I Simulated a Forest with Evolution 

For anyone interested in the source code or just want to play with it: www.gallahat.com/public/waitlist It's not ready yet, but you can join the waitlist and I'll let you know when the website is ready. And i'll get the audio better next time 😬

to make the trees not as clumped, you could add wind that carries the seeds a few meters

The cool thing about forests is that they are so much more than individual trees. They can control their own temperature by collectively releasing water into the air, they can exchange nutrients via fungal networks, and so on. Really cool simulation so far and so much potential yet.

Here's an idea: make the trees also require water. So they would grow more around the banks of water in your terrain. With this you could also experiment with what tree types form in different climates.

Great video! Small tip: Turn off the day/night cycle during timelapses, or at least include a night light. The continues flickering of the cycle during timelapses is quite jarring. Good luck!

You didn't need to stop the logs from growing in the same space. You just need the growth to depend on how much light they have. If a log is already there, the new branch can't get enough light and won't grow in that area. That's how it works in nature. Also, trees in shadows shouldn't grow as well. This is why seeds evolved to scatter far away from the mother, so they don't drop under the canopy.

I think a good factor for realistic trees is max branch depth. A tree with a maxBranchDepth of 3, can have the trunk, branches of the trunk, branches of the branches and branches of of those. No more

The way you made the energy transfer from leaves to logs downward, is basically how mosses do it, cell-to-cell

"L-systems were introduced and developed in 1968 by Aristid Lindenmayer, a Hungarian theoretical biologist and botanist at the University of Utrecht.[1] Lindenmayer used L-systems to describe the behaviour of plant cells and to model the growth processes of plant development. L-systems have also been used to model the morphology of a variety of organisms[2] and can be used to generate self-similar fractals." - Wikipedia about L-system I wanted to create a game that grows procedurally generated plants, trees, and flowers but I'm too ignorant :'( Anyways, maybe you can use the math to make an improve tree/plant growth simulation.

Surprised how you only have 350 some subscribers, good video!

Genes ideas: Amount of energy before generating new log (making trees smaller/higher) Shape of leaf (could influence other things) Number of leaves per log (more leaves = more energy production but requires a higher amount to generate new log) Also for aesthetetic, you the leaf should have a fixed width mini log

*99 missed calls from RockStar* 💀

This is great! Thank you for pouring so much time and effort in to it ❤

A suggestion would be to add nutrients in the ground that slowly replenishes so not a lot of trees can be in the same place

being inspired by trees is a new one for sure

Time of day (night) you can make not a completely black screen, but just a “blue” color. To justify this, one can imagine that this is “moonlight”, which does not provide energy, but ignites and one can see what is happening. Besides, it’s more comfortable for the eyes, otherwise it takes a few minutes to “turn the light on/off” 🚬🌚

I think most plants gain energy in the day, but actually do most of the growing in the night cycle. So it would make sense to store the energy gained in day (without growth) and then the trees grow in the night time from the energy that they collected in the day.

This is so cool. Im sure you will blow up soon! Edit: I have finished the video, and im excited to see your future projects. Another comment mentioned having a script and I think that would be very beneficial for the more complex parts of the video to help explain what is going on better. Personally I can find it hard to get the right words across so it can help to plan it out. I also believe it would help structure the journey you had better for the viewer. Have a good one :)

I’m studying plant growth and physiology in university and I think a really good tool for simulating plant growth is using a plant Relative Growth Rate equation. Here’s one: RGR = LAR * NAR Relative Growth Rate = Leaf Area Ratio * Net Assimilation Rate. There are good explanations online of how this works.

im not botanist but if i remember plants grow faster at night because they focus on photosynthesis during the day. So exposed at light tree should store energy (max_energy = tree_level* base_value; increment_energy() energy += tree_level * base_value) and at night old tree should grow slower than young ( log(tree_level + 1))

Yooo crazy. Evry time I see a video about simulating nature I smile

I feel like people who code these cool random things could make a whole functioning changing ecosystem

What might be interesting is something like "depth-modified branch chance" which is a function (maybe a polynomial with the factors being the genes) which controls the chance for a log to branch as a function of the tree depth. This might allow trees to develop certain shapes.

