Welcome to UniverseStable
Fantasy Version - max cfg 5.5
Version 3.0 - better detail, gorgeous backgrounds.
Ladies and gentlemen, this model was merged for no particular purpose, but as an experiment and to enjoy generating images.
This model is mainly designed for artistic portraits and full-length photos. Photorealistic landscapes, monsters and more, this model is versatile for any request.
Recommend: DPM++2M SDE Karras, Clip skip: 1,2,3 Sampler, Steps: 25-35, CFG scale: 5-7
A merge model from a dozen diverse models, with the expectation of high-quality and interesting generations.
I will be glad and grateful for the placement of generations using this model.
Description
FAQ
Comments (12)
I'm confused. Is each model just an update to the previous one or are they all different for different uses? Thanks..
Different...
pruned version plz
whats the point if there's no option of selling images?
Since the model uses data from other developers with the label not for sale. This model is also not for generating images for sale.
Honestly, i dont rly get it- how will they even know?
@x1234 Perhaps special images with a special unique token are trained in the models. However, I don’t know how to check the image itself in which model it was made.
@wildzzz i guess, but i figure using ti, loras / lohas on top of that that alter the style or push the style outside the boundaries should be enough not not even be recognizable
isnt there an invisible watermark that gets embedded?
@DukeNukem47 havent seen a source on that more than the simple invisible watermark around edges that you can just crop - i might be wrong, unsure on the newest models
how did you know that?
@x1234 All the same, I think when creating a model, an image is used with a certain unique token of the type m57tcac5978et, when used, when generated, for example, an inscription will be displayed, I don’t think something lorma can be interrupted ...