Count Rate % - Chaos count % settings of distributor

[mattpesquet]

Hi all,

This is my first post on the board, but i have been playing around with Growfx since few years (i didnt manage to get enough time practicing as i would have wanted, but now i have more free time and i want to go next level).

First i want to congrat peoples developping this master piece, great work Eduard!

 

I m trying to find the best workflow to create the most realistic trees, talking about their shapes and branching architecture (that information itself for each species is already kinda complex and hard to spot on with real data references sometimes, also it might be changing during successive ages regarding the same species, that is one of the point that makes the job hard to get really natural shapes based on tree phyllotaxy).

 

 

Here is my specific problem:

One of the most important parameters in physical and natural architecture of a tree, is the capability to give birth to a branch in each place where a leaf appeared, or not (i mean if we say each starter of a leaf on a tree is a "node", each branch of the tree will grow from a node, but not all the nodes at all will necessarily give birth to a branch, at least a growing one, and that happens totally randomly even if its linked to few general laws, for examples most of the time the nodes near the top of the  branches are the ones the most supposed to give birth to new branches, etc.)

 

So to get that random aspect (with affects and modifiers driving the fine tunning of "species" laws), the most easy way to me with growfx should be to adjust the count rate % of the distributor, or the Chaos count rate % of the 'distributor in point ' settings panel (i get what might be the difference between both of them, i mean for the 'distributor in point' setting the random% is applied on each nodes independantly, and the Count Rate % should be on the whole amount of "children").

Actually, adjusting those settings doesnt get a linear rate %, it more or less totally jumps from 100 % distributor visible to 70% hidden, to 100% hidden, even if i input whatever values between 0-100.

i m not sure if you see what i mean, for example lets say we have 1 Pathdistributor that generates 20 "children" on the parent path, if i want 30% of the children to be off randomly, putting 30% to Count Rate % will hide almost all children.

actually 98-100% values at count rate gives 100% visible

65-98% gives like 80% hidden

64% gives sudenly 100% visible

60-63% gives 10% visible

27-60% gives 5% visible

<27% gives 0 % visible

 

The exactly same problem occurs with distributor in point Chaos count rate %.

I think those settings are not working as they should, the random% algo behind is kinda weird, i guess it is not the most difficult thing to code.

It would be heaven to have it working linearly but also with some random settings options (like % ranges regarding various affects), and ability to use affects with those settings, for example path position/lenght path affects, if i want branches under 1m to generate branches at 100% children nodes and branches over 1m lenght generates only 30% children nodes.

 

 

I hope Eduard will come for rescue !

 

 

 

Some old first attempts renders linked as a preview.

 

I have various old wips, and actual ones where i have few issues like this, i should come back with lot of more advanced questions and examples !! 

Should i start a new post for each question or can i start a kind of Work in Progress post, with all the questions there ?

 

Thanks

 

[ptrain03]

I think the chaos% and count rate% are based on the density within a base unit set in the preferences (default every 10 inches), not for the entire tree. But I could be totally wrong about this. 

 

It sounds like you could achieve the effects of node proliferation or death by affecting the density with the path position.

[mattpesquet]

I think the chaos% and count rate% are based on the density within a base unit set in the preferences (default every 10 inches), not for the entire tree. But I could be totally wrong about this. 

 

It sounds like you could achieve the effects of node proliferation or death by affecting the density with the path position.

 

thank you for your reply ptrain, i was sure someone would suggest to use density settings, unfortunatly it doesnt drive the same effect at all as what i m talking about in my post.

I already make extensive use of density curves, it can at least fake the fade effect of twigs from the end to the start of a branch for example, but it just gradiently change the space between each nodes (according to custom density curve too), but it doesnt help to make a % value of the generated paths to be "invisible", "ungenerated" whatever you want to call it, and it definitly doesnt change the tree shape as it should.

 

Once again i m working on trying to use real data and research about trees architecture and phyllotaxy, i will have to share more about my process actually and the kind of data i try to translate with growfx, but the goal is to produce the most realistic trees species regarding their shapes at various ages too (not talking about 500 millions faces models, but the right shape), ages categories give different architectural structures and branching patterns for most species, a young tree doesnt have the same growing architecture and branching pattenr model than a mature tree, or later a "dying" tree.

The "dying" time is on a long periode of time at human life scale, so in real life we see lot of post mature shapes that we identify as being caracteristical for a given species.

 

i already know about start, end settings too, but i m just talking about the random % working nicely (for count rate and chaos count, maybe with the ability to choose if the algo works on the whole tree, or per "children groups", i could expand more the idea later), giving a true random distribution of the % we want on a given path category.

