As many of us, I have some models being for sale by third party shops without authorization/license.
I found some of then using Google Lens, other by help of other MW users (thank you).
As I’m not capable of, are any fellow member with programming skills to develop an script that can use Google Lens to search for each of the images on a particular profile?
It would be a great tool for looking for not authorized use of designers models.
I’m open for others alternatives too. Any idea welcomed.
Depends how robust and reliable you’d want it to be.
You could scrape all the images on a profile and feed them into a reverse image search, then look at the results and make a decision on what’s pulled back. But… it’s likely MW and the Search engines will periodically throw up a occasional capture, or are already implementing some form of anti-scraping tech, and things will fail.
Paid-for API for images searches exist and are likely to be more robust, but as the name suggests, you’ll be paying to use them. That may make sense if it’s something you’d run frequently or that outlay make commercial sense. Only you can make that judgement,
Alternatively, for infrequent use, you could code something that mimics what you’re going with Google using a browser and just accept that occasionally it’ll fall over.
Rather than scraping a MW profile, if if it’s your own profile you may be better off downloading the images to a local drive and using those - it removes MW from being one of the potential sources of failures.
(1) create a dir with a copy of all your source images. name them the string you want to search for as a match
(2) ask claude to get you a script that will iterate over that dir with reverse image search, getting you the top 20 results. lazy ■■■■■■■■ wont even change your photos. use the string for better matching via LLM vibes, or dont
3a) make it give you a report
3b) make it maintain a small database, so it can prompt you to assign each of those top 20 results a status [ex: ok, unrelated, infringement], track those for future runs, and only report on unknown stuff. first run will be a lot of time analyzing, but after that youll only see new stuff.
It’s probably not worth the effort, because almost every API will block you if you don’t pay for it. It will identify you as a bot, and if you don’t pay, the API will say no.
This might work with just a few 3D models and images, but once you have dozens of them and want to search through them regularly, the API will say, “Time to pay up.”
If you can afford it, use a service like Brandshield.
Nothing existing is accurate enough to trust it to send automatic C&D, DMCA etc. You’re going to have check everything it finds manually unless you don’t care and can deal with the legal consequences of sending false DMCA etc.
Am I the only one that finds it amusing the forum that is staunchly anti-AI when it comes to model creation, suddenly does a 180 turn to become staunchly pro-AI when it goes for writing code.
It’s like “No, don’t create models using AI, that’s just AI slop, no one needs that. Spend dozens of hours learning modeling software instead!”
But “Obviously, use AI to create code, that’s going to be the easiest thing in the world. You don’t need to spend dozens of hours learning to code, just make AI do it for you!”
This may be a good use case for vibe-coding, if you have access to an AI tool like Claude Code.
I’ve had good luck with using an AI agent through claude to search the internet for listings of my models.
ask claude to get you a script that will iterate over that dir with reverse image search
If this was a modeling question, the response would be something like:
You can easily do it yourself, just take a couple of hours to read about bash/perl scripting, and use an available API. Or better yet deploy a local open-source image comparison model and run in on your GPU pre-trained on your image set. It’s always better to learn to code yourself, than churn some AI-slop code.
It’s funny, because he quoted me… and yet I recently had this post, where I learned a lot from this community:
Alex is conflating two entirely separate issues related to using AI… one of which gets tons of support here at the forum because it allows creativity, while the other does not because it is used for dishonest purposes.
After my journey through that thread I linked, I was able to help my mom bring her idea into reality, and improve it in ways that she hadn’t even imagined. She was absolutely ecstatic about it.
Sure seems like that was a big “win” for AI assisted model creation.
That’s exactly the main point, and I mean this in a general way - not directed at any specific users.
The community isn’t broadly against AI; actually, most people just see it as a tool. But users can tell whether someone is using it for personal needs or creative support, or whether they’re using it to flood MakerWorld with low‑effort AI content. The issue is often not the AI itself, but how it’s being used - and yes, people do look closely at public MakerWorld profiles when forming their opinions.
People aren’t generally against AI; they’re mostly against low‑quality stuff - and even then, only when MakerWorld gets flooded with it. In private, everyone can do whatever they like. It’s always important to distinguish between private and public.
