General Custom Cards: The Lab

I mean, Pacts as Force spells?

Force of Summoner's Pact {2}{G}{G}
Instant
You may pay 1 life and exile a green card from your hand rather than pay this spell's mana cost
Search your library for a creature card, reveal it, put it into your hand, then shuffle.

I'd play the hell outta that.

Admittedly doesn't do much with the Red or White pacts. Or black really.
Nevermind
 
The tricky thing with the Forces is the FoW is actually only worth 2 mana. Obviously, the reactive and protective nature of it makes in almost invaluable, but the low mana value of the effect must be taken into consideration, as well. As a result, I avoided the "exile two nonland" because I couldn't imagine WotC actually printing that under 5, but maybe I'm wrong.
The white Forves make me nervous when comparing to Spectral Procession or to Anguished Unmaking, but I very well may be too conservative.
The blue ones are great, as we know.
The Zombify is exciting on turn 1 with some form of Looting. The Murder is a bit boring, but will be useful and is a fair hardcast as a game drags.
The red ones were hard because Pyrokinesis looks strong but apparently isn't. I figure that hitting mana sources is always going to be good.
I think that another 0 mana tutor (I have Summoner's Pact) will be great for any creature combo decks I put in. Good suggestion.
EDIT: Immediately noticed wrong artist credit on Riches. It's Dominik Mayer.
 
I'm using downloaded Cardconjurer and I just noticed that it doesn't have the NOT FOR SALE text on there anymore. Will this be an issue with MPC?

Is there a good place to get high quality MtG art? Sometimes scryfall's is kinda ass. Looking for Commander Collection Black Sol Ring art.



The "not for sale" missing part on cardconjurer isn't a problem. I made an order with the "downloaded" cardconjurer on mpc and it arrived this weekend.
 

Onderzeeboot

Ecstatic Orb
I started a new project: Dominaria Reforged. Which is basically a "What if I designed Dominaria Remastered with cube in mind?" I'm using the same basic design constraint as WotC did, with the addition of Dominaria United and Brother's War as eligible sets, because of obvious reasons. If you want to follow along, you can check the W.I.P. here. Be prepared for a glacial pace of progress and potential project abandonment though :p Also note that, while you can draft the cube, I haven't fleshed out many of the new themes, so you might find the draft experience a bit/lot wonky.

Anyway, I wanted to show off the five Onslaught fetchlands with new artwork I created with MidJourney, because I think they came out really nice!

Flooded Strand.jpgPolluted Delta.jpgBloodstained Mire.jpgWooded Foothills.jpgWindswept Heath.jpg
 
not gonna lie, on principle i really prefer art that a person sat down and used their hands to make (whether that be holding a brush or a tablet), but i can’t sit here and tell you the AI art LOOKS bad. cuz it looks real nice
 
AI Art as a new tool for generating unique and interesting works = good
AI Art as a mass-mechanism for stealing existing artists works without credit to derive its results = bad

Is how I see it.

They do look real nice.
 

Onderzeeboot

Ecstatic Orb
AI Art as a mass-mechanism for stealing existing artists works without credit to derive its results = bad
Interested to learn what you mean by this. You can use AI to imitate an artist's style like you could do in real life (if you put in the hours), but can you truly steal it?
 
Folks have used one of the AIs to basically replicate existing works without meaningful modification.

Thinking about springing for the paid MidJourney plan? I'd love to go halfsies on one! It's stressful trying to come up with the perfect phrasing so that it doesn't have to be re-tried.
 

Onderzeeboot

Ecstatic Orb
Folks have used one of the AIs to basically replicate existing works without meaningful modification.

Thinking about springing for the paid MidJourney plan? I'd love to go halfsies on one! It's stressful trying to come up with the perfect phrasing so that it doesn't have to be re-tried.
I am on the cheaper of the two general paid subscriber plans for MidJourney, which allows for 200 minutes of fast time per month, but no relaxed mode (which is basically a slower mode that doesn't eat up your monthly allowance). You can read more about the subscription plans here.

I pay 11.52 euro per month. On average, 1 minute generates 1 image, and generating one final piece takes at least 3 minutes. The first step is to generate 4 small scale options, e.g., for Polluted Delta, I got the following options.

Polluted Delta options.jpg

I upscaled the top right option, then did a light redo (which reduces noise and makes the image tighter overall, though sometimes you lose character this way), then upscaled the result to max resolution. So in total, this particular image cost me ~4 minutes. However, it took me a few tries to get to a cue that generated art I liked.

Polluted Delta option B.jpg

Polluted Delta options A.jpg

So far, I used up ~90 minutes (so almost half) of this month's fast time to generate these five fetchlands plus art for these.

Stormbind.jpg Dakmor Salvage.jpg Barbarian Ring.jpg

And 11 15 failed attempts at trying to create art for Braidwood Sextant, because MidJourney doesn't seem to understand what a sextant is... Also, Dakmor Salvage took a long time, because no matter how much emphasis I put on leaning, slanted, at an angle, toppling over, etc., MidJourney just won't paint a leaning tower!

