Just adding dialogue to AI illustrations. The story of how I made my own 'Speech Bubble Maker'
When you're generating AI illustrations with ComfyUI, don't you ever think, 'If I added some dialogue to this character, it would feel like a real piece of work'? Viewing them as a single image is fine, but adding just one speech bubble creates a sense of 'timing' or a 'punchline' like in a 1-panel or 2-panel comic.
You see this a lot on X and similar platforms, right? Especially the ones that hint at something a bit suggestive lol
That said, firing up Photoshop or high-end image editing software just for that is overkill, and using a smartphone app means leaving the PC environment where you're working with ComfyUI. So, I built my own browser-based tool that focuses solely on 'adding speech bubbles to generated illustrations.' This is an introduction to that tool.
What I made: 'Speech Bubble Maker'
It's a free, lightweight, and snappy web app that lets you add speech bubbles and text to illustrations to create simple 1-panel or multi-panel comics. It requires no installation, runs entirely in your browser, and doesn't process image uploads to a server, so you can edit images generated in ComfyUI directly within your browser.
▼ Here is what the UI looks like.
I created this tool out of the feeling that 'it's a waste to just post an illustration as-is when I've managed to output such a great expression.' Just adding a line of dialogue or an inner thought can significantly change how an illustration is perceived.
▼ It's free for anyone to use, so please give it a try (bookmarking is recommended).
What it can do
The main features are simple.
Open images (drag & drop is also supported)
Select and place speech bubble shapes (9 patterns: text only / oval / cloud (thought) / square / burst / rock / hedgehog / speed lines)
Double-click to immediately switch to text input
Drag to move, use the squares at the corners to resize, and use the red circle to adjust the direction and length of the tail
Fine-tune colors, line thickness, fonts, etc., in the right panel
Supports creating multiple pages (useful when you don't want to stop at one panel and want to create a 2-panel or 3-panel exchange)
PNG export (batch export of all pages is also possible)
For example, I envision using it to take 'two illustrations generated with different expressions' and adding dialogue to each to create a 2-panel comic. You can use it for exchanges like a straight man and a funny man, or just for light posts where you add a single comment from the character.
However, please note that you cannot create panel layouts within a single page!I personally want a tool that can handle panel layouts, so I'll share it again if I come up with something good.
Also, while it can technically be used on a smartphone, it's honestly difficult to operate, so a PC is recommended.
About saving layouts
Speech bubble layouts currently being edited can be saved and loaded in JSON format (you might not be familiar with it, but it's just a simple text file). However, there is one point to note: JSON does not include the background image itself (because including images would make the data extremely heavy). Since it is a mechanism that only saves the 'speech bubble placement information' such as position and text, the speech bubbles will be restored when you load the JSON later, but the background image will be blank. I have set it up so that the file name of the image that was placed is displayed in the message when loading, so you will need to re-place the same image.
Conversely, this specification is convenient when you want to reuse the same speech bubble layout for a different cut of an illustration. It is useful in situations like, 'I like this panel layout, so I want to use it with the same placement for the next illustration I generate.'
Summary
I created a browser-based tool to turn AI illustrations generated with ComfyUI into 1-panel or 2-panel comics by adding speech bubbles.
No installation required, and images are processed locally, so it's easy to use.
You can also create panel-based interactions with the multi-page feature.
Note that while layouts can be saved as JSON, background images are not included.
Shortcut keys and basic operations are also summarized on the 'How to Use' page within the app.
It is quietly fun to see the expressions in the illustrations stand out just by adding a line of dialogue, rather than just finishing after generating them. I hope this is helpful for others who are working with AI illustrations in the same way.
Bonus
I have published comics created with this tool with different variations on the following two sites. Please feel free to check them out if you like.
They are a bit suggestive, so please skip them if that's not your thing.
▼Poipiku (For followers only)
▼Pixiv
#ComfyUI #AIImageGeneration #AIIllustration #SelfMadeTool #WebApp #GenerativeAI #note #macOS #SpeechBubble #Manga