File size: 2,807 Bytes
dd8a686 f0cb6c9 48f9484 2f2dcdb 27befd0 2f2dcdb 22706b0 2f2dcdb e3275b7 db99474 5f8c854 db99474 5f8c854 db99474 5f8c854 db99474 5f8c854 db99474 5f8c854 90d4fc8 2f2dcdb 55ffda1 6acf9f8 4a66990 5e76631 f0cb6c9 22706b0 f0cb6c9 55ffda1 f0cb6c9 22706b0 f0cb6c9 55ffda1 fe2d790 f0cb6c9 22706b0 f0cb6c9 cd2cd2c d30a723 375f8b4 0f2d693 f0cb6c9 9308cc0 6c92cb9 3ec07c2 19de83a f0cb6c9 fe56506 6970361 fe56506 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 |
---
license: wtfpl
---
**Swizz8-V1** (SD 1.5) = semirealistic 2.5D painter. For Prompt Gamers. 360 degree looping illumination patch. Baked in VAE, FP16. (<span style="color:red">Deleted</span>)
**Swizz8-V2** (SD 1.5) = photorealistic 2.5D painter. Prompt Gamers and Prompt Engineers. Baked in VAE, FP16. (<span style="color:red">Deleted</span>)
**Swizz8-V3** (SD 1.5) = 2.5D + 3D painter. Only for profs (Inpainting, ControlNet, Upscale). 3 quality patches to immitate SDXL or MJ. Baked in VAE, FP16.
**Swizz8-ART** (SD 1.5) = high quality fantasy, digital art checkpoint merge. Only for Prompt Engineers. Baked in VAE, FP16. The Golden CFG is 4.000 (ComfyUI)
--- License ---
Selling generated AI art? <span style="color:green">Allowed</span>
Merge this model and share? <span style="color:green">Allowed</span>
Not crediting the model creator? <span style="color:green">Allowed</span>
Run on commercial AI service? <span style="color:green">Allowed</span>
Have different permissions when sharing merges? <span style="color:green">Allowed</span>
Open Source checkpoint merge Swizz8-ART (a semirealistic SD 1.5 painter model) with out of the box high-end results in the first generation.
It is SFW and does not show (illegal) pornography but sometimes normal nudity. To avoid nudity just put these two terms in negative prompt: nude, nudity
Recommended age for use: 13 years.
+++ All Swizz-8 models are Low CFG
Lives only in the ComfyUI / A1111 parameter world of 0.500, 1.000, 1.500, 2.000, 2.500, 3.000, 3.500 - maximum 4.000 otherwise you will see only artifacts / crap.
+++ Most competent lighting concept of all existing SD 1.5 models. 360 degree looping illumination (competitive to SDXL 1.0). (Only V1)
No lighting LoRA s needed. Lights from all sides if neccessary. Dark is really dark and bright is really bright - at the same time.
Fixes the problem that (tiled) upscaled images are to bright.
+++ Massiv faces, eyes, hairs, skin variations. (Only V1, V