mean more dean up in some situations, but in the same time it gives you more possibilities in other situations.
-
(https://forum.corona-renderer.com/index.php?action=dlattach;topic=15892.0;attach=63402;image)
Effect includes/excludes.
"Same object only" i would assume gave the edge down the middle, like it is 2 objects and you wanted the edge.
"Same material only" i would assume gave no edge down the middle, like it is 1 object merged together.
^seems like it already is a tick, thats possibly not working right?
-
This is what I'm getting.
Not sure anymore is it a bug, feature or am I doing something wrong :)
-
It works in the same way as if you put actual physical chamfer modifier on top of it. If you did that to your current setup, you would need to attach and connect those meshes too.
-
I don't think that this is actually the point anymore. It is not about why it is like this. It is about what is useful and what not. As other mentioned before, the new behavior of the edge modifier breaks older scences. Also it requires additional modelling / cleaning work.
For me the old behavior was just right. I never had any problems with my modelling workflow or rendering. Now I have to spend additional time to check the model, clean it or render again, if an inner face was overseen. Corona Renderer is about simplicity and speed. But the new handling of edges is not. An additional checkbox like "exclude inner faces" or something like this would be welcome. Then everybody could use it or leave it as he/she wishes.
-
I vote for the old method, More simple, you dont need to enter the round edges map to select the mode that will work here..
-
It works in the same way as if you put actual physical chamfer modifier on top of it. If you did that to your current setup, you would need to attach and connect those meshes too.
It doesn't though, if you see the examples posted by maru, 2 different objects create a round corner where they meet, this is not how a chamfer modifier would work, but it is incredibly helpful.
In that case, Corona is smart enough to recognize that the 2 objects have the same material and are intersecting, so it assumes they are the same object.
I think the same assumption can be made for 2 objects with the same material that have absolutely no gap between them, or even 2 objects that intersect but share a face.
I think this image shows that this is indeed a bug, look at the 4th and 5th examples, if Corona is treating the 6th example as a single object, it should do the same with all of the others IMO.
Also notice what happens when I add the chamfer modifier to the second image, the images don't match.
If you are used to modeling in Polygons, this is not an issue, as you have absolute control and you probably wouldn't build your scene like that, but working with CAD data is very different and this "bug" is a problem for imported geometry.
-
The older handling of this is better I agree.
When my office models, we do not connect geometry for many elements (walls, mullions, etc) as during design things are constantly changing/moving. Such elements are left as extrusions that intersect. With the new behavior we would have to change our modeling approach and redo everything from scratch time and time again versus leaving everything as it was. Very annoying.
-
Should this thread be moved to bug reporting?
-
I don't think so. However, if you want round edges behaviour to be changed, you can open new feature request.
-
I also noticed some odd behavior with rounded edges in 1.6.
There are cases where I can't/wont join meshes together but I want rounded corners just on the visible edges only - this worked perfectly before 1.6 but now it doesn't. I have a visible seam on objects touching each other no matter what rounded edge options I select.
To me the rounded edges feature was about simplicity; being able to knock up a super simple model without having to worry about supporting loops/polygon flow to get nice chamfered corners.
-
Well, if the different options doesn't make a difference (if they all give the same result), it must be a bug?
-
Could it be that you simply don't understand how it works?
-
I understand perfectly well how it's supposed to work. I haven't used 1.6 in production yet (I'm wating for confirmation that there aren't any major bugs), I'm going by what has been said in this thread. Someone posted an example showing that the different options gave the exact same result.
-
Well, as you can see, the difference is there. It works as intended. As i said before, if someone doesn't like current behaviour, they should open new feature request and not a bug report.
-
True, it doesn't work like chamfer if you have overlapping faces, but if you have overlapping faces anywhere in your scene, then you have a much bigger problem than rounded edges not working the way you want them.
-
Well, as you can see, the difference is there. It works as intended. As i said before, if someone doesn't like current behaviour, they should open new feature request and not a bug report.
If something worked one way, and no one complained. Then it was changed and people complained, and 0 people are happier with the newer way, wouldn't that make it a bug?
I understand that this doesn't affect your or Rawalanche's workflow, but it clearly breaks many peoples'.
I used overlapping faces as an extreme scenario to demonstrate that the system doesn't make sense, you can't argue that it is correct for corona to add round corners in the middle of an object, regardless of that not being a "correct" workflow in your mind. The real case scenario is 2 objects touching, 0 gap, 0 overlap. What we are arguing is that it would be better for Corona to treat THAT case as if both objects are one, the way it used to handle it, and EVERYONE was happy with.
