This is an intro pack rare, at least. Not the worst in that context, but Grave Titan sure has spoiled us.
I'm still waiting for the Devoid payoff. I'm not seeing the mechanical incentives to play with them other than a few odds and ends. Maybe a colorless land, or an uncommon lord?
T: Add one mana of any color to your mana pool. Spend this mana only on colorless cards or activated abilities of colorless cards.
This card though... 7 power for 6 is a lot for blue, but for constructed this looks awful. In limited I'ld be happy to open a large body that can break stalemates.
In my mind, the biggest problem with the current card pool is the quality of the big creatures. Really bummed they cut Sentinel of the Eternal Watch, she would have been great in this format.
Wait... can someone elaborate on this? I was under the impression that the entire Origins set was available on this Duels Online thing. How many cards aren't included? Why?
No, not the entire set. There are multiple cards missing, and Sentinel of the Eternal Watch is sadly one of them. I have no idea why they didn't just include it though.
Edit: I am most disappointed by the exclusion of Pyromancer's Goggles though. I would have loved copying the big red spells My best guess why they didn't include every card is to not step on the toes of paper Magic and MtGO. If you want the real deal, with all of the cards and no limit on how many uncommons, rares and mythics you can play, you have to invest money. It makes sense from a company's perspective to be honest.
I have extremely bad graphics on my Win7 laptop. Like, there's no background during the game. It's pixely cards floating in black space. Of course there's also the server overload, but apparently they want us to upgrade to Win10 first...
Do you have DirectX 11? Many old laptops (like mine) only support DirectX 10, and for me, it's the whole screen that's black, I can only see the cursor.
I have extremely bad graphics on my Win7 laptop. Like, there's no background during the game. It's pixely cards floating in black space. Of course there's also the server overload, but apparently they want us to upgrade to Win10 first...
Creessavage Aty 2WW
Creature - Elemental Wizard
Flying, Cycling 2
You may cycle Creessavage Aty whenever you could play an instant. If you do, prevent the next 2 damage that would be dealt to you this turn.
3/3
#Pretty nice, eh?
The last line is redundant. Cycling has always been instant speed.
Ah, my bad, I thought it was sorcery speed. Been a while since I actually played.
Very nice card for limited though! I can't imagine you would want to cycle it often though, as a 3/3 flyer is worth too much I think.
You're right. I think we want it to do something slightly different than it says to make it more interesting. That seems to happen a lot with these cards where it's "so close."
But if it nudges us into interesting design territory, doesn't that mean it's doing its job?
Ethershore Snape*****ter2B
Creature - Vampire Rogue
Flying
Whenever Ethershore Snape*****ter attacks, each other creature you control gets +1/+1 until end of turn.
1/1
#Poor Snape!
Edit: Ooo! This is an insteresting design!
Esper Deathmest2B
Enchantment - Aura
Enchant creature
Enchanted creature gets -1/-1 for each attacking creature
Hello all! My two loves are machine learning and magic, so this thread has been so very much fun for me. I've been working with neural nets for a while now, but only have light experience with RNNs so I hope my following thought makes sense.
Every prediction the net makes is based on the previous n characters. Which means if you format the data as name|color|mana cost|cmc|type|text (for example), then when it's predicting this new card's name, it's basing it off of the end of the previous card's text. So my question is this: would it make sense to alter the training data by adding n blank characters before each new card (where n is the memory length), so as to 'flush out' its memory of the previous card before showing it a new one? That way there would be no contamination from one card to the next.
Let me know what you think, and keep up the great work everyone!
If that is so, wouldn't that result in far more wrong cards. Power and toughness based on the previous card, etc?
#Oh wow! Playable and on-cost? It's instant land destruction, which is not likely to happen, but still!
Stormscare Soldier3
Creature - Elf Shaman (R/P): This gets +2/+2 until end of turn.
