DALL-E 3 Hailed For Endowing Users With Incredible Comic-Making Superpowers
By Mikelle Leow, 02 Oct 2023
Our prompts: “Create a dynamic black and white comic book panel with a bespectacled thin man drinking from a can of soda” and “Create a dynamic black and white comic panel of the same man, this time more muscular, and he’s gained the power of invisibility.” Images generated on Bing
DALL-E’s latest version is saving the day for those who possess the power of imagination but consider execution their liability. OpenAI’s DALL-E 3 is built to understand prompts better and deliver precise results.
While the art-generating artificial intelligence tool is currently in its research phase, AI enthusiasts have discovered that it’s been quietly integrated into Microsoft’s Bing Chat, where it’s available to use for free.
With great power comes a whole world of creation, it seems. Users are learning that DALL-E 3 has a rather supernatural prowess at creating comics, willing them out of thin air (and text prompts) in a flash. Take, for example, product designer Ammaar Reshi, who has dreamed up a mini-story via Bing Chat “in under five minutes.”
OpenAI’s latest image generation model, DALL-E 3, makes it SO easy to create comic books!
— Ammaar Reshi (@ammaar) September 30, 2023
Here are 4 panels for a fan-made Batman comic made in under 5 minutes.
Prompts included in the ALT.
Enjoy!ð¥ pic.twitter.com/9LhZj4I0KW
To put pen to paper, Reshi asked the chatbot to “create a black and white comic book panel with Batman standing looking panicked and afraid in front of his Batmobile in a dramatic way… include a bubble above his head saying ‘ROBIN!’” and “create a black and white comic book panel with a close up of the Joker’s eyes… include a dramatic sound effect text saying ‘HAHAH’.”
For a consistent art style, Reshi advised others to retain the same descriptors across prompts. In this case, he kept the “dramatic” adjective and the “black and white comic panel” phrase.
A couple of things!
— Ammaar Reshi (@ammaar) September 30, 2023
- keep the style consistent by saying “black and white comic panel”
- I picked Batman because I felt like it would’ve been trained on those characters and so it wouldn’t vary too much
- maintain adjectives: use of “dramatic” is consistent in all prompts…
Batman’s presence in pop culture would’ve helped to paint a more harmonious picture too, as it’s likely tools like DALL-E have been trained on copious images depicting the DC superhero.
Compelled by the results, coder Dreaming Tulpa conjured up their own eerie black-and-white comic centered around the AI art illusion trend that’s currently also taking social media by storm. Its look is, admittedly, quite resemblant of Junji Ito’s work.
Inspired by @ammaar's Batman comic, I took OpenAI's DALL-E 3 for a spin to create a short story about the recent spiral craze here on X ð
— Dreaming Tulpa ð¥ð (@dreamingtulpa) September 30, 2023
Used GPT-4 to generate the image prompts (check ALT).
The models ability to understand text and create readable text is super impressive 𤯠pic.twitter.com/vLMFOkzHO7
Separately, AI educator Chase Lean decided to inject more color into his comic. Instead of crafting panel by panel, though, he got the chatbot to create an entire comic strip “in which a boy discovers a mysterious egg that hatches into a baby dinosaur.”
OpenAI just dropped DALL-E 3 a few hours ago, and it's a game changer for making comics.
— Chase Lean (@chaseleantj) September 30, 2023
I made the comic strip below in less than 15 mins.
Here is my process: pic.twitter.com/09HfYbMqhO
There are pros and cons to this approach. It does save time, but the story might not play out in the right sequence. Fortunately, you could easily sort out the panels on an image editing software, Lean noted.
AI also isn’t great at deciphering text, which means dialogue could sometimes read as gibberish. That’s another hiccup that can be solved using apps like Photoshop.
Step 3: Add text to the speech bubbles
— Chase Lean (@chaseleantj) September 30, 2023
DALL-E 3 can make text, but when there are too many words it can fail.
If the words don't make sense, just replace them with whatever text you want.
Here's my final image!
10x this & in theory you can create a kids' comic book. pic.twitter.com/mrccAt5lpS
Although DALL-E is to use on Bing Chat, generations are limited to two comic panels per conversation, which will make composing an entire graphic novel pretty tricky. You can, however, initiate a new chat to create more, but the output might not be as consistent. To launch the art-making model, you’ll also need to switch to the GPT-4 version.
OpenAI just dropped DALL-E 3 a few hours ago, and it's a game changer for making comics.
— Siddhant Adlakha (@SiddhantAdlakha) October 2, 2023
I made the comic strip below in less than 15 mins.
Here is my process: pic.twitter.com/waqamuSNX1
The generator’s ability to cook up quality artwork in a jiffy will likely proliferate concerns about copyright infringement. To that end, OpenAI underscores that DALL-E 3 is engineered to turn down requests mentioning the names of public figures or the style of a living artist. With that said, as proven in the outputs above, Batman-themed images appear to soar above these guardrails.
Additionally, to prevent disinformation, all artwork generated on Bing Image Creator will also retain an invisible watermark embedded with its origin, timestamp and date stamp.
On the heels of Bing, DALL-E will also make its way to ChatGPT for Plus and Enterprise customers to turn text into masterpieces this month.
Ok, Stop the presses! We can now visualize our ideas in a fun comic strip/storyboard style! DALL-E 3 is next-level stuff for this and could change the game for quick storyboarding. Being a visual thinker, I often mentally sketch ideas as comics in my head. Now, I can share them.… pic.twitter.com/Uk7eSw7oXT
— Thomas Alan Smilie (@tasmilie) October 1, 2023
DALL-E 3ã¯ã³ãå²ããã漫ç»ãæãã¦ãããã¨ç¥ã試ãã¦ã¿ã¾ããã
— ã¢ã·ã«ã¬ã³ã¦ã¸ï¼ãã¸ã¿ã«æ´»ç¨Podcast (@ashikagacast) September 30, 2023
ããããããã³ã®ã³ãå極ããæ¥æ¬ã«ãã£ã¦ãã漫ç»ã
ããã³ããã«comic stripã¨ããè¨èãå ¥ããã®ããã¤ã³ãããªãDALL-E 3ã«å¯¾å¿ãããããBing Image Creatorã§çæãã¾ããã pic.twitter.com/wA28fTb0yY
OpenAI’s latest image generation model, DALL-E 3, makes it SO easy to create comic books!
— Abhay Vinayak (@abhayvinayakiaf) October 1, 2023
Here are 2 strips of 6 panels each for a fan-made Baby HULK beating Thanos made in under 5 minutes.
Prompts included in the ALT. Enjoy!ð¥#chatpgt #DALLE3 #Hulk #Marvel #comicstrip pic.twitter.com/LOFV9fl6Xt
[via X (1)(2) and MSPoweruser, images via various sources]