The"Split"Node_bug
#1
The s node particle output is incorrect.

For example, the amount of particles output by 50% is wrong, and the value is still wrong when adjusting the value.
  Reply
#2
Can you upload your file please Smile
  Reply
#3
(04-16-2019, 02:11 PM)tyFlow Wrote: Can you upload your file please Smile

Sometimes it will be normal, sometimes it is not normal, sometimes it is wrong to transfer the value. In short, there are always some strange problems.
  Reply
#4
Hi,

I checked the file...there doesn't seem to be anything wrong with the Split behavior.

In the file you attached, you have it set to "percent of new particles"....but because there's only a couple new particles per frame, "20%" of those is always going to be zero.

The setting you want is "Random Chance" and then set timing to entry. That will split particles based on a probability, rather than a percent of the total new. What that setting turned on the result is similar to the pflow result.
  Reply
#5
Hey cgdvs,

I saw your latest video on the facebook group (you closed comments for some reason so I'm replying here).

The behavior you documented is still 100% expected behavior for the "percent of new particles" method.

When you have one particle stream flowing downwards, there is only 1 new particle per frame being generated. So the algorithm is getting all the new particles (1), multiplying by the value in the spinner, and then returning the rounded result. So for example, round(1 * 45%) = 0. round(1 * 55%) = 1. So it makes sense that either all or none of the particles are being sent, depending on the percentage you enter.

Then, when you generate several streams of particles flowing down, you have multiple new particles per frame (1 new particle in each stream). So now a subset of the total number of particles will be sent. So if you have 5 particles creating streams, that's 5 new particles per frame. round(5 * 45%) = 2. round (5 * 75%) = 4, etc. So that's why only some of those particles are sent each frame.

I hope you understand a bit better now how Split works, using the "new particles" method. You need to choose "random chance" if you want particles to be sent based on random probabilities, instead of percentages of new particles.
  Reply
#6
First of all, thank you for your attention.
I closed the comment, mainly because I sent the video just for you. I found out that you saw it, so I turned off the comment so that it doesn't make this post take up too many resources.
I found that you are right. Still an old problem, I didn't know enough about TFYLOW particles, which led to my mistake. Thank you for your patience.
Today, I carefully observed the description of the parameters. Now understand what you mean. There are no problems with the node. Thank you
  Reply
#7
Thanks cgdvs, I appreciate your careful consideration and critique of the various operators. It will help me find and fix future bugs Smile
  Reply


Forum Jump: