Chaos Scatter feedback

[mike288]

Hi, from the latest Corona Renderer 8 Daily Builds Changelog:

• Corona Scatter has been deprecated in favor of more advanced and standalone Chaos Scatter. Chaos Scatter is automatically installed alongside Corona. You can find it in Chaos -> ChaosScatter after installing.
• If possible, Corona Scatter button has been removed from the toolbar because Chaos Scatter now has its own toolbar.
• Legacy Corona Scatter scenes get automatically converted on scene open, the randomization may change in this case, if you need to maintain the same scattering we advise to open and render the scene in Corona v7.

Download Link: https://drive.google.com/file/d/1UbEYs_wGvGY_nkw4so69oeuam915awBD/view?usp=sharing

Please, what do you think? :-) The changes announced recently in our blog post now apply.

• Slope limiting (no trees growing out of cliffs!)
• Include/exclude via splines (closed or open)
• Surface color map (patterns on rugs, etc.)
• Better viewport display
• Limit scattering only inside camera field of view (frustum culling)
• and more

[marchik]

It seems that everything stated works fine, except

• Limit scattering only inside camera field of view (frustum culling)

I cant find this option, and unfortunately I did not wait for the function of normal orientation of objects scattered along the spline, as in the forest pack which I wrote about several years ago
https://forum.corona-renderer.com/index.php?topic=26898.msg159486#msg159486

but overall feelings and performance are great!

[mike288]

It seems that everything stated works fine, except

• Limit scattering only inside camera field of view (frustum culling)

I cant find this option, and unfortunately I did not wait for the function of normal orientation of objects scattered along the spline, as in the forest pack which I wrote about several years ago
https://forum.corona-renderer.com/index.php?topic=26898.msg159486#msg159486

Stay tuned. Frustum culling (Camera clipping) is WIP and will be added before Corona 8 release.

There is Follow Spline spinner available for 1D spline scattering, do you mean that? Anyway, ideas there are recorded in our internal tracker (thank you). The thing is that ATM we are not focusing on 1D features as other missing features compared to other scatters are more needed.

[marchik]

Stay tuned. Frustum culling (Camera clipping) is WIP and will be added before Corona 8 release.

There is Follow Spline spinner available for 1D spline scattering, do you mean that? Anyway, ideas there are recorded in our internal tracker (thank you). The thing is that ATM we are not focusing on 1D features as other missing features compared to other scatters are more needed.

Nice to hear, we will wait for all the features to be added!
As for spline scattering, personally, in my case, for large exterior projects I will most likely still use the forest pack, but for small everyday tasks, for example, curb stones or stitching on fabric, I would gladly use a faster and simpler tool like Chaos Scatter (performance is really great), ofc I remember about the follow spline option, but it seems to me that the instance orientation algorithm itself is different here, in any case I will conduct more tests for myself, and again thank you so much for the work done!

[shortcirkuit]

looks great - does anyone know how this compares to forest?  I mean, i will do tests but keen to see if anyone has done some tests.  By tests i mean, with large forest pack objects (lots of scattered items) - the interactive render takes up to 15 seconds to start in some extreme cases.

[Frood]

This one seems like an inherited defect from CoronaScatter. Trackview names are empty for ChaosScatter.



Good Luck


Edit: Sorry, this should rather go to
https://forum.corona-renderer.com/index.php?topic=35061.0

[mferster]

Is there any way to have an option to scatter instances on a model's surface post material displacement? This would be an incredibly useful feature, but not sure how complicated it would be to implement.

[TomG]

In Max this can be done by using a Displacement modifier in the stack rather than relying on displacement in the material (as that stack-based displacement is before render time, while other displacement is at/during render time, making it "impossible" to include it in the scattering in the same way it doesn't show up in the viewport).

[marchik]

In Max this can be done by using a Displacement modifier in the stack rather than relying on displacement in the material (as that stack-based displacement is before render time, while other displacement is at/during render time, making it "impossible" to include it in the scattering in the same way it doesn't show up in the viewport).