2/2
#This definitely needs the Rootwalla "once per turn" text, and it should be (G/P), and the name is wrong, but still... This is perfectly printable with some tweaks!
n:Great Harmane|m:{1}{1}{1}{1}{1}{1}{1}|cmc:8|r:Affinity for artifacts|t:[Artifact Creature]|p:4/4|s:[Beast]
#It recreated Myr Enforcer!!! It's wrong about the converted mana cost, by 1, though
Trestchous Resurtance4W
Creature - Human Samurai
Vigilance
When Trestchous Resurtance enters the battlefield, you may search your library for a Goblin card, reveal it, and put it into your hand. Then shuffle your library.
2/2
#Sweet! I mean, low powered, but fun! A Goblin Matron in white
n:Steel the Sean|c:[White]|m:{W}{W}|cmc:2|r:Flash\nEnchant creature\nEnchanted creature gets +2/+2 and has flying.\nEquip {2}|t:[Artifact]|s:[Equipment]
#No! Bad NN! Equipment does not need the Enchant creature text!
#But the best one of the bunch, by far (even if it's slightly overpowered):
Mystic Frog1B
Creature - Horse
Megamorph 2B
When Mystic Frog is turned face up, you may shuffle a card from your hand.
3/3
In order to enhance the results, you should train the algorithm. Not manually change the input and hoping for better results.
That's like teaching a kid how to ride a bike, and the first time it falls, you get another kid.
Eventually you'll run into a natural talent, but that doesn't mean you taught it how to ride a bike.
At the same time, if the wheel on the bike is bent, you straighten it before trying to teach the kid to ride. As people have been discussing here, there are definitely ways to present the data to the NN that will be more conducive to it learning what we want to teach it. Messing around with the input is a perfectly valid approach to improving the results.
My network is learning to make good cards, but tweaking the way you present you input as well as tweaking the parameters of the lua call both have a great impact on the impact of your network's ability to learn. For example, my first try resulted in a lot of creatures without power and toughness, and noncreatures with p/t. When I put the p/t next to the cardtype the neural network did pick it up. In that sense your analogy doesn't make sense. I'm not getting a new kid, I'm getting a new bike because the old bike sucked. Yes, of course I can try to teach the kid to drive on a wrecked and rusted bike, but it's a whole lot easier to learn on a working bike.
I started a new run this morning with a new dataset. I split out the mana cost as someone suggested, stripped out reminder text, and tweaked the order of fields a bit again. I also have a much larger set this time. Basically Beta + everything from Mirage onwards, or something like that anyway. I cut some stuff, non-standard layout, un-sets, Sol Ring, Ancestral Recall and the original Moxen, etc.
Settings:
-rnn_size 300 -layers 3 -seq_length 100
Example card from my dataset:
n:Tibor and Lumia|c:[Blue,Red]|m:{1}{1}{U}{R}|cmc:4|t:[Legendary Creature]|p:3/3|s:[Human Wizard]|r:Whenever you cast a blue spell, target creature gains flying until end of turn.\nWhenever you cast a red spell, $THIS deals 1 damage to each creature without flying.
On my way to work I had a flash of insight, the mana cost really should be {U}{R}{1}{1} if we want to make it easier to learn the colors based on the mana cost.
So I thought a little too hard about how to justify flying on a black cat who raises things from the dead, and realized that mounts don't contribute to creature type. A necromancer cat who rides undead flying things really captured my imagination, and I needed to do something fun to break out of an art block, so here you are!
PS Just saw Izzy's post. Really not trying to steal the spotlight, started the painting a couple of days ago and was eager to share it!
I feel this didn't get nearly enough thumbs up
Would love to see the Slidshocking Krow by Izzy, that surely must have gathered the most votes!
T: Add one mana of any color to your mana pool. Spend this mana only on colorless cards or activated abilities of colorless cards.
This card though... 7 power for 6 is a lot for blue, but for constructed this looks awful. In limited I'ld be happy to open a large body that can break stalemates.
No, not the entire set. There are multiple cards missing, and Sentinel of the Eternal Watch is sadly one of them. I have no idea why they didn't just include it though.
Edit: I am most disappointed by the exclusion of Pyromancer's Goggles though. I would have loved copying the big red spells
Someone compiled a list of missing cards on Reddit: https://www.reddit.com/r/magicduels/comments/3ckfrj/cards_from_origins_but_not_in_duels/
Yup, I got DX11. Don't know what's the problem.