So far I've been using MidJourney to create art for those cards that I can't find the high res art I wanted to use for, and, in the case of Stormbind, to generate art that is something other than a naked white lady whipping up a storm.
 
Last edited:
Those Fetch lands looks amazing!
And they fit together because they share a common theme. They all have that watery current. Also they look like they could be on the same plane. Maybe not Dominaria but another plane.
 
Interested to learn what you mean by this. You can use AI to imitate an artist's style like you could do in real life (if you put in the hours), but can you truly steal it?
Basically what Spice Witch said. Like, say you want a book cover in a certain artists style. Previously, you would commission that artist and they would make a book cover. They would get a nice little check and move on with their day; you would get a nice little book cover and move on with your day. Now, in theory, you can feed that artists entire library into one of these and generate a book cover acceptable to you for minimal opportunity cost (the time put into the task).

So what it comes down to, is.. is it "imitation", or is it a derivative work? The latter which would be covered under copyright and the artist due royalties or whatever else they deem under their copyright protections, as would be the case for any derived work not covered under fair use. In the most technically straight-on look, it definitely seems a derived work. It is literally being fed into code in the direct pursuit of making an output based on it. Unfortunately for artists, copyright case law has absolutely not caught up to this.

So yeah, basically back to what I said. As a new tool for developing unique work, AI is cool. But used to basically auto-derive works, using artists handmade crafts that they likely rely on to make a living, in the pursuit of circumventing them/copying them without credit/etc.... not so cool.
 
And I should say, I think the derived work bit applies even to works not a direct structural copy of some composition. This isn't you spending 2 years mastering the same techniques they use so you can also create art in that style. This is using their works to... derive a work in their style. Directly utilizing their works. I hope case law agrees when it catches up, so at least artists need to either give permission or be given their dues when fed into the machine, as it were.
 

Onderzeeboot

Ecstatic Orb
So, my opinion is a little more (or less?) nuanced, I think. Basically, a) technological advancements have made and/or are making certain professions obsolete. Court jesters, plague doctors, cavalry, scriveners, switchboard operators, cashiers, there are a lot of occupations that fell by the wayside in history because technology caught up to them. This is very lamentable for individuals affected, but technological advancement isn't going to slow down, and in general made people's lives better. On average, living conditions are better now than they were 100 years ago. I know capitalism isn't fair, and many people are still exploited by the day, but on the whole, quality of life has increased. And b) AI art is not the death knell of artists, because it can not create art intelligently and with understanding as a human can. At least not yet. Case in point, I have tried to get MidJourney to draw me a leaning tower, and a wooden sextant a lot this month, and it seems incapable of drawing either. Had I given this assignment to a human artist, they would have had no problem drawing me what I wanted.

Instead, what AI art is capable of doing, is a) making it possible and affordable for people like me to generate original and appropriate art for at least a number of projects that I would otherwise have just use low quality or poached art for, and b) provide a tool for adapting artists to create proof of concepts, brainstorm ideas, and even draw backgrounds that they can then paint over and perfect to deliver on assignments quicker. Some artists may fall by the wayside, but new ones that are willing to adapt to this new reality will arise, and traditional artists still have plenty of opportunities to nail assignments that the AI is (as of yet) incapable of. The fact that I can now create art with AI without two (or way more) years of practice is irrelevant, both because this advancement in technology can not be undone, and because I actually still can't do the same things an artist with experience can.
 
I certainly agree with a lot of aspects of what you said, and in general is the whole "if lives weren't at stake this would be viewed as a great tool by more". As a proof of concept engine? Yeah, absolutely. As a way to knock out some quick mock ups without having to deal with finding pre-existing works? Neat. As a way to explore styles here-to-fore unheard of? Sounds amazing.

One issue, and distinct difference from the professions you mentioned, is that the court jesters previous jestering wasn't directly used to develop the post-jester. The computer switchboard didn't "steal the switchboard operators works" because they didn't have any persistent works. The issue here is the combination of a replacement technology with the fact that it's exploring uncharted copyright territory on existing artist works. AI art would literally not be able to exist without artists making works in other ways, at least in the form it exists now. And law and regulations and public sentiment hasn't caught up on what it means for intellectual property rights, etc. So a new technology that supplants artists, in theory, by using the breadth of all the work they've already done, is certainly a new combination. And as the tech gets better and can draw sextants and towers (which only take the proper inputs into the algorithm for it to catch on), the issue will become more urgent and stark.

This isn't like traditional to digital art; a new tool artists could use to generate art in their same or similar style. This is a tool random people can use to derive works almost indistinguishable from existing artists work with their work without their input; perversely still requiring that they have put in the work, without providing them the fruits of having done it.
 
Top