As far as I remember, even ForestPack has a "Follow Displaced Surface (Corona)" effect to do this, so I think it is not "impossible" at least

[arqrenderz]

Please add a Global Scale for the objects

[denisgo22]

please add options for areas with surface edge mode for scaling and density/
Also need adding free paint areas mode for painting include/exclude areas, not only with splines

[mike288]

This one seems like an inherited defect from CoronaScatter. Trackview names are empty for ChaosScatter.

Good Luck

Edit: Sorry, this should rather go to
https://forum.corona-renderer.com/index.php?topic=35061.0

Hi Frood, done, thanks, this will be resolved in next Scatter daily.
UPDATE: released.

[mike288]

Please add a Global Scale for the objects

Hi, what do you mean by global scale? You can scale your original model object and if you turn on Preserve model scale option it gets reflected in your scatter.

[mike288]

Anyway, thanks for your feature ideas. We are taking notes and will of course consider them for future versions.

[Majeranek]

I think that there's some problem with translation tab.
For example setting min 0cm and max 50cm on Z axis moves instanced objects just a little bit - like 0,5cm, definietly not 50cm.

[GeorgeK]

I think that there's some problem with translation tab.
For example setting min 0cm and max 50cm on Z axis moves instanced objects just a little bit - like 0,5cm, definietly not 50cm.

Hi, can you please send us a reproduction scene of the issue you are reporting, (or if possible your current scatter setup)? I am afraid I wasn't able to reproduce this.

[maru]

Keep in mind that if Z translation is set to 0 - 50 cm, then the instances will be moved randomly in this range, so:
- a tiny number of instances will be moved 0cm (or almost 0cm)
- a tiny number of instances will be moved 50cm (or almost 50cm)
- a large number of instances will be moved somewhere between 0 and 50 cm; 25 cm on average

[romullus]

I have small request. Currently include/exclude splines are projected in world space only. Please consider to also add a local (object) space option. This would be very handy to define precise inclusion/exclusion areas on more complex terrains.

[aaouviz]

Hi,

Not sure if it's a bug or an intended feature, but if I use a closed spline (surrounding the edge of a surface I'm scattering on) as the exclude spline, the entire scatter doesn't work at all.

I doubt this is intended... seeing as there is no edge texture (like FP has) this is my best shot at having sharp edges for grass scattering (please let me know if I'm missing something and there is a better workflow).

Hope this is clear and replicable.

[romullus]

I don't think that's a bug - if the spline is completely covering all the distribution are and you are using it as exclude, then obviously nothing will be scattered. Try to use spline as include, but don't expect that it will be somehow better for edges than normal scattering - it still uses distributed object's pivot to determine its placement.

[aaouviz]

But it's about distance from the spline, right?

So if I scatter on a square that's 5000x5000mm and a bordering closed spline around it as an exclude set to 100/150mm the WHOLE scatter shouldn't be disabled...

I think the proof is that if I break that same spline even by a tiny amount so it's not closed, the scatter works as expected...

[maru]

But it's about distance from the spline, right?

So if I scatter on a square that's 5000x5000mm and a bordering closed spline around it as an exclude set to 100/150mm the WHOLE scatter shouldn't be disabled...

I think the proof is that if I break that same spline even by a tiny amount so it's not closed, the scatter works as expected...

To me it sounds like the current behavior is expected, but maybe I am missing something. Could you share some screenshots / renders showing what exactly you are doing and why? That would greatly help us.

Other than that - maybe you can use the Corona Slicer Material for edge trimming? https://support.chaos.com/hc/en-us/articles/4956761514897-How-to-use-the-Corona-Slicer-Material

[Bjoershol]

Sounds like this is by design, you need to be able to use a closed spline while still being able to use distance on the outside of the spline. The solution would be to break the spline, unless I'm missing something. :)

[romullus]

But it's about distance from the spline, right?

So if I scatter on a square that's 5000x5000mm and a bordering closed spline around it as an exclude set to 100/150mm the WHOLE scatter shouldn't be disabled...

I think the proof is that if I break that same spline even by a tiny amount so it's not closed, the scatter works as expected...

