Just a merge of various models and some detailer lora's, aimed to make a realistic 3D look as in the sample pictures.

Models used for merging are anillustrious_v10VAE, illustrij_v50, korone3dmixIll_v15, bubbleHentai_v10, knkLuminai_v10. Detailers used StS-Illustrious-Detail-Slider-v1.0, zy_Detailed_Backgrounds_v1 (last one for pony but somehow works).

Sample images are created and upscaled with the same model. Initial generation 40 steps, cfg 5.5, euler exponential. Upscaled with checkpoint, image to image, 35 steps, no tiled decode, denoise .38, cfg 4, euler exponential, with prompt info (SD Prompt reader and SD upscale). All generation and merging was done in ComfyUI.

Sample images have the same stable prompt and a bunch of wildcards. Depending on the style you look for, it might work well with this as stable prompt: "HDR, 8K, high contrast, depth of field, focused subject, , best quality, masterpiece, BREAK [...wildcards/environment/subject descriptions, etc.,...], BREAK vivid, bright, depth of field, focused subject, best quality, masterpiece, realistic, highly detailed.

Works with various resolution like 1024*1024, 1216*832 (and vice versa), 896*1152 (but horizontal in this resolution not very good imo) and probably other resolutions.

This model is not intended for sale/commercial use of the model as I'm not the creator of the original models, it's just a merge. What you do with the generated images is up to you.

Please share your creations below!