Great video! This topic fascinates me and I’ve worked on some of my own for a few years now. Great to see another enthusiast :)

You are so underrated!

Wow you are so underated you deserve more subs this video.

Basically trees from lumber tycoon on Roblox, that’s my simplified variation of what’s going on I do know what’s going on without the simplification

When I studied evolutionary algorithms at uni one of the most interesting things I read was a paper on recessive genes. They used a 20(iirc) char string to encode the angle between each segment of a wing, and a value for wing length, and had a second copy used for recessive genes. Then culled any sequence that exceeded a bounding box. As you'd expect, when simulated the ones with straight wings did better, until the recessive gene for two 90° angles fell in the right place and made a biplane. I think in this simulation you could get some very interesting results if you changed what genes you encode. E.g. if you don't simulate energy transfer within a tree can you get the performance budget to encode variable branch angle based on height or something?

Nice video you are Crazy underated you deserve more subs!!

really interesting video, I've read a bit about the math of how trees grow and you did a great job demonstrating the complexity of the problem and simplifying it. Passing down 75% of the energy is a a great way of doing it, kinda simulates how auxin is passed down from the top of the plant which inhibits growth and causes that node to release more auxin so the cumulative affect is similar to what you ended up with. One interesting thing plants do with the seedlings is younger plants tend to have leaves which are tuned to use light that's already passed through the leaves above it - the high leaves absorb more red light because it's more efficient so this leaves more of the blue in the light which younger plants with more chlorophyll a are turned to use - you could probably simulate this by giving trees under a certain age a buff that decreases the shade on them? not sure if it would change how things evolve in your model but i think if trees died at a certain age then having ones half grown to replace them would be important, especially if seeds could only be produced on mature trees. anyway yeah love the vid, given me loads to think about and subscribed looking forward to your next video. (oh and if you're looking for things to simulate how plants grow towards light by using cell elongation would be a fascinating thing to play with and you could try out all sorts of weird light positions and motions to see how it affects them and what odd shapes you can get them to grow into)

pls make more of this stuff, and make the simulation have even more things, like small plants, weather (and then the trees can evolve to handle more water to become tropical), make it so that the plants have to fight for the water (so that the more rain, the more plants), and maybe add mushrooms and animals. Plants first, though. Me subscroobed from this video! So pls take my ideas into consideration.

Wth is this, I just love them trees. I was also waiting the whole time for a species that would build up high and fast so that it could live in the water, I assumed it creates a shadow inside and hinders the saplings there.

so cool!!

seeds contain a huge amount of energy, so really only mature and thriving plants can create them. and they depend on being carried away by weather and animals, that's why fruits exist. and trees always invest in the strongest branch, mostly rhe trunk. that's why small trees and big trees are the same shape. Yes, small trees branch near the ground, but those branches don't just get longer and longer and the tree wider, the energy is actually put into the trunk and other stronger branches, so the big tree still has the same shape and because it's bigger the branches are not near the ground anymore

This is so nice. Hope you're avinng a good day!!

now make mycelyum to eat the fallen off logs otherwise the place is gonna be a mess in a few million generations

to make it more interesting/realistic you could also make it so once the trees reach max height, they slowly start dying off. this would also make it less intensive on ur computer

now add a woodworking mechanic and make it in to a lumber game

I think you should make it so you can get leaves as well (depending on the energy of the log (or leaf since I think that grows into logs), if it has less energy, it can opt for leaves instead of trying to grow more logs) and make each leaf collect less energy, this way you should see more leaves and make the entire tree look like a tree and not like a glorified dead bush

thats amazing! you just earned a sub, also btw is it fine if i make somethign like this in my game cus i have never seen anyone do something like this!