That's right, but the thing is, that near/far feathering radiates outside of closed spline, not inside, so if your spline is exactly bordering your distribution mesh, then no matter what you'll do, it will always exclude everything.

As for the splitting the spline - as soon as you do that, there's no longer an inside or outside and inclusion/exclusion will propagate from the spline in all directions.

[maru]

So to me it still sounds like expected behavior, but if we can make our users' lives easier, then why not? :)
I would still appreciate taking a look at what exactly the use case is so that we could report it to the devs. Thanks!
Another option is contacting us about this at https://support.chaos.com/hc/en-us/requests/new

[romullus]

So to me it still sounds like expected behavior, but if we can make our users' lives easier, then why not? :)

Maybe you could do that near and far would take negative values as well, then user could choose in which direction to extend inclusion exclusion area. The only downside to this, is that negative values would make no sense with open splines.

[aaouviz]

Ok,

Thanks for all the input.

I guess the main takeaway here, for me, and probably other users, is that the spline include/exclude works differently if it's closed or not. This is handy to know; If it's closed, it excludes everything inside the spline, but if it's open it excludes in a feather in both directions of the spline.

I think this double situational/conditional use is a bit confusing, personally, but can live with it :)

I guess the best case use of the closed spline is if the user wants to make 'islands' of no/only scatter. Understandable and useful.

Screenshots attached to show my 3 test findings. I no longer have a particular beef with the workflow, but perhaps other users can chime in?

----

As for the clipping material for clean grass edge surfaces... ugh, I don't think this is a great scenario either. When using Corona scatter my previous best option was to use a shelled version of the scatter surface to act as a distance object for the opacity of the grass. This new suggested workflow isn't much of an improvement - I still have to use a separate object. If I edit the scatter surface this breaks the shelled version of it (I know, I know, it can be a referenced instance, but this isn't always clean/possible).

Is there a better workflow I'm unaware of for clean grass edges?

Thanks again.

[aaouviz]

So to me it still sounds like expected behavior, but if we can make our users' lives easier, then why not? :)

Maybe you could do that near and far would take negative values as well, then user could choose in which direction to extend inclusion exclusion area. The only downside to this, is that negative values would make no sense with open splines.

Or some sort of tick box 'act as closed island spline' or something like that, which is only active when using a closed spline. Sorry to be suggesting adding complexity!

[romullus]

If you want that thsi secnario would work as open spline, then simply select one vertex in that spline and break it :]

[aaouviz]

If you want that thsi secnario would work as open spline, then simply select one vertex in that spline and break it :]

Yes, thanks. I mentioned this earlier :)

[maru]

Random idea:
I think 3ds Max does not allow any out of the box way of adjusting the normals of a spline (e.g. if you extrude it along normals - there is no way to specify in which direction it should extrude - this is predefined depending on which viewport you created the spline in). Maybe "flipping normals" of a spline would work here? But how to do it is an open question. :D

[mike288]

...

As for the clipping material for clean grass edge surfaces... ugh, I don't think this is a great scenario either. When using Corona scatter my previous best option was to use a shelled version of the scatter surface to act as a distance object for the opacity of the grass. This new suggested workflow isn't much of an improvement - I still have to use a separate object. If I edit the scatter surface this breaks the shelled version of it (I know, I know, it can be a referenced instance, but this isn't always clean/possible).

Is there a better workflow I'm unaware of for clean grass edges?

Unfortunately I am not aware of any. We should do something about it. We already have this logged but thank you. :-)

[aaouviz]

Perhaps, just a little bit of friendly UI; when the user selects a spline as the object to scatter on, ChaosScatter could detect this and automatically set the scatter type to 'on spline'?

Food for thought :)

Thanks

[aaouviz]

Hi,

Another query/request.

When I use the near/far includes splines on a scatter across a surface the number of scattered items is kind of confusing/messed up.

I understand the way the scatter works it takes as sample of surface and divides it up into x number of points to place items on. So this might be a very unreasonable ask:

But if I have a big surface (say a football field) and I want to scatter an object in a small inclusion area I must make the number very very big to ensure the inclusion area is filled. So I must enter say 100,000 items, but only say 5,000 will be scattered.

