I actually came across this tool two months ago, but I never clicked on it.
To be honest, the function of colorizing black-and-white photos seems like a toy that doesn’t make much money no matter how you look at it.
But the other day, I saw it being wildly recommended on Reddit again, with the title: “The most natural old photo coloring tool, bar none.”
I just clicked in to try it out – and the result was a bit shocking.
Product Address👉 https://palette.fm
A product that only restores old photos is actually making money
Its core functions are : AI automatic coloring and style transfer
The technical architecture is speculated to be an image generation and color transfer model based on deep learning, which may call self-developed or open-source diffusion models (such as Stable Diffusion) and perform accelerated inference on cloud GPUs.
I dug into the data, and surprisingly, the monthly traffic is nearly 370,000 (it was launched only 3 years ago in 2022)
In other words, for such a small feature, it is highly likely to earn more than $3,000 to more than $30,000 per month.
The business model is also straightforward:
Free use allows you to generate previews, but if you want high definition downloads, batch processing, or API access, you need to pay.
It doesn’t make money by showing off skills, but rather by stable, smooth, and mindless operation, squeezing out the users’ “emotional value”.
Who’s paying? Emotional users + content teams, all paying quite happily
Restoring old photos sounds like something for “old people”, but after actually experiencing it, I found that:
- Some people use it to retouch their grandparents’ photos
- Some people use it for wedding souvenir albums
- Some content creators directly use this as the opening of a vintage video
- Even museums/film and television restoration institutions have left messages on Reddit saying that their teams are using this
This is quite telling: It is not a toy, but a tool, only that this tool looks very gentle.
So I think although this tool is small in scale, it has opened up a highly stable niche market
There are indeed similar tools on the market, such as the open-source version of DeOldify, Hotpot AI, and MyHeritage In Color, all of which can perform colorization.
But the reason Palette can succeed is that it doesn’t create a “functional toolbox”, but instead, it says little and only gives you the most beautiful.
It’s not about the volume function, nor the algorithm, but the delicate sense of the volume experience . This is exactly what I think makes this tool “amazing”.
If I were to do a similar project now, I would choose these two directions to approach it.
1️⃣ The batch repair tool is for B-side image restoration companies. Batch upload, style locking, and task management are not currently implemented, but the task of restoring old images already has a stable demand among cultural heritage, imaging, and documentary teams. If done well, it could be a project worth $5,000 per job.
2️⃣ Create a template tool for turning old photos into short videos, designed for TikTok creators Imagine this: upload a photo → automatically add subtitles → automatically add music with a sense of the era → automatically generate a vintage video intro. Do you think this thing can go viral? I think it can.
Echo’s Judgment
This is not a “stunning” product, but a product worth copying.
It uses a seemingly simple feature with no obvious technological barriers to take the “user experience” to the extreme, and finally achieves monetization through “emotion + speed”.
It is not in the popular track, does not develop complex systems, nor tell grand stories, but it can generate revenue, and steadily at that.
To be honest, I like this kind of product so much.
Do it professionally, do it smoothly, make it monetizable, and avoid being flashy. If I were working on a side project, I’d rather create a Palette.fm than compete for the next Claude.
🍊 I’m AI Blue-Haired Witch Echo
8-year Product Manager, specializing in AI × Going Global × Toolkit Practice
Test products daily and deconstruct business logic using methodologies
Only focus on those AI products that have completed the whole process and are worthy of reference
📬 If you’re also looking for the next money-making AI direction, welcome to follow me
📬 Overseas / Tools / Agent direction, welcome to add my WeChat 👉 echoaipm
Echo AI PM