 

The best would be to be able to add a new 'bias' setting to be able to kinda compress the effect near the start of the branch or at the end.

For example, i want 30% of my path02 elements (if path02 elements are generated by a path distribution on path01) do be visible only, and i want it to be driven by a random value for positions, and a bias value (lets  say bias 0.0 = more random effect near 0% path position, bias 0.5= random effect on the whole path, bias 1 = random effect near 100% path position)

 

Eduard, can you give your point of view, do you think count rate% and chaos count% are working fine ? basically, i think there might be some code issues around the random functions, and i also think the kind of options i m talking about should be easily implemented, and would give a much more deep power to make realistic branching patterns (we could talk more about dead branches, and how those "disapearing" branches during treelife time, produces the caracteristical shapes for each tree or more generally plants, we can see in real life, at a T time)

 

cheers!

[3DMK]

 

Eduard, can you give your point of view, do you think count rate% and chaos count% are working fine ? basically, i think there might be some code issues around the random functions,

 

 

 

I also believe the Chaos count% is not working as it should or rather, as we would like it to function. The input values do not give the expected results and it may need to be modified/fixed.

The child chaos option would also be nice although I think this can already be done using the affects panel with better results. The Affects panel can control the chaos% of children paths depending on their position along the parent path. [see attached example]

 

 

Currently there is no way to control this Chaos Count% value along a path such as you can do with Chaos Density%. It would be good to add the same path control for the Chaos Count.

 

Hope this helps

Cheers

Jamie

Chaos Count.zip

[mattpesquet]

I also believe the Chaos count% is not working as it should or rather, as we would like it to function. The input values do not give the expected results and it may need to be modified/fixed.

The child chaos option would also be nice although I think this can already be done using the affects panel with better results. The Affects panel can control the chaos% of children paths depending on their position along the parent path. [see attached example]

 

Hey Jamie, thanks for your reply and the file.

Let me first congratulate you for the good work, i see your trees and plants are getting great attention to realistic patterns and trees/plants architecture just as i m trying to do using growfx, with on my side a maybe more obsessive will of using real life measurements/values and already well described tree species architecture

(i use personal observations and some biologistical studies/books with the last and most advanced knowledge about each of the 22 architecturals tree models possibles, and used by the whole plant/tree species on earth, how each species can move from one model to other one, during tree life, or how it is mixed among the whole hierarchy).

 

For example, tempered region trees (my main work in progress) use only 7 of the 22 architectural models, but the way they are combined for some species

(few species only use 1 architectural model during the whole life, or among the whole hierarchy, expect the trunk most of the time, that makes them easier to understand and to generate with growfx),

or the way they are switching from one level to another of the hierarchy can be really tricky (it is also evolving with tree life, and therefore branches time life), but i m sure and already experienced this complex knowledge can be translated to generate the most realistic patterns (with right branching pattern, right length/radius ratios for Parent/children elements, right density values).

 

We just need few things like Chaos count and count rate % to be working fine (as i said in my first post, i suspect most of the chaos parameters to not work correctly, as the Chaos angle for vector start, it applies the same random value of the defined range, to the same hierarchy group of branches, not a random value per parent path, what makes it almost useless..) 

And 2-3 other really important things to me that i would love to discuss with Eduard.

 

For example i really, really think that growfx miss a "vertical" offset % (accessible by affects) for path distributions children, does that make sense or should i say more ? we badly need to break the "computer" pattern that is repeated among all element of the same level (i should put some screenshot here, with problem about chaos angle for vector start, and why we need "vertical" offset along parent path).

 

 

More generally Chaos parameters should be able to give random value among the range we set, for each element of this hierarchy level, not the same random value to all elements.

 

Regarding your file and your reply Jamie, few screen shots with explicit comments, let me know if something is not clear enough.

But, to make it short, the affect cant help if Chaos % are not working fine, we still get like 3-4 solid step among the 0-100 range, like is said in my first post, the value just change according those misterious steps whatever value we put there, displaying or the 100% or something like 0-5% as in your example file.

 

 

and another screenshot with how it should be working to me, and the point of a "per children group" option :

 

 

Cheers.

[ptrain03]

I looked into this a little more.

 

the chaos density is a little strange but it is predictable and seems to be working as described;

In the help file it is defined as follows:

 

Example: Density = 10, Chaos dens.% = 50, so the density varies from 5 to 10.

 

If you had a 10' distributing path with a 12" module and a density of 10 (paths = 100), then at a chaos % of 0 you would get 100.

 

at chaos % of 10 you would get the mean of 9 to 10 (so apx 95)

at chaos % of 20 you would get the mean of 8 to 10 (apx 90)

at chaos % of 30 you would get the mean of 7 to 10 (apx 85)

 

and so on. 

