Introducing improved community ratings system

I'll be honest here - the rating game is dumb. Just rated a few products and saw this:

https://www.cgtrader.com/3d-models/architectural-details/decoration/mum-al-ma - average rating of 4, I gave it 1 because it's just one image, I can't even understand what it's supposed to be and it is just a useless product (product page just proves me right).

www.cgtrader.com/3d-models/furniture/lamp-light/sigma-desc-lamp - average of 8, I gave it 3, because it has only 1 preview, is very simplistic, does not show wireframe - useless product, can be replicated in minutes.

https://www.cgtrader.com/3d-models/plant-tree/other/pumpkin--3 - average of 7 (gave it 2) - same issues.

So how about some basic guidelines? I.e.:
- no wireframe -2,
- no closeups -1 for basic model, -2 for more complex,
- less than 5 previews -1,
- 1 preview -5 for basic model, -8 for complex,
- can't even understand what it is -9,
- model so simple that it should not even be sold -3,
- etc.

How about a minimum of at least 3 preview images to be able to publish a model in the first place? As I see it, a good (from what is shown) model with one preview should not be purchased (cat in the bag), a model so simple that it can be properly represented with one image should not even be on sale.

This is one of my issues with Turbosquid's version of this as well. I see the value in it but it's also the issue you run into when you have a wide range of ability levels.

Some things that I see as a mid quality product (for whatever reason. Lack of images, quality in general, etc.) could be someone else's highest ranking.

The way I think squid fixes this to a degree is to show the average ranking for a piece after you've voted. It shows how off the mark you are on a piece and give you a bit of feedback if you should be more hard/raise or lower your standards.

But even then you are leaving quality rankings up to an unknown pool of people. It's still a good initial filter though at least.

I think so many guidelines like Ricardas mentioned may be hard to enforce unless you break each piece up into multiple rankings then average them. For example a piece would have, instead of a 1 to 10 ranking alone, a list of rankings from image quality to model in general, etc. then average those up and there is the pieces final rank from that user.

BUT I think restrictions on model uploads is a fantastic place to start and remove some of the guess work. Requiring x number of images, etc. Then you wont have to vote on those and you could probably narrow down each piece to just a few key things to vote on.

If a model came up and you voted on core things like 1 to 10 model quality, 1 to 10 presentation, 1 to 10 would you purchase this item

something like that. Just tackling it from both sides with upload restrictions and then voting on multiple facets of the upload instead of one overall vote.

Either way, it's great to see you guys improving the site. I know this kind of system is a lot of work and getting criticism in the first few days sucks as a lot of it is normally knee jerk reactions to something new. Still, info is info. Thanks guys!

I agree with Ricardas.
Why do people upload a product and try to sell it with only 1 image and no description? At least have a simple "This is a model of a (something), which is great for use in (fantasy, sci-fi, or whatever) images."

However, I disagree with models that are considered "simple". Sometimes it's quicker for someone without much modelling skills to buy a "simple" model than to make it from scratch.
Then again, the seller could also add some tweaks to make the simple model more advanced. For example, a "simple" lamp could have a ray-traced metallic surface or a luminous/ ambient bulb or a fancy lampshade.

Thank you for your feedback. That's another good point that should be fixed moving forward. We are considering two possible solutions:

- Not allowing models with less than X models to participate in Community Ratings (and not getting additional points for being rated by community). This one is faster though it does not solves the root problem.
- Creating additional rating that would be calculated automatically based on factors that you mentioned, like number of images, length of description, etc. It would have significant impact on overall rating of a model.