I adjusted the art style of the Omega version to maintain the consistency of Qtea. It can still output large pictures and stable chibi style. I have replaced the uploaded files and display pictures. The uploaded files are indeed the original pictures. I don’t know why the generated data cannot be detected, but the prompts used are completely consistent with the previous display pictures.
Negative prompt: verybadimagenegative_v1.3
Omega, this is the final version of QteaMix, which is an upgrade of the model. It is trained on new materials and fused with corrections. It can also adapt to various resolutions (even up to 1280*1280) and ratios while maintaining sufficient understanding of prompt words. It can easily generate male characters through 1boy.
Compared with previous versions, Omega’s lines are more prominent and unnecessary highlights have been eliminated. It is more flattened and requires less negative prompt words. Even without any negative prompts, it performs better than previous versions.
At the same time, QteaMix can also serve as a qualified training bottom model for chibi-style training sets. Considering that the garbage data in the model has no effect on the output results, it has been streamlined and only half-precision versions have been uploaded.
If you prefer the old version’s style, it is recommended to download Omega version.
QteaMix is now complete and will not be updated anymore
Note: Omicron is not an upgrade iteration of the model, but should be seen as another model.
Regarding omicron, it is more stable than the gamma version, not easy to get off Q, and can adapt to more resolutions, but at the cost of losing the ability of 1boy + male to produce men, please refer to the comparison chart for specific style differences.
In particular, omicron presents two different styles when using easynegative or verybadimagenegative_v1.3, the former has a more delicate tone, the latter has more prominent lines, in general, using the style of verybadimagenegative_v1.3 will be more suitable for the Q version, please try it yourself.
The following content is from machine translation: the general model of the quaternary fusion chibi version, with less detail and a flatter style, adapted to expression materials, standing cards, and character chibi version, more general. Usually it can be achieved simply by adding "chibi", or even without it. The recommended native resolution does not exceed 960
There is no difference between the two versions, if you do not want to take up too much storage space, it is recommended to use the half-precision version.
You can also find this model on the hugging face
You are free to use the model, including for product manufacturing, but you cannot sell the model itself for profit. The model is open and free.