Realistic renderings of some Legendary humanoid cards in Splinterlands

in #sps2 years ago

Sadly for me, I am not an artist. However, with the recent boom in the capabilities of AI tools, I've been able to play with the creative visual arts more, which has been a dream for me as deep inside I've always wanted to be an artist. In this post, I took four beautiful Legendary cards from the Chaos Legion set and used Midjourney to try to render some more realistic versions of each one. Enjoy!

Runemancer Florre

image.png

Although I was able to get more "real" looking versions of Florre, I absolutely love how this one turned out. If you look closely, there are still some issues with the fingers, although v5 of Midjourney has MASSIVELY improved on that front. Also I sadly wasn't able to get the nice little birdy in this image, however I feel the detail and coloring turned out perfectly other than that.

Runemancer Atuat

image.png

This one was almost an immediate success. Somehow Midjourney took one look at Atuat and said - here's the guy! Everything in this photo seems perfect to me, and Atuat looks completely realistic.

Lira The Dark

image.png

Lira was a bit harder to get as hyperrealistic as Atuat, despite many attempts. When I tried to get her smiling or "arrogant" her face would turn out strange and unconvincing, so I kept one that felt like it got the majority of her other features, as well as the coloring, right. However I feel I missed the expression and "tone" of the original card.

Runemancer Kye

image.png

Kye was by far the hardest to get right (and there are various bloopers to show that below). I think MJ got confused by the variety of features in the original card (the fact she's a siren, has demon wings, the colors and tone being in-human despite many human features. I landed on this one which I feel gets the expression and "mood" rather well, though it's missing features like her signature staff and frankly the "wings" look more like decorative swirling dragon tails.


Bloopers

With Midjourney, it's all about experimentation and lots of re-rolling. While looking for the right prompts to generate the images above, I got lots of less satisfying (and sometimes hilarious) results.

Here's Lira looking like the mom in Addam's Family. "Polaroid" was something I found as a suggested keyword to get realism but in this case I got a weird retro feel that didn't work.

image.png

In this following one, I'm not sure how Lira ended up wielding a giant pencil... The pen is mightier than the sword I guess?

image.png

It was really difficult to get Runemancer Kye to turn out properly. Lots of funky arms and legs, wrong tone/expression. This following set was close but each image has problems (weird stumpy arms, giga-shoulder in another, missing leg in the third...).

image.png

Here's a hilarious set, where I tried specifying the type of camera I wanted used for the shot (which is frequently recommended by MJ "prompt experts"). Unfortunately in this particular case it took the camera literally, and inserted it right on Kye...

image.png

Here's a blooper of Atuat where I described him as a "Grizzly Warrior" and MJ apparently took the word "Grizzly" to heart, offering me a bear in return.

image.png

And finally here's a different set of attempts for Kye, where I tried to pack the query with words that would indicate a realistic portrait. I felt these ones took it too far, and I couldn't recognize Key anymore. For reference the prompt on this one was "soul, soulful portrait, spirit, slightly evil, shadow overhead, person, real, young female portrait, long bright red hair, dark purple eyes, running mascara, smeared lipstick, purple skintone, personal portrait, smoke, underwater, chest level, silver chains, dramatic light, dreamlike light, photorealism, ultra detail, 100mm, f 1.2 --ar 6:9 --q 2 --v 5"

image.png

Hope you enjoyed these and are enjoying AI as much as I am! Despite all these incredible capabilities, I still wish I could draw with my own two hands, but given my lack of artistic skills AI will have to suffice for me :)

Sort:  

Midjourney has gotten a LOT better since I last used it. Sheesh!

Thanks for sharing! - castleberry#6859

Me gustó bastante la de Lira, con un estilo sombrio y a su vez elegante

These turned out really cool! Thanks for being upfront about using AI and sharing your prompts. I have tried a few runs to make some SL characters and didn't have much luck. To be honest, i didn't try too hard. I will say, it would be better if you added the card image to show the contrast of character results from the originals.. for those less familiar with who you are naming. Also, i don't know if negative prompts work on Midjourney... but if they do, you can try [deformed hands, too many fingers, etc.]
Whatever you don't want or aren't liking... just put inside brackets and it might help smooth things out a bit. Cheers!

Congratulations @bravetofu! You have completed the following achievement on the Hive blockchain And have been rewarded with New badge(s)

You made more than 200 comments.
Your next target is to reach 300 comments.

You can view your badges on your board and compare yourself to others in the Ranking
If you no longer want to receive notifications, reply to this comment with the word STOP

To support your work, I also upvoted your post!

Check out our last posts:

The Hive Gamification Proposal
Support the HiveBuzz project. Vote for our proposal!