-
same problema here, would like much more how worked in the old builds...
-
@lolec - IIRC there has been some request to make round edges work exactly as it does know. I think one of the requested scenarios was Windows, where the inner and outer frame was seperate elements in the same object, and the wanted round edge behavior was exactly as it is now.
But hey - that might be from some of those users you're counting as 0....
-
@lolec - IIRC there has been some request to make round edges work exactly as it does know. I think one of the requested scenarios was Windows, where the inner and outer frame was seperate elements in the same object, and the wanted round edge behavior was exactly as it is now.
But hey - that might be from some of those users you're counting as 0....
I'm not counting anyone as 0, I searched and couldn't find anyone complaining about the old behavior, that's why I said that. If there is a case to be made for this to work both ways, I think the sensible way forward would be to add a checkbox maybe?
I don't think "This is the expected behavior" or "This is the most logical way" are good arguments, because there are so many different workflows and uses for corona, that what is expected or logical for someone doesn't make sense for someone else.
The window argument you brought to the conversation is 100% valid, as it reflects an actual benefit and not just a conceptual rightness.
It's funny how you try to make me look like an asshole for dismissing people, it's not a personal attack on anyone to give my opinion on this feature change. It's just that the way this thing used to work was very VERY useful for many people, I think that deserves to be said.
-
I just tried to get the new rounded edges to work and for my/our purposes it just doesn't. For example, I'm working on a hospital, 15 floors with 1000's of objects straight from revit (that sit flush against eachother), are you telling me that going through that model, merging objects and removing interior faces by hand to be able to use rounded edges is a better workflow than before? In 1.5 it just works.
That the new rounded edges behaves like the chamfer modifier doesn't make any sense. When I chamfer stuff I don't select all border edges and apply chamfer, I chamfer edges that needs to be chamfered, ie edges on the actual border of the form. And again, in 1.5 it just works.
I'm sorry, but you just made rounded edges non-usable for me and a lot of users.
-
+1 for at least the option to have the behaviour like it was in 1.5.
And would go a step further and say it should be the default. This way old scenes (we have multiple projects which are running 1-2 years or even longer) aren´t broken.
While with other changes in the renderer everyone was eager to maintain compatibility as far as possible, this one comes along quite anoying. I don´t get it. This issue alone is a reason for a v1.6.1 for me personally.
This is what it´s all about (https://forum.corona-renderer.com/index.php?action=dlattach;topic=15892.0;attach=63420;image). Reworking scenes with such elements is hell of a work and you have to drop all parametric parts usually. It seems in fact less work to revert to 1.5 when picking up a long run project than rework them.
Apart from this I agree to others claiming that usual workflows (and no, I don´t mean modelling Neuschwanstein Castle from a box with edit poly in half an hour like Rawa does) are also broken.
Good Luck
-
We do need this behaviour updated so that if you have a scene saved in 1.5 some kind of "legacy" checkbox in RoundedCorners will be enabled, so that images render the same. We've just re-rendered some 1.5 scenes in 1.6 and have all sorts of issues with this. Yes the geometry shouldn't in an ideal world be modeled this way but as someone said earlier, when the design is changing almost daily sometimes it's best to keep this sort of overlapping faces for flexibility in re-modeling....
-
If something worked one way, and no one complained. Then it was changed and people complained, and 0 people are happier with the newer way, wouldn't that make it a bug?
Usually people are much more vocal when they are not satisfied than when they are. I already got some people telling me in private that they like the way way more, asking me not to change the behavior back.
I can see reasons for both ways of doing it, perhaps a switch would be solution
-
I can see reasons for both ways of doing it, perhaps a switch would be solution
Yes please, that would be fantastic. I'm sure the new way is good in a lot of situations, but when it doesn't work it's a dealbreaker. I can't use it for the things I need to do.
-
I mean there's no doubt at all that the new way is better and correct and should be maintained. But the issue is rendering old scenes from 1.5 in 1.6 causing differences. I reckon a legacy checkbox that is enabled by default (that's important) on 1.5.x scenes is the way to go. This way new scene = correct+better, old scene = identical image result.
-
perhaps a switch would be solution
That would be great, thanks in advance!
Both behaviours definitely have its own advantages.
-
Yes. I think a switch would be a perfect solution to support people with different workflows.
-
Yes. I think a switch would be a perfect solution to support people with different workflows.
It's already there - checkbox legacy behaviour
-
oh! I missed the hotfix. Thank you.