If I was scattering on a surface the size of the inclusion area (but not using any inclusion spline) and I set this scatter count to 5,000, it would scatter 5,000 pieces.

Is it possible to somehow get these numbers to be somewhat more controllable/predictable?

I haven't done the best job here explaining myself, so hopefully it's understood? Cheers!

[romullus]

I think it's done this way to prevent scattered instances from randomly jumping when the user edits a spline. I'd recommend to switch from fixed number of instances to instances per area - not only you won't need to enter very big numbers, but you also will be sure that you will get consistent results even if you drastically resize scattered surface. Personally i very rarely define total number of instances, because defining density is so much more convenient.

[Aram Avetisyan]

Perhaps, just a little bit of friendly UI; when the user selects a spline as the object to scatter on, ChaosScatter could detect this and automatically set the scatter type to 'on spline'?

Food for thought :)

Thanks

Hi,

Splines can be both linear objects and surfaces to scatter on. It is better that user decides which type of scattering should happen instead of auto-detecting but then changing.
The suggestion maybe the case for closed spline. But this can lead to another issue - splines, open or closed, can be made renderable and they become spline controled surfaces and in that case scatter cannot be sure what exact scattering is needed.
And this is actually a great feature. Because...

Ok,

Thanks for all the input.

I guess the main takeaway here, for me, and probably other users, is that the spline include/exclude works differently if it's closed or not. This is handy to know; If it's closed, it excludes everything inside the spline, but if it's open it excludes in a feather in both directions of the spline.

I think this double situational/conditional use is a bit confusing, personally, but can live with it :)

I guess the best case use of the closed spline is if the user wants to make 'islands' of no/only scatter. Understandable and useful.

Screenshots attached to show my 3 test findings. I no longer have a particular beef with the workflow, but perhaps other users can chime in?

----

As for the clipping material for clean grass edge surfaces... ugh, I don't think this is a great scenario either. When using Corona scatter my previous best option was to use a shelled version of the scatter surface to act as a distance object for the opacity of the grass. This new suggested workflow isn't much of an improvement - I still have to use a separate object. If I edit the scatter surface this breaks the shelled version of it (I know, I know, it can be a referenced instance, but this isn't always clean/possible).

Is there a better workflow I'm unaware of for clean grass edges?

Thanks again.

you can make the spline - open, closed, doesn't matter - renderable, probably only in viewport and not in render.
And you will get what you need. You get double control - the physical size of the renderable spline plus the near and far from scatter parameters.
Hope this helps.

Thanks,
Aram

[tradstown]

I hope displacement follow will be added to this scatter and clusters as well.

[romullus]

I hope displacement follow will be added to this scatter and clusters as well.

Make sure to cast your vote here: https://forum.corona-renderer.com/index.php?topic=37064.0 ;]

[tradstown]

I hope displacement follow will be added to this scatter and clusters as well.

Make sure to cast your vote here: https://forum.corona-renderer.com/index.php?topic=37064.0 ;]

Thank you, already done! Fingers are crossed!

[Bjoershol]

I'm using Chaos Scatter to distribute trees on a 1800x1800m terrain surface. I have two scatter types, one for deciduous trees and one for pine trees. The two types are scattered on the same terrain, but driven by a black/white map that defines where the trees should grow according to how high up they are. I also have multiple splines set up to exclude scattering in some places.

My problem is that the scene has become incredibly slow, selecting the terrain or scatter objects and making adjustments takes so long that it's impossible to work with. Finding it hard to work if I have to disable the scatters every time I need to make a change, and simply selecting them and disabling takes a long time. Also, I'm planning on scattering other vegetation like shrubs and grass on the same terrain, but it looks like this isn't going to be possible using this technique.

I think the mask is slowing things down a bit, but the spline excludes slows even more.

Disabling the spline excludes in the list doesn't help the performance, but deleting them eliminates the issue.

Is this to be expected using spline include/exclude, or is this a bug?

