Tried this model myself and got some nice looking results. (Scheduler: Euler Ancestral, CFG: 5, Steps: 50).
No use of any LoRA's. The model seems just fine to me.

i find sometimes, some models don't like a ton of weird negatives, and you have quite a few if those are the ones you used. I'm trying this model now so maybe i'm wrong. But I would try removing or cleaning up the negatives you have.

unnatural body without any imperfections, non-natural leather without wrinkles and pigmentation defects, main subject have noticeable anatomical issues, picture have no strong emotions and lacks of drama, calm peaceful atmosphere, picture doesn't show any pleasant visuals or intricate details and image is quite boring and have bad composition, main subject is not obvious, image doesn't follow the rule of thirds BREAK ziprealism_neg that's a lot of odd stuff in there. -------------The negatives need like 1 word. I don't think it understands things like no strong emotions. If you want peaceful, say that in the prompt. If you don't want like, cartoons, say that in the negative. Otherwise you just removed a bunch of keywords and its probably getting confused.

@frankmike I can see your point and I can expect an aesthetic effect of not understanding negatives, but I'm not expecting the hard damage like on this image... Let me check the same with another prompt and regular negatives.

Looks like the LoRas is the root cause... Will delete my posted image and post new ones.

@homoludens  i did one model and all I got were weird robots. All because of the negatives, try these, its what i'm currently using------- Low resolution,low rez,unfocused,undefined,extra arms,long neck,missing arms,low quality,mutation,mutated hands,mutated legs,unsharp, illustration,normal quality,too many legs,worst quality, low quality, normal quality, lowres, low details, oversaturated, undersaturated, overexposed, underexposed,(cartoon, 3d, 3d render, Photoshop, sketch, sketches, video game, draw, paint, painting, render, cgi, computer graphics, anime, manga, 2d art, 3d art, illustration:1.1), (canvas frame, watermark, signature, username, artist name:1.1),black and white,watermark, candles in bottles,text,OUT OF FRAME,BAD CROP,SNAPSHOT, lomo, black and white

@homoludens  now it doesn't always listen these negatives, but it seems to bump the quality . it might also be the sampler, I don't know how sensitive this one is, i've found that while slower DPM+++ SDE Karras works well on everything. I'm also not using loras. So far while the results aren't all that artful, and it looks like almost all the other models i've used, it is sharp and it is coming out.

@frankmike Most issues disappeared after I removed the Dystopian LoRas

@homoludens Remember that SD generates two images at each iteration, one from the positive prompt and one from the negative. It can be useful to try negative prompts in the positive to see if they are understood by the model and what effect they have. Elements from the negative image can also bleed into the final image in some cases, because, loosely speaking, the presence of elements in the negative may guide the positive / final result in directions that it wouldn't have gone were they not in the negative image in the first place.

@nickfli121 That's exactly what I tried to test by my more "organic/natural" negatives.

https://civitai.com/models/460013?modelVersionId=511957
I've created the merge-fix of CinEro XL v1.5 with HillSide v1.0.

Let me know if you don't want this merge to exist.