I was here at 405 subscribers

i've spent decades on procedural gen. three days a week i chainsaw for a golf course, nice to spend time observing nature periodically. the most important part of any development is the time outside smoking weed :)

Alright this is epic, imma have to see where this channel goes Expect me to stay for the long run

That water is definitely radioactive

Amazing video! Looking foward to your future projects :)

that first clip is really close to where i live :D anyways great video!

Quando a energia já tiver descido pelo tronco da árvore você poderia tranferir ela para o chão. Talvez em uma área circular ao redor. de modo a permitir que, mesmo em regiões sem iluminação direta, por conta da copa das árvores, plantas de porte menor pudessem crescer. Pensei que uma maneira legal de pensar em plantas menores poderia ser atravéz da taxa de transferência da energia.

1. You can make apples or a fruit to make seeds. 2. The trees should have more green

When implementing evolution in this manner, it's essential to always include a mutation gene. This gene determines the likelihood and magnitude of each evolutionary change. Consider the scenario of evolving a human for sprinting. In the initial generation, individuals struggle to move efficiently. After several generations, they begin to jog. Millions of generations later, they achieve the ability to run. Finally, a nearly perfect runner emerges. However, their offspring often inherit mutated legs, leading to decreased performance due to random alterations. Yet, if the probability of mutation is inherited, subsequent generations experience faster improvements in fitness. The first individual to undergo significant leg mutation gains a considerable advantage, passing on this advantageous trait to their offspring. Consequently, successive generations explore increasingly radical leg modifications more rapidly. However, once nearly perfect legs are attained, slight adjustments become crucial for optimization. Thus, descendants with finely tuned alterations tend to outperform others. A similar outcome can be achieved through simulated annealing, gradually reducing the mutation rate over time. However, accurately determining the optimal mutation rate at any given stage is challenging. Therefore, allowing natural evolution to determine mutation probabilities proves more effective. Experimenting with this approach demonstrates a remarkable acceleration in evolution and a notable reduction in simulation duration.

Very cool video. This makes me think about using something similar for terrain generation in a procedural game, especially if you can get it to respond appropriately to altitude and water (and it already does shade, which is nice), so you just naturally end up with tree variety, denser or sparser regions based on lighting/water availability, etc. Though it'd probably take too long to do in full scale whenever you generate terrain, many games do procedural generation and then sort of hand pick the best result, and this could be used for adding flora to something like that, in a natural and varied way I think it needs more/larger leafs, so it can produce enough shade to actually discourage growth near big trees, and some rudimentary water logic such that water is less available at higher altitudes (and less available on more sloped land, where the water would quickly run down and away). Then basically more tree height = more water usage (or similar), and you should see some real species variety depending on nearby conditions, and some nice placement that fits with whatever terrain you do it on. Even nicer if they could 'use up' the water, like one big tree is taking all the water so other trees can't grow nearby, but that's a lot harder and the shade logic should kinda make that happen anyway

This is awesome!

hi, I rly like these kind of simulations and got also inspired by CodingAdventures. but its rly hard for me to get started in these kind of projects. any idea how to learn programming simulations like this? or maybe u could do a little tutorial or some more explanation in your vids. would be awesome!!

Simulations go brrr

Wonder, if anyone has done a simulation like this, but with simple plants, plant eaters, carnivores and possibly some form of decomposers together.

Super cool!

Interesting! Do you try using l-system to implemented? Some advice I think you can plan main topic for each section it will more clear.

YESS, SEB!

neat!

This is cool

I have a new alternative to grass now

please turn this along with other evolution-like mechanisms into a game(maybe Survival?) and release it on steam

can you please upload the project files or make a tutorial?

maybe add a weight factor so they die if there is to much weight on them also incurrages thinner branchs at the top

good OWO

plants grow taller in the shade

The reason your trees keep growing taller is same as it happens in the nature. They are competing for sunlight. But in the real world, there are physics for trees to follow, mainly gravity (on trunk structure and water transport). So in the real world, at place where water is sacred, plants tend grow low. But the behavior of competing for light won't change. If you've ever been in a forest, you should notice that all tree trunks are straight. It is because they all want to grow taller than their neighbor. Also gravity prevent them to have a big branch which may break if it is too heavy). 2nd reason is the survivability without sunlight. Some plants are able to grow in low light situation, some may even be afraid of strong sunlight. They are typical small, slow growing, but produce many offsprings(with spore). I suggest that you have some research in plants' "behavior"(growth in react to environment stimulants and stress). Knowing that will help you write a better simulation.

