this looks awesome, can't wait to try it

This looks fun I will try it later. I bet this would be wild with pokemon loras.

Why "BREAK" in prompts?

Doesnt seem to like me using CLIP 2, tends to end up a single coloured image :S

Of the tags on the examples, one is rather key to getting the specific effect that this model is good at:

• detailed_eyes — required for this model's amazing eyes, but tends to encourage closeups of faces (I haven't yet experimented with adding it only at the hi-res/face detailing stage).

Others seem less key:

• adorable — the girls typically are even without this, and note that using it significantly reduces variation in face shape etc. across seeds. So adding it is basically insisting on peak adorability. (I haven't tried negating it.)
• source_cartoon — source_anime also works well, but you get a slightly more anime face shape and the eyes seem to glow more. Removing this entirely increases face-shape variation.
• semi realistic — altering this to realistic or photorealistic have relatively little effect. Altering it to photograph or traditional media can have slight interesting effects (the eyes can get even better, clothes rendering may look more or less realistic), but they're still pretty small: the model doesn't actually look like a photograph or an oil painting, it still looks semi-realistic, like a high-quality render (I haven't tried applying style LoRAs for this).

This model is also very fond of bokeh; you can reduce this by adding bokeh as a negative prompt at some strength. and the resulting less-blurred backgrounds are often pretty good. However, it's hard to eliminate the bokeh entirely this way, even with a negative (bokeh:2.5): at this point it's starting to affect the content, and while the blur is finally gone, now the backgrounds look hand-painted not rendered, and still not as sharp as the subject (High-res would likely help here.)

In ComfyUI, this model doesn't need a "CLIP Set Last Layer" node (or you can use one set to -2, which has the same effect as not having one).

Which resolution is recommended?

I'm loving the look of this one. Is there any chance this might become available for onsite generator use?

nice!

I think this is going to be my favorite checkpoint from now on

Why did you decide to go with Lightning instead of Turbo?

If I may ask :)

hi, I loved this model its my fav but it seem is no longer available in tensor, will it come back? And if no can I re-uploaded as a reprint?

What makes this model suddenly stop genning, and only produce solid blue pictures?

Did something happen to on-site creation?Wont let me pick it

What hi-res settings and upscaler would you recommend for this checkpoint?

what happened to this awesome model?

Last time Tunix was taken from the generator, it was only a few days, guess it's gone for good now. RIP to one of my favorite models. :'(

can you umm bring it back pretty peas?

podrias volver a traerlo porfa queria hacer unas imagenes

Try using Euler SGMUniform sampler, it produces stunning results with this model! You can also make the prompts very long with a lot of random stuff, it will find it's way

How much Buzz for v2? There are a few issues I've been facing I'd love ironed out. I adore this model.

Haven't even used it yet but already amazed by the detail and style. Great work

Any chance for a Illustrious version?

I like this checkpoint, very good to use with LORAS 3D, this checkpoint will replace the ones I used for my LORA models

We, the people, deserve an updated Illustrious version of this immaculate and incredible checkpoint.

Thanks in advance!

with changes coming to this platform for uk, would love for this model to be implemented on tensor.art so i can continue to create artwork from it at a later date.

Nice