The old was gold 100%

Hate to "overreact" but let's be honest - it's been wrecked!...I just don't understand the thinking behind the update,it didn't need improving...why would anyone make it a 100 times worse?...I pray that they listen & revert to the old system.

There are no sides... some (me) prefer the old model because of the consistency that existed in it, and because of "projects" that already existed and would need a sequel, other people were lucky with the current model and style, it is a free platform, unfortunately it seems that the time of S.D 1.5 will not return, and for that reason, I will use other sites that I like better. But I can deny that it is sad D:

For me, the new image model just doesn't work at all, even wait times are horrible. And the result is unfortunately not worth that wait to me.

Does anyone know what was used as the old model? I would love to be able to continue using that in any way.

Wish we'd been told it was going to get updated for the worse - I would've "stockpiled" a load of images I was happily creating on the previous version - can't see me using this at all anymore.

You know, a lot of the trouble could have been avoided if people would simply read the notes about the change and have a little patience. It says right in there that the model is still training, so it literally doesn't have all the data needed to make images similar to the old one, yet. It's starting at base settings and parameters, so expect a lot of strange variance and "sameness".

For those wondering why we can't have the old one as an option still, the answer is simple: memory (RAM). These image models don't spin up instantly, so they need to remain loaded in memory in order to field requests. To have the old one running at the same time would mean effectively doubling the resources required when the dev is already running on a shoestring budget supported entirely by ad revenue. It's either one running or the other. In order to train the new model, it must be running, which means the old one can't run at the same time.

At first i didn't like the update but now its better its just slower and the artsyle i used to use is too different but art styles that used to look like crap now look great. the only downside is that its slower so while images are more likely to be good in my first try if they are not its more annoying. on the old one i had to make like 100 images before there was a good one and it can do things that used to just be impossible or looked totally weird

I really liked the old model, even though it took me a LONG time to fine tune and get the results I liked.

I'm really having fun with the new model and will continue to play with it.

The differences that I currently find lacking with the new model are:

1. The lack of artist styles recognition.
2. a certain generic default e-girl sameness look.. which is made worse by #1.
3. multiple images with the same prompt can look extremely and randomly different (paradoxically, considering #2... but also made worse by #1).

I think (hope) these can/will be improved with time, model training, and prompt tuning.