 

I am not good at math but this this seems to work predictably.

 

 

On the other hand, I can't really make sense of how the count rate is calculated.

In the manual it says: 

 

Count rate "Specifies how many paths in percent to calculate using the selected distributor"

So it sounds like is meant to work as you have described in your posts, but for some reason it does seem to be very unpredictable. In situations like the application of modifiers It appears to work as described but It seems to run into problems when defining a distribution.

 

weird.

[ptrain03]

count rate % does seem to be working correctly when applied to a path position distribution. you might try converting your path distribution to a path position distribution and then applying the count rate.

 

 

good luck !

[mattpesquet]

ptrain03 >> Thank you for your replies, but you keep pointing the chaos density...that would be another thread at least, but i m not complaining about anything related to chaos density. 

And count rate just working with path position distribution isnt really the solution, i m not sure you would plan to turn a path distributor of second or third level hierarchy element to path position distribution (that means lot of childrens sometimes), plus this would apply the same position distribution to each of the "parent path", not much randomness here.

 

 

I m kinda disappointed that all the writing and screenshots painting cant get a reply from Eduard, i have been used to see way better feedback to growfx customers since i purchased this great plugin, i really think almost all chaos functions are not working fine, and i really think that it is an important question (as the "vertical offset or chaos" request).

[Eduard]

Hi Matt,

 

I've attached a file-max, where you can see how to work the "Chaos Count%" and "Count Rate%" but for the modifiers. And seems the "Count Rate%" for distributors is not working properly, and we'll investigate this issue, and later I'll let you know about it.

 

To solve your problem, I think you can use the Slice modifier, and there you can specify the "Count Rate%" as you like, and it will work fine. How it works, please see the attached max-file.

 

Thanks!

Chaos_1_2012.max

[mattpesquet]

Hi Eduard, thank you for your time.

I carefully explored your max file, and first of all i agree that for modifiers "Chaos count %" and "Count Rate %" seems to be working fine, and just as we would like it to work for distributors.

 

>> Regarding your first example with use of "Chaos Count %" in distributor in point on Path_03, i still noticed something weird just as described in my initial post about solid steps and no linear variation, for distributor count set at 20 as in your original file it seems to be working fine, at least dropping from 100% to 0% almost gives variation at each 1% step (the variation still doesnt look totally random to me, and it would be great to be able to choose if we want % to be applied at "node" level, path level, or whole distributor paths of same level, i already talked about it with some screenshots).

But if i lower distributor count to something under 10 (for example 6 or 8), dropping "Chaos Count %" from 100% to 0% starts to give variation at few "solid" steps (like 7-8 steps over the 100%-0% range), even if we have still way enough "children" at the whole level state to get a variation at 1% step.

 

There is something about your file and settings that i noticed, and that makes me think about something i wanted to ask since a long time (what is the best Max units preferences and settings to work with growfx if there is one ? i guess units settings can affect some internal growfx computation, the reason this Unit Calc setting exists i guess, even if i m not sure i 100% understood how it works, and the benefits).

I notice when i open your .max file that in the preference object of any growfx primitive Unit Calc is set to 1000 cm (and all values in growfx panels display centimeters).

i work on max 2012, my max units preferences are set to 1 unit = 1m (i mostly do archviz), usually if i create a growfx object with those settings, my default Unit Calc value is 10m (that sounds good and the same than 1000cm), but why with your file i see centimeters, even if my max settings keeps saying 1 unit = 1 m ?

Is it all good having 1 unit = 1m for max preferences, and Unit calc = 10m (or 1000cm) ? or should i switch max settings to 1 unit = 1cm for my growfx work ?

 

>> about example 2 and 3 from your file, as i said i agree count rate seems to be working way better, and the slice modifier definitely helps to get what i was talking about ( just a simple "pruning" parameter working fine and linearly at each levels, with the ability to connect it to affects (that is a big difference, if i m not mistaken i cant use for example path position affect to adjust the % count rate of the slice modifier).

 

I hope you can fix the Count rate % and Chaos Count % of distributors easily, that would be a really good point (i still think other random parameters of distributors like chaos angle are acting weird, cf. first screenshot on Post #5, and all the blabla about it).

 

i would love to hear what you think about the "random vertical offset" or "vertical offset noise" kind of request for distributor settings, do you think it should be easily implemented ? but i would also like to know if i m the only one seeing all the potential benefits of this little function ?

 

(to me, this is the only thing that is impossible to do with growfx right now and that is wildely present in plants architecture, this is not the same than playing with density curve, but in nature grow units have an important random factor, i mean density curve drives kind of logical and "perfect" knots position taking into account the turn axis, and we want them to get a noise variation around this too "perfect" distribution just "vertically" not changing turn axis that can have his own random parameters from distributor chaos angles parameters. This could be a "vertical noise" modifier, vertical regarding the vector axis of the parent.)

