Creature Fusion Discussion

Avatar

By Leo 17 Jan 2016 21:00

Moderator · 70 comments

I've been wondering for a while what you meant by "fusions".
If you really want to make any pair of creatures "fusable", well, then every sprite would have to be specifically planed so that you can algorithmically cut and paste parts of a creature onto another, switch palettes, etc. It's a quite complex endeavour, because you need pixel perfect fusion (because of the sprite size).
Then would you allow unlimited level of fusions? Why?
Why do you want fusions? because of the pkmn fusion meme? A fusion system would work fairly better in 3D.

Avatar

By ShadowApex 18 Jan 2016 19:16

Lead Developer · 374 comments
Leo wrote

I've been wondering for a while what you meant by "fusions".
If you really want to make any pair of creatures "fusable", well, then every sprite would have to be specifically planed so that you can algorithmically cut and paste parts of a creature onto another, switch palettes, etc. It's a quite complex endeavour, because you need pixel perfect fusion (because of the sprite size).
Then would you allow unlimited level of fusions? Why?
Why do you want fusions? because of the pkmn fusion meme? A fusion system would work fairly better in 3D.

I was never a big fan of how breeding worked in the Pokemon series. Cross-breeding Pokemon would just result in one kind of Pokemon and figuring out which Pokemon are in which egg groups is cumbersome.

When I saw Alex Onsager's post on how his Pokemon fusion site worked, I tried implementing his ideas in Python and found they worked really well. You can see the code I wrote to actually perform the sprite fusion here. After I knew that this kind of fusion was technically viable, I thought it would be interesting as a game mechanic.

Right now the fusion mechanic hasn't been a very high priority, so it's still open to discuss how or if it should be implemented. I think if done right it could be a very interesting part of the game and one which could further set the project apart from the Pokemon series.

Also, with the current fusion system, you wouldn't have an unlimited number number of fusions. Each monster is divided into their "head/face" with their color palette and a "body". Fusing monster A with monster B would result in a monster with monster A's face on monster B's body with monster A's color palette. Since there are only 2 possible fusion combinations between any two monsters, there is a finite number of different variations.


Avatar

By benneti 18 Jan 2016 21:34

Member · 38 comments

As this is my first post, i wanted to say this is a awesome project (i say this as pokemon fan and a free software enthusiast), thank you to all contributors and i hope to contribute in any way as soon as i have enough spare time.

But i wanted to second Leos opininion about fusion.
I think the point of a catch and train advanture is to get a virtual bond between the trainer and the tuxemons and to catch them all (in my opinion these two are in contrary to fusion).
But I think fusion on a small scale would be a nice perk.
With small scale i mean some (proprietary?) fusions created by the evil side (proprietary in this case could mean that they cant  recreated their selfes in the wild since they were scientifically produced and the organisation would have no interest in a self growing product (comperable to some genetic manipulated herbs?)). Another aproach would be to make fusions of a few selected mosters possible so the player has to try it out in something like the pension.

i hope everything was understandable smile
Keep the good work smile

Avatar

By Leo 18 Jan 2016 21:46

Moderator · 70 comments

Well, I also find the pkmn breeding system quite horrible.
The downside with this type of algorithmic fusion is that:
- Onsager's method is not pixel perfect, so as is it is not suitable, so in Tuxemon we would need to put more restrictions on sprite design (like standard head size).
- If you fuse a bird with a lion, you expect a griffin, not a chicken-headed dog. It would be possible to make some special fusions (like the aforementioned) but it would make the others boring.
- Some fusions may look terrible. If we allow a variety of positions in sprite, this number will go through the roof. If we don't allow much variations, things will probably be better.
- Even then, if you look the drawings inspired from "fusion pkmns", they are usually quite changed from the direct output of the algorithm, because they're not that good. 

Note that if the game was 3D, the discussion would be completely different. You could easily graft part of the skeletons, textures and motifs to make very interesting fusions without losing anything.

I'm not totally against it (squaring the number of available creatures is quite huge), but it comes with a price, and may compromise the aesthetic integrity of the whole game (compare hand-drafted sprites with odd fusions...).
Also, yo-kai watch already does fusion (as well as draqwe monster I think), so I'd rather set Tuxemon apart with other ideas (dynamic battle systems, forms instead of evolutions, etc). Also we can more variations in creatures from color palettes and alternative sprites.

Avatar

By ShadowApex 19 Jan 2016 00:19

Lead Developer · 374 comments
Leo wrote

- Onsager's method is not pixel perfect, so as is it is not suitable, so in Tuxemon we would need to put more restrictions on sprite design (like standard head size).

