DEV Community: 吴sir

Introducing Retouchia – The Best AI Image Editor

吴sir — Sun, 20 Sep 2026 01:54:26 +0000

Why AI Photo Editing Still Feels More Complicated Than It Should

AI image editing has improved dramatically over the past few years, but actually using it can still feel more fragmented than it should.

A typical photo-editing workflow might look like this:

one tool to remove a background
another tool to upscale an image
another one to remove an unwanted object
a different app to restore an old photo
and yet another tool to change a hairstyle, outfit, or scene

None of these tasks are especially unusual anymore.

The frustrating part is having to move between several different tools just to finish one image.

The real problem isn't always the AI model

When people discuss AI image editing, most of the conversation tends to focus on model quality.

Can it preserve faces?

Can it generate realistic clothing?

Can it restore missing details?

Can it upscale an image without introducing strange artifacts?

Those things matter, but there is another problem that gets less attention: the editing workflow itself can become unnecessarily complicated.

For many users, the desired result is actually very simple.

Remove this background.

Clean up this photo.

Change this outfit.

Restore this old picture.

Most people don't want to learn the terminology behind every model or editing technique.

They simply want to describe the result they need and get a usable image.

A simpler workflow

That suggests a much simpler interaction:

Upload an image
Choose or describe what you want to change
Let the AI process it
Compare the result
Download the version you want

The problem is that keeping this experience simple becomes harder as more editing capabilities are added.

If one editor supports background removal, image enhancement, restoration, object removal, outfit changes, hairstyle changes, background generation, and other transformations, the interface can quickly become a wall of buttons.

So the challenge isn't only adding more AI features.

It's deciding how those features should be presented without making the product harder to understand.

Users don't always think in technical terms

Developers and users often describe the same task very differently.

A developer might call something image inpainting.

A user might simply say:

remove the person standing behind me

A model may technically be performing segmentation and generation.

The user just thinks:

change the background

That difference matters when designing an AI product.

People shouldn't need to understand how the underlying technology works before they can edit an image.

Natural-language instructions often match the way people already think.

For example:

Remove the background and keep the product unchanged.

Change this outfit to a black business suit.

Restore this old photo and reduce the visible scratches.

The easier it is to describe the desired result, the less time someone has to spend figuring out which tool or feature they are supposed to use.

Bringing different editing workflows together

I've been exploring this idea while building Retouchia, a browser-based AI image editor that brings several common photo-editing workflows into one place.

You can see it here:

https://retouchia.com/

The idea isn't to create a separate experience for every small editing task.

Instead, I'm experimenting with a single editor that can handle things such as background removal, object removal, image enhancement, image upscaling, old photo restoration, black-and-white photo colorization, hairstyle changes, outfit changes, and background or scene changes.

The interesting part isn't simply how many features can be added.

The harder question is how to keep the experience understandable as the number of available editing options grows.

Building around outcomes instead of technology

I'm increasingly convinced that AI image editors should be organized around what someone wants to accomplish rather than around the technology performing the task.

Instead of showing users terms like segmentation, inpainting, super-resolution, or image-to-image generation, the interface can focus on much simpler actions:

Remove something
Improve the image
Change something
Restore something

The underlying AI model can stay in the background.

For most users, the result is more important than knowing exactly which model or technique produced it.

More features don't always mean a better product

AI products make it very tempting to keep adding features.

A new model becomes available, so another button gets added.

Then another.

Eventually, an editor that was supposed to make image editing easier starts looking just as complicated as traditional editing software.

That's the tradeoff I'm currently thinking about the most.

At what point does flexibility become complexity?

Should users see every available editing option, or should the product try to understand their intent and guide them toward the right workflow?

There probably isn't one answer that works for every product.

But I suspect the most useful AI tools will eventually make the AI itself feel less visible.

Instead of thinking:

Which AI model should I use?

people may simply think:

I have this image. I want this changed.

And the product figures out the rest.

Still experimenting

Retouchia is still something I'm actively improving, and I'm interested in how other developers approach the same problem.

If you're building or using AI image tools, do you prefer separate tools for individual tasks or one editor that handles multiple workflows?

Do you care which AI model is being used?

At what point does adding more features make an editor harder to use?

Would you rather select a specific tool or simply describe the result you want?

I'd be interested to hear how other builders are thinking about the balance between flexibility and simplicity.

I Needed a Faster Way to Make VTT Tokens for DnD, So I Built a Local-First Token Maker

吴sir — Tue, 19 May 2026 09:57:28 +0000

I built a simple token maker for my DnD and VTT workflow
Tags: dnd, webdev, gamedev, tooling
Making tokens for a tabletop campaign should not take that long.
But if you run DnD games, you probably know the small routine:
• find character art
• open an image editor
• crop the face or body
• add a border
• resize it
• export a PNG
• repeat it for every NPC, monster, boss, and player character
Doing this once is fine. Doing it for a full encounter is annoying.
That is why I built a small browser tool for this workflow:
Token Maker: https://www.tokenmaker.one/
It is a free VTT token maker for DnD, Roll20, Foundry VTT, and Owlbear.