[rowmanns]

I'm using Chaos Scatter to distribute trees on a 1800x1800m terrain surface. I have two scatter types, one for deciduous trees and one for pine trees. The two types are scattered on the same terrain, but driven by a black/white map that defines where the trees should grow according to how high up they are. I also have multiple splines set up to exclude scattering in some places.

My problem is that the scene has become incredibly slow, selecting the terrain or scatter objects and making adjustments takes so long that it's impossible to work with. Finding it hard to work if I have to disable the scatters every time I need to make a change, and simply selecting them and disabling takes a long time. Also, I'm planning on scattering other vegetation like shrubs and grass on the same terrain, but it looks like this isn't going to be possible using this technique.

I think the mask is slowing things down a bit, but the spline excludes slows even more.

Disabling the spline excludes in the list doesn't help the performance, but deleting them eliminates the issue.

Is this to be expected using spline include/exclude, or is this a bug?

Hey,

I'd like to investigate this issue, it sounds like something isn't quite right there. Would you be able to share your scene with us?

If so I'll PM you a link to upload a file

Thanks,

Rowan

[romullus]

@Bjoershol, you could try to disable automatic update in the scatter's display settings and see if that helps.

[Bjoershol]

I'm using Chaos Scatter to distribute trees on a 1800x1800m terrain surface. I have two scatter types, one for deciduous trees and one for pine trees. The two types are scattered on the same terrain, but driven by a black/white map that defines where the trees should grow according to how high up they are. I also have multiple splines set up to exclude scattering in some places.

My problem is that the scene has become incredibly slow, selecting the terrain or scatter objects and making adjustments takes so long that it's impossible to work with. Finding it hard to work if I have to disable the scatters every time I need to make a change, and simply selecting them and disabling takes a long time. Also, I'm planning on scattering other vegetation like shrubs and grass on the same terrain, but it looks like this isn't going to be possible using this technique.

I think the mask is slowing things down a bit, but the spline excludes slows even more.

Disabling the spline excludes in the list doesn't help the performance, but deleting them eliminates the issue.

Is this to be expected using spline include/exclude, or is this a bug?

Hey,

I'd like to investigate this issue, it sounds like something isn't quite right there. Would you be able to share your scene with us?

If so I'll PM you a link to upload a file

Thanks,

Rowan

Sure, I made a copy of my scene and I can upload it. Thanks :)

@Bjoershol, you could try to disable automatic update in the scatter's display settings and see if that helps.

Thank you romullus, that makes it easier to work with. Hitting the update button takes about 8-10 seconds, but at least the wait is confined to pushing that one button. :)

[rowmanns]

Sure, I made a copy of my scene and I can upload it. Thanks :)

You should have a link in your PMs :)

Thanks,

Rowan

[mike288]

Thank you romullus, that makes it easier to work with. Hitting the update button takes about 8-10 seconds, but at least the wait is confined to pushing that one button. :)

Hi, BTW you should be able to hit that button also from the Lister, to avoid iterating through the individual scatters in the Command panel.

[rowmanns]

@Bjoershol,

Thanks for the scene, my last question for the time being. What is the specification of your computer?

Thanks,

Rowan

[Bjoershol]

@Bjoershol,

Thanks for the scene, my last question for the time being. What is the specification of your computer?

Thanks,

Rowan

Hi, these are my specs. Let me know if you need anything else.

Intel Xeon W-1290P CPU @ 3.70GHz
NVIDIA GeForce RTX 2080 Ti
128 GB RAM

[romullus]

Maybe you could do that near and far would take negative values as well, then user could choose in which direction to extend inclusion exclusion area.

I didn't see this request acknowledged, so i'll repeat it again... just in case ;] Quite often i find situations where i would want that include/exclude spline areas could have falloff in both directions, not only outwards, like it is now.

[aaouviz]

Hi,

A quick query; is beta (or daily) version of Scatter available for our use, much like Corona is?

I see on the trello roadmap that edge trimming is done in V2... would LOVE to get my hands on this

Woah, nevermind! I see it was just released, exciting!