Add gravity and make it so that there is a very small chance for leaves to grow on the side of branches except for the main log Also storage of energy and age

cool

Could you give each organ a weight and make every trunk section have a limit to the total weight it can support above it?

add more L E A F 🍃

Evolution of the environment and terrain as time passes is the next step for games, in fact, is it possible to create erosion scripts in real time? For example, when it rains a lot over a period of time, will erosion occur that creates waterfalls or something like that? This is one of the mechanic ideas I want to put into my project, I don't know if it's possible. Sorry for my bad english.

You should add animals to consume the trees and animals to consume animals you know like ecosystem

I love evolution :D

please make a tutorial

That’s how god is feeling

in which software did you use for graphics unity or unreal? greate explanation 👍

Great Video! i would recoomend looking at Sebastian League's channel, as this reminds me of some of his earlier videos. :p

I support.

Are the trees spending energy to produce the seeds? Is that energy divided up amongst the seed's attributes?

❤

Maybe could use this for game development.

You could also try using the leaves as a factor for energy and not just sunlight the more leaves or even larger leaves the more chance to grow and see diverse species and chance of extinction for some not doing as well

This video is very nice but i feel like you should improve the video design a bit, maybe try practicing the script of your video a bit

Cool, can I see the code?

i think im addicted to these videos 😭

Gg

Release the code lol. It could be fun playing with it.

Will you remember me when you get famous?

I have a challenge I would like to pose to you. Could you make a dirt simulation? Like a "tiny" area, say 3 meters by 3 meters where the bottom meter is dirt. I have seen meat tearing simulations in real time so why can't we make realistic dirt? Hell ive seen minecraft block physics animations with better dirt than any game ive ever played. If you're interested let me know. Discord: hadies_alchemist

אחלה סרטון! כיף לראות סרטוני סימולציה ואבולוציה כחול לבן! זורק בקטנה הערות בונות למקרה שאכפת לך: 1. הזכרת את Sebastian Laugue, אחד הדברים היותר נחמדים בסרטונים שלו זה שהוא מסביר את מה שהוא עשה מספיק בפרטים כדי שתרגיש שתוכל לעשות את מה שהוא עשה בעצמך. מרגיש לי שאצלך ההסברים היו הרבה יותר בגבוה וזה קצת פחות קוסם. סתם לדוגמא הייתי שמח להבין איך דאגת לזה שהעצים לא מתנגשים אחד בשני. 2. נראה שהיה יכול להיות הרבה יותר מגניב אם היית חוקר יותר בפירוט למה המדדים שהצגת משתנים כמו שהם משתנים, זה היה חלק קצת קצר. ממליץ על Primer כערוץ לדוגמא. 3. ואחרון חביב: אחי - קצת קשה עם המבטא :) כל הביקורות בונות בלבד ואני חושב שעשית עבודה מדהימה, אני מנוי ומחכה בקוצר רוח לסרטונים הבאים. שיהיה בהצלחה!

bro, dont chenge volume of music when u talk

Try writing a script before recording voice. This way video will have more structure and be more engaging.

Audio is in one word: Terrible. You've set it up to auto-increse music volume while you are not speaking, but you often make short breaks between words which causes the music to kick in for a split second. You could have easily avoided this by manualy increasing the music volume when needed. Also the mic pop is very annoying. You could have easily gotten rid of it using a tool like Audacity. It might seem like a small issues to you, but I literally can't watch the video because the sound is so annoying. Too bad. The video idea seemed interesting.

This is cool