You're right that Onsager's current method does scaling, which would not look very good in-game. My thoughts have been that we would have a separate set of head and body sprites for each fuse-able monster, which is completely independent of the normal monster sprite. This would let us perform pixel-perfect fusion with other monsters and introduce variation in the bodies and faces of different size.

Here is an example Tigrock with separate custom-drawn head sprites depending on if the target body is small, medium, or large:

wZpEKlk.png

There is also an additional "face" size that can be a set size larger than the head for things like long hanging jaws, antennae, etc.

Of course, not all of the monsters will be fuse-able until these sprites are made, but that's all right. Not all of the monsters need to be fuse-able right away. We could even have a select few monsters that can be fused if we want.

benneti wrote

I think the point of a catch and train advanture is to get a virtual bond between the trainer and the tuxemons and to catch them all (in my opinion these two are in contrary to fusion).

I think we could solve this problem by making fusions a breeding system. Instead of combining two of your monsters together to make one monster (which would make you lose both original monsters), you can breed two monsters together to make an offspring that would resemble both of their parents.


Avatar

By benneti 19 Jan 2016 13:07

Member · 38 comments
ShadowApex wrote

I think we could solve this problem by making fusions a breeding system. Instead of combining two of your monsters together to make one monster (which would make you lose both original monsters), you can breed two monsters together to make an offspring that would resemble both of their parents.

Yeah this is similar to the second idea i meant (wich was inspired by some things i read on this forum), where you could breed them, but i think it may be easier and seem more natural if it is only possible to breed selectet  TXMNS together, this would in addition add a speciality to them (wich i think would not be the case if you could combine them randomly).

Avatar

By ShadowApex 23 Jan 2016 23:36

Lead Developer · 374 comments

I've moved this discussion into its own thread for further discussion on the fusion mechanic.


Avatar

By Leo 25 Jan 2016 17:13

Moderator · 70 comments
ShadowApex wrote

You're right that Onsager's current method does scaling, which would not look very good in-game. My thoughts have been that we would have a separate set of head and body sprites for each fuse-able monster, which is completely independent of the normal monster sprite. This would let us perform pixel-perfect fusion with other monsters and introduce variation in the bodies and faces of different size.

I think this solution would work, but it doesn't adress any of the other issues.

There is an alternate solution: introduce a class of creatures, say "chimeras", that are built of random elements.
We conceive these elements like bricks with specific ways to arrange them, maybe with some constraints (like a body of type A cannot go with legs of type B, etc). Then provided with enough bricks it's easily going to make millions of possible creatures, that are naturally fusable.
This solution would require less effort, and without compromising the quality of the game sprites.

What do you think?

Avatar

By ShadowApex 29 Jan 2016 19:27

Lead Developer · 374 comments
Leo wrote

There is an alternate solution: introduce a class of creatures, say "chimeras", that are built of random elements.
We conceive these elements like bricks with specific ways to arrange them, maybe with some constraints (like a body of type A cannot go with legs of type B, etc). Then provided with enough bricks it's easily going to make millions of possible creatures, that are naturally fusable.
This solution would require less effort, and without compromising the quality of the game sprites.

What do you think?

This could work really well. I'd like to hear more about this idea from the community. It would certainly be easier to only have a small subset of creature "bricks" that we could use to make a separate class of creatures.

We could still conceivably base all the creature "bricks" off of existing creatures if we wanted to (E.g. Rockitten's ears design, etc.). The main concern with this would be ensuring that there are enough constraints to make the resulting sprite look like a cohesive new creature.


Avatar

By Leo 29 Jan 2016 21:48

Moderator · 70 comments
ShadowApex wrote

We could still conceivably base all the creature "bricks" off of existing creatures if we wanted to (E.g. Rockitten's ears design, etc.). The main concern with this would be ensuring that there are enough constraints to make the resulting sprite look like a cohesive new creature.

Yeah sure.
About the cohesiveness even simple fusion already has this problem, think Charikarp, the terrible fire-breathing sardine ...  neutral
There's a simple way to solve this problem: each brick belong to a category (leg, body, etc) and has a class (fishy, birdy, wolfy...).
Any creature made of bricks from one class should look coherent. Then to spice things up (and get things like griffins or capricorns) we can allow 80-20% mixes (or more if it holds well).
It could even be something that you unlock little by little, e.g. at first you can get only 100% chimeras, then (90,10), etc.