[tennet]

Any solution to this issue with the slow viewport performance? I'm using a "circular" Spline Mask (as 'include') to grow some trees (just 30 trees on a small area) and C4D freezes for 5-10 seconds every time I try to adjust anything while this spline mask is used in the Chaos Scatter Object. If I delete the mask, then everything runs smoothly again (and trees are growing all over).

1) Is it possible to use 'vertex maps' to distribute Scatter objects?

2) Is it possible to use 'Selection Tags' to distribute scatter objects?

3) Would be nice if it was possible to use other objects as masks (not just splines).

*** I'm using Corona 9 in C4D R21 (PC i9-9900K @3,6GHz and 64GB RAM). For the specs.

[Beanzvision]

Any solution to this issue with the slow viewport performance? I'm using a "circular" Spline Mask (as 'include') to grow some trees (just 30 trees on a small area) and C4D freezes for 5-10 seconds every time I try to adjust anything while this spline mask is used in the Chaos Scatter Object. If I delete the mask, then everything runs smoothly again (and trees are growing all over).

1) Is it possible to use 'vertex maps' to distribute Scatter objects?

2) Is it possible to use 'Selection Tags' to distribute scatter objects?

3) Would be nice if it was possible to use other objects as masks (not just splines).

*** I'm using Corona 9 in C4D R21 (PC i9-9900K @3,6GHz and 64GB RAM). For the specs.

Hi there, are you able to share a short video of this happening? My PC isn't the best but still performs ok.
(AMD Ryzen 7 2700 Eight-Core Processor 3.20 GHz & 32.0 GB)


As for your points:
1 & 2. Not currently. If there's a way to convert a vertex map to black & white values, this may work. I haven't tried it myself. For point 3, sounds like a feature request to me ;) I will make a mention of it but no promises :)

[NereuLZK]

I have realized that chaos scatter is very slow before a new release. I didn't use to it in my projects because I prefer itoo forest, but in some cases I have to use Chaos Scatter. Then, I realized that is very slow compared with itoo forest. I test a render with a simple Ivy and It is slower than a big area with grass with itoo forest for start render and in viewport. This is so bad, very very bad.

Do you have any tips?
Thanks.

[Aram Avetisyan]

I have realized that chaos scatter is very slow before a new release. I didn't use to it in my projects because I prefer itoo forest, but in some cases I have to use Chaos Scatter. Then, I realized that is very slow compared with itoo forest. I test a render with a simple Ivy and It is slower than a big area with grass with itoo forest for start render and in viewport. This is so bad, very very bad.

Do you have any tips?
Thanks.

Hi,

Can you please provide both scenes - ChaosScatter and ForestPack - for which you tell that ChaosScatter is much slower than ForestPack.
You may have set up something in ChaosScatter not optimally. In our various tests, it was at least as fast as ForestPack, if not faster.

[d4l1]

Maybe you could do that near and far would take negative values as well, then user could choose in which direction to extend inclusion exclusion area.

I didn't see this request acknowledged, so i'll repeat it again... just in case ;] Quite often i find situations where i would want that include/exclude spline areas could have falloff in both directions, not only outwards, like it is now.

I'd like that too! :)

[romullus]

Some observations after briefly playing with new Chaos Scatter features.

Group/hierarchy scattering - for some reason groups' frequency values gets truncated in distributed objects list, even though there seems to be enough space for its display in full. Could it be fixed, please? Also if possible, it would be nice that group's swatch colour would be taken from its first member, instead of leaving it black. I think it's standard 3ds Max behaviour.

Slope / Altitude - it would be nice if those limits could have adjustable falloff, ideally the one which could be mapped.

Look at - not sure if that's a bug, but its falloff distance is not actually a falloff, but rather a cut-off , i.e. inside its radius look at effect is fully on and outside it's completely off. Could we have an actual fallof here with near and far values, please?

Another look at feature request - i think this feature would even more useful if we could define multiple look at targets, each with their own influence and falloff radii.

Overall, i'm very happy to finally see some new Scatter features!

[Aram Avetisyan]

Thanks for the feedback, Romullus. And sorry for the delay.

Things are still being worked on, I will pass your findings to the devs team for now.