Seems like it’s solely Makerworld’s responsibility to identify and ban these accounts. With all the AI powered tooling available these days, it should be much easier than in the past to detect them, assuming they care enough to do so.
2 Likes
AI is good at finding known patterns, this was about discovering the new patterns, what “feels” wrong and worthy of investigating.
Yeah ^^` because as for now, they are relying heavily on user reports to spot those abuse.
1 Like
I think that would be the idea though - you train it with a bunch of known accounts who are not misbehaving, then have it identify anything that an anomaly flag for manual review.
Honestly though you don’t need AI to detect a lot of the obvious weird things being pointed out in this thread. They just need to make it a priority if they want to do something about it.
1 Like
At some point AI should step in to sort out fake profiles, point farmers but also everyone uploading print profiles without real pictures. It seems that MW is too generous and some non legit people are taking advantage of it.
3 Likes
Don’t give people ideas, it’s pretty bad as it is 
ha good point, i’ll just delete that then
1 Like
Not really possible on here, sadly. But yeah, that’s how they do it. I also believe some are cancelling prints partway. It lets people review it, but then they can’t rate it higher than 4 stars because it didn’t finish. Thus negative reviews with no explanation. Not completely positive about that, but it seems the most likely to me.
I had one guy like that today. But the comment/rating included a picture. I’m wondering if they are not taking a picture already send by someone else. (And I have way too many comments on this model to check if it’s true)
Perhaps reverse image searching it? Tried on one of my reviews but it wasn’t able to pull the original review image up. Got the actual model though 
I tried that too 
, but nothing come up. So it’s easier to just assume that it’s a bug on BL side.
1 Like
Since this topic is still open…
My wife was searching for a specific model to print and came up to this guy. I was in the couch and the wife asked me if i knew a certain model that the guy had. As soon as i got in his profile i saw that almost everything is a “remix” of other models but without any change at all. 13k downloads until now for downloading models from printables and thingiverse and uploading them here, stating that “changes were made”. Opening his models and the then original ones show no changes. They are all exactly the same…
https://makerworld.com/en/@maxerek
This is what i call point farming at it’s finest. 13k downloads and no one noticed 
So you can simply steal models and add them as remix. Cool! 
BTW I bet your wife was looking for the Bat-Cat mask 
4 Likes
Just an fyi, early adopters of MW were allowed to upload models as a “share” (vs a remix) which was literally reposting from other repositories. I don’t know if those are classified as shares still or remixes now, but it’s a possibility of how that was rampant for a long while. It’s absolutely not allowed today, but there’s still users who have a lot of these. By the time I got on MW, I didn’t understand what “share” meant when uploading because it was grayed out, but it’s just gone now.
Actually yes, it was 
That’s why i got into this profile, because we had already printed one years ago and the model was the same, comparing to the original
Checking the release date of the first model “remix” he uploaded, nah, it was already a no go at the time. That means that this guy was really stealthy not to get noticed. And it worked, it seems.
Released 2024-02-14
Another one, but this one got caught as it can be seen on the comments. And why is makerworld still allowing it when is clear that the user is even selling the product on his socials like it is his own creation? That is the question i would love to see answered…
2 Likes
I thought - there’s no way no one noticed! Then I checked original batman cat mask vs bat-cat “remix” and sure enough: they’re different! The original is bigger and more detailed… But then I wondered “what if it’s scalled” - it is. Scale the original to 40% in bambu studio and the dimmensions are identical. Simplify the original model and I bet you’ll be able to hit the vertex count, but I didn’t want to play with that and “spray and pray” so I’ve looked for a way to compare STLs and sure I’ve found one: CloudCompare. There’s a bit of learning curve to it… But I’ve managed to load original and the “remix”, hit align meshes and boom: 100% overlap and calculated scale is 0.39995 so hit nearly perfectly my 40% 
I don’t have time to play further and learn how to actually confirm whether the meshes are identical and/or what simplification brings original to remix but for me it’s clear: that’s not a remix in any way that agrees with MW’s remix guidelines.
4 Likes
That’s the hard way to do it, but yeah, you got my point. Now please explain how the heck does makerworld let’s a person get so many downloads and not even move a straw? This i cannot understand but since i have been in previous discussions about this same aspect, now i start to understand how it works and when does makerworld close an eye to it… starting to see some patterns everywhere i look
1 Like
I went to look for some easy info on how to compare simplified meshes in CloudCompare and there’s distance tool. So I’ve looked into it and the mean distance is 0.00023 with std dev of 0.002. So it’s “identical.” Last thing would be to calculate decimation ratio to go from one to another
//EDIT:
Decimation ratio: 93.6%
After that (scale and simplification) there’s so little difference it’s indistinguishable.