The problem I wanted to solve

Most image tools are too broad for this job.
They can make tokens, of course, but they are not built around the actual tabletop workflow.
When I am preparing a session, I usually do not want a full design tool. I just want to:

upload character art
place it inside a frame
adjust the crop
export a transparent PNG
move on to the next token That repeated flow is the real problem. The tool does not need to be complicated. It just needs to remove the boring steps. ## What the tool does Token Maker is focused on VTT token creation. The current workflow is simple:
Drop an image into the editor
Choose a border or style
Adjust the image scale and position
Pick an export size
Download the PNG token It supports different token shapes and borders, including circular, square, and polygon-style tokens. The export is a transparent PNG, which works well for virtual tabletop platforms. ## Why I made it local-first For this kind of tool, I wanted the workflow to feel quick. You should be able to open the page, drop in an image, make the token, and export it. • No heavy setup. • No long design process. • No need to turn a small token task into a full graphics project. That is also why the editor is built around one main workspace instead of spreading the workflow across several steps. ## Where this helps I think this is most useful for: • DnD character tokens • NPC tokens • monster tokens • boss encounter tokens • quick campaign prep • Roll20 and Foundry VTT games • token pack preparation It is especially helpful when you need several tokens at once. One token is not a big deal. Twenty tokens is where the small time savings start to matter. ## A few details I cared about ### Fast preview When making a token, you need to quickly see if the face, weapon, or character pose fits inside the frame. A slow preview makes the whole process feel worse. ### Different border styles A plain ring works for many tokens, but not every token should look the same. A rogue, a mage, an undead creature, and a dragon boss usually need different visual treatment. ### Transparent PNG export This matters for VTT use. Transparent PNG tokens are easier to place on different maps and backgrounds. ### Simple controls The tool should not feel like Photoshop. For most token work, scale, position, border, style, and export size are enough. ## What I learned while building it Small tools are often useful because they are specific. This is not trying to replace a design app. It is just trying to make one repeated tabletop task faster. That is the kind of product I enjoy building: not huge, not over-designed, just focused on one workflow that people actually repeat. If you make tokens for DnD, Roll20, Foundry VTT, or Owlbear, you can try it here:Making tokens for a tabletop campaign should not take that long. But if you run DnD games, you probably know the small routine: find character art open an image editor crop the face or body add a border resize it export a PNG repeat it for every NPC, monster, boss, and player character Doing this once is fine. Doing it for a full encounter is annoying. That is why I built a small browser tool for this workflow: Token Maker | https://www.tokenmaker.one/ It is a free VTT token maker for DnD, Roll20, Foundry VTT, and Owlbear. The problem I wanted to solve Most image tools are too broad for this job. They can make tokens, of course, but they are not built around the actual tabletop workflow. When I am preparing a session, I usually do not want a full design tool. I just want to: 1.upload character art 2.place it inside a frame 3.adjust the crop 4.export a transparent PNG 5.move on to the next token That repeated flow is the real problem. The tool does not need to be complicated. It just needs to remove the boring steps. What the tool does Token Maker is focused on VTT token creation. The current workflow is simple: 6.Drop an image into the editor 7.Choose a border or style 8.Adjust the image scale and position 9.Pick an export size 10.Download the PNG token It supports different token shapes and borders, including circular, square, and polygon-style tokens. The export is a transparent PNG, which works well for virtual tabletop platforms. Why I made it local-first For this kind of tool, I wanted the workflow to feel quick. You should be able to open the page, drop in an image, make the token, and export it. No heavy setup. No long design process. No need to turn a small token task into a full graphics project. That is also why the editor is built around one main workspace instead of spreading the workflow across several steps. Where this helps I think this is most useful for: DnD character tokens NPC tokens monster tokens boss encounter tokens quick campaign prep Roll20 and Foundry VTT games token pack preparation It is especially helpful when you need several tokens at once. One token is not a big deal. Twenty tokens is where the small time savings start to matter. A few details I cared about Fast preview When making a token, you need to quickly see if the face, weapon, or character pose fits inside the frame. A slow preview makes the whole process feel worse. Different border styles A plain ring works for many tokens, but not every token should look the same. A rogue, a mage, an undead creature, and a dragon boss usually need different visual treatment. Transparent PNG export This matters for VTT use. Transparent PNG tokens are easier to place on different maps and backgrounds. Simple controls The tool should not feel like Photoshop. For most token work, scale, position, border, style, and export size are enough. What I learned while building it Small tools are often useful because they are specific. This is not trying to replace a design app. It is just trying to make one repeated tabletop task faster. That is the kind of product I enjoy building: not huge, not over-designed, just focused on one workflow that people actually repeat. If you make tokens for DnD, Roll20, Foundry VTT, or Owlbear, you can try it here: Token Maker - Free VTT Token Maker | https://www.tokenmaker.one/ I am still improving the border styles, token templates, and batch workflow. If you run VTT games and have suggestions, I would be interested to hear what would make token prep faster for you.

Publishing checklist
Paste the title into DEV title field.
Paste the article body into the editor.
Use tags: dnd, webdev, gamedev, tooling.
Keep only the two natural Token Maker links already included.
Preview once before publishing.
Token Maker - Free VTT Token Maker: https://www.tokenmaker.one/
I am still improving the border styles, token templates, and batch workflow.
If you run VTT games and have suggestions, I would be interested to hear what would make token prep faster for you.

my site

吴sir — Tue, 05 May 2026 08:05:26 +0000

This is my game site: https://www.samuraisudoku.pro 
This is my tool site: https://www.tokenmaker.one