 

Let me know if i m not clear enough on some points.

Thanks

[Eduard]

Hi Matt,
 

I carefully explored your max file, and first of all i agree that for modifiers "Chaos count %" and "Count Rate %" seems to be working fine, and just as we would like it to work for distributors.
 
>> Regarding your first example with use of "Chaos Count %" in distributor in point on Path_03, i still noticed something weird just as described in my initial post about solid steps and no linear variation, for distributor count set at 20 as in your original file it seems to be working fine, at least dropping from 100% to 0% almost gives variation at each 1% step (the variation still doesnt look totally random to me, and it would be great to be able to choose if we want % to be applied at "node" level, path level, or whole distributor paths of same level, i already talked about it with some screenshots).
But if i lower distributor count to something under 10 (for example 6 or 8), dropping "Chaos Count %" from 100% to 0% starts to give variation at few "solid" steps (like 7-8 steps over the 100%-0% range), even if we have still way enough "children" at the whole level state to get a variation at 1% step.

We'll investigate the algorithm and try to fix this problem.
 

There is something about your file and settings that i noticed, and that makes me think about something i wanted to ask since a long time (what is the best Max units preferences and settings to work with growfx if there is one ? i guess units settings can affect some internal growfx computation, the reason this Unit Calc setting exists i guess, even if i m not sure i 100% understood how it works, and the benefits).
I notice when i open your .max file that in the preference object of any growfx primitive Unit Calc is set to 1000 cm (and all values in growfx panels display centimeters).

The Unit Calc is mainly used to calculate the density in the Path distributor. Also to calculate the density in the Surface distributor, and in some modifiers.
You can specify any value. Just when you change this value, you'll need to change some parameters such as Density, and others.
 

i work on max 2012, my max units preferences are set to 1 unit = 1m (i mostly do archviz), usually if i create a growfx object with those settings, my default Unit Calc value is 10m (that sounds good and the same than 1000cm), but why with your file i see centimeters, even if my max settings keeps saying 1 unit = 1 m ?

Maybe you have 1unit = 1m in the System Unit Setup, but in the Display Unit Scale you've specified "Centimeters"?
 

Is it all good having 1 unit = 1m for max preferences, and Unit calc = 10m (or 1000cm) ? or should i switch max settings to 1 unit = 1cm for my growfx work ?

Yes. Some of our customers are using this (1 unit = 1m) settings.
 

>> about example 2 and 3 from your file, as i said i agree count rate seems to be working way better, and the slice modifier definitely helps to get what i was talking about ( just a simple "pruning" parameter working fine and linearly at each levels, with the ability to connect it to affects (that is a big difference, if i m not mistaken i cant use for example path position affect to adjust the % count rate of the slice modifier).

If you want, we can add the Count rate% of the Slice modifier to the Affects options, so that you can influence on this parameter.
 

I hope you can fix the Count rate % and Chaos Count % of distributors easily, that would be a really good point (i still think other random parameters of distributors like chaos angle are acting weird, cf. first screenshot on Post #5, and all the blabla about it).
 
i would love to hear what you think about the "random vertical offset" or "vertical offset noise" kind of request for distributor settings, do you think it should be easily implemented ? but i would also like to know if i m the only one seeing all the potential benefits of this little function ?
 
(to me, this is the only thing that is impossible to do with growfx right now and that is wildely present in plants architecture, this is not the same than playing with density curve, but in nature grow units have an important random factor, i mean density curve drives kind of logical and "perfect" knots position taking into account the turn axis, and we want them to get a noise variation around this too "perfect" distribution just "vertically" not changing turn axis that can have his own random parameters from distributor chaos angles parameters. This could be a "vertical noise" modifier, vertical regarding the vector axis of the parent.)

Regarding the "random vertical offset", to solve this issue, you can try to use an additional dummy path.
Please see my posts in this thread: branch-offset-on-z-axis
 
Cheers

[mattpesquet]

Hi Matt,

 

We'll investigate the algorithm and try to fix this problem.

 

 

If you want, we can add the Count rate% of the Slice modifier to the Affects options, so that you can influence on this parameter.

 

Regarding the "random vertical offset", to solve this issue, you can try to use an additional dummy path.

Please see my posts in this thread: branch-offset-on-z-axis

 

Cheers

Hi Eduard, any news about those bugs ? about the old thread link from marsh 2013, i noticed you said you should implement this kind of parameter for random Z offset soon. 

Does the last release provide any solution on this side ?

 

Thanks in advance