First, a huge shoutout to @CaylaCatz . I wouldn't have had the first clue how to do a comic strip without her brilliant guides. Second, thank you to everyone who has liked or commented on my Steamforged Robotics and V.E.N.O.M work. "So, tell me oh Great Oracle", you all say, "How does one make a comic strip?" First, I would recommend considering a reason for a character or concept to exist. Sure, you can make a comic about a cute cat, or a Bioengineered Hamster, or IDK, a psychopathic, murdering budgie. But if you can think of a personality trait, or some kind of experiment gone wrong, then IMHO it really helps when developing a storyline. Here is the Prompt for the Cover image : Imagine a steampunk style victorian villainess, head of an Evil organisation called V.E.N.O.M,she is a redhead beautiful busty refined aristocrat who despises the way men belittle women, she is strong , independent and fierce and has a genius level IQ, she has unpinned flowing hair, low cut top, right eye replaced with an embedded steampunk lens,show her in her office,world map and library shelves in the background,, chart on the wall showing detailed goals for replacing men in all levels of society,The lighting is soft and diffused, characteristic of a 19th-century glass studio, casting a gentle vignette at the edges. Wet-plate photography aesthetic, subtle authentic film grain, daguerreotype quality, historical realism, 8k resolution This was done in grok-image-imagine To achieve the look, I just asked AI on Google how to prompt for a Victorian-style photograph and added that in at the end. As for the name, Baroness Morwenna Scrivener again, good 'ol AI came up with the answer when I asked for Good Steampunk Villainess names. Creating the Comic Strip Method One Gpt image 2 is, imho the absolute best for interpreting your ideas and adding in amazing little details. When I tried the first test image, I forgot to add what V.E.N.O.M was an acronym for, and I am both mad and impressed that it came up with Victorious Empowered Network of Omnipotent Matrons . My original idea was Very Evil Nemeses of Men . Damn you, AI, damn you! I have found through a bit of experimenting that less is in fact more and generally speaking, GPT nails it on the first attempt. The K.I.S.S ( K eep I t S imple S tupid) principle really is the best. Prompt using Image to Image comic book panels, maintain steampunk style aesthetic and palette, Baroness Morwenna Scrivener, head of V.E.N.O.M, which stands for Victorious Empowered Network of Omnipotent Matrons, explains her plan for replacing men at all levels of society by all any means necessary, end with a tounge-in-cheek jokes about how they would hardly be missed as women pretty much run the world anyway Ta-Da.. Method Two GPT is also pretty good with Text to Image by again using the K.I.S.S principle. Cartoon panels, title "Progress", subtitle "Then", a gamer playing Space Invaders on an Atari VCS subtitle "now", a gamer clicks a icon on desktop, then a notice on screen saying updating with percentage and estimated time, then download complete, then update found and another notice saying updating and percentage and estimated time, panel of gamer asleep in gaming chair Drum Roll.. So in conclusion... Honestly, I am totally going with "unbispired" as my new go-to word :)
Tags: comic books, gpt image 2.0, grok
Comments (31)
monkeydude69
Yeah, I did the same before I read CaylaCatz's guide. I was describing each panel and the dialogue seperately which led to either the wrong character saying something or a panel with nothing in it, and then one panel filled with dialogue. Less is definitely more.
monkeydude69
Just one, the main image on the header.
DavidP
SHort and sweet, excellent post @monkeydude69
dezie
Loved it! Well done!
pikpoo
Dropping this in for the technical registry loop. If you've been fighting with background crowd subjects melting into terrifying zombie artifacts, itβs usually an asset-density allocation glitch in the base model. I just mapped out the exact workflow to lock human proportions and force distinct, sharp silhouettes across high-occupancy frames: https://budgetpixel.com/blog/breaking-the-hivemind-how-to-render-massive-crowds-without-the-zombie-glitch Let me know if you run into any token conflicts implementing the layout boundaries.
MagnusGibson
I worked drawing comics ( professionally) in the 80s and 90s. And I used to draw them since the sixties. I even gave classes for 17 years. The AI evolved a lot , and I think that it is a great tool to save time for traditional artists. In my personal opinion, it still needs to improve to be really good.
JMN
clap icon has a dumb timeout window. iirc we need to click once on icon, then wait a beat, then it's okay to click again. spamming click and not moving the mouse away from the clap icon will glitch it and not count them properly
monkeydude69
Look at the first panel on the last comic :) . It came out so well for such a short prompt that I just decided to go with it π
JMN
Hmm wow, I assumed that was the case, but even so...
Makes me think I've been over engineering my approach to comics/graphic novel style altogether...
JMN
wow I'm surprised it came up with that from such a compact script. how many ref images did you use for VENOM?