Very nice card for limited though! I can't imagine you would want to cycle it often though, as a 3/3 flyer is worth too much I think.
But if it nudges us into interesting design territory, doesn't that mean it's doing its job?
Ethershore Snape*****ter 2B
Creature - Vampire Rogue
Flying
Whenever Ethershore Snape*****ter attacks, each other creature you control gets +1/+1 until end of turn.
1/1
#Poor Snape!
Edit: Ooo! This is an insteresting design!
Esper Deathmest 2B
Enchantment - Aura
Enchant creature
Enchanted creature gets -1/-1 for each attacking creature
If that is so, wouldn't that result in far more wrong cards. Power and toughness based on the previous card, etc?
I also apparently forgot to remove planeswalkers, d'oh! On the plus side, it handles color and converted mana cost relatively well, though not always.
n:Gerrard's Goblin|c:[Red]|m:{1}{1}{1}{R}|cmc:4|r:Exile target artifact.\nDestroy target nonbasic land.|t:[Instant]
Gerrard's Goblin 3R
Instant
Exile target artifact.
Destroy target nonbasic land.
#Oh wow! Playable and on-cost? It's instant land destruction, which is not likely to happen, but still!
Stormscare Soldier 3
Creature - Elf Shaman
(R/P): This gets +2/+2 until end of turn.
2/2
#This definitely needs the Rootwalla "once per turn" text, and it should be (G/P), and the name is wrong, but still... This is perfectly printable with some tweaks!
n:Great Harmane|m:{1}{1}{1}{1}{1}{1}{1}|cmc:8|r:Affinity for artifacts|t:[Artifact Creature]|p:4/4|s:[Beast]
#It recreated Myr Enforcer!!! It's wrong about the converted mana cost, by 1, though
Trestchous Resurtance 4W
Creature - Human Samurai
Vigilance
When Trestchous Resurtance enters the battlefield, you may search your library for a Goblin card, reveal it, and put it into your hand. Then shuffle your library.
2/2
#Sweet! I mean, low powered, but fun! A Goblin Matron in white
n:Steel the Sean|c:[White]|m:{W}{W}|cmc:2|r:Flash\nEnchant creature\nEnchanted creature gets +2/+2 and has flying.\nEquip {2}|t:[Artifact]|s:[Equipment]
#No! Bad NN! Equipment does not need the Enchant creature text!
#But the best one of the bunch, by far (even if it's slightly overpowered):
Mystic Frog 1B
Creature - Horse
Megamorph 2B
When Mystic Frog is turned face up, you may shuffle a card from your hand.
3/3
#Let me shuffle that card!
I get what you're saying, but
My network is learning to make good cards, but tweaking the way you present you input as well as tweaking the parameters of the lua call both have a great impact on the impact of your network's ability to learn. For example, my first try resulted in a lot of creatures without power and toughness, and noncreatures with p/t. When I put the p/t next to the cardtype the neural network did pick it up. In that sense your analogy doesn't make sense. I'm not getting a new kid, I'm getting a new bike because the old bike sucked. Yes, of course I can try to teach the kid to drive on a wrecked and rusted bike, but it's a whole lot easier to learn on a working bike.
I am not quite sure I am following you. Are we talking about the same thing?
Me? Pretty much this I think. Tweaking input.txt and parameters to see what works well.
Settings:
-rnn_size 300 -layers 3 -seq_length 100
Example card from my dataset:
n:Tibor and Lumia|c:[Blue,Red]|m:{1}{1}{U}{R}|cmc:4|t:[Legendary Creature]|p:3/3|s:[Human Wizard]|r:Whenever you cast a blue spell, target creature gains flying until end of turn.\nWhenever you cast a red spell, $THIS deals 1 damage to each creature without flying.
On my way to work I had a flash of insight, the mana cost really should be {U}{R}{1}{1} if we want to make it easier to learn the colors based on the mana cost.
I feel this didn't get nearly enough thumbs up
Would love to see the Slidshocking Krow by Izzy, that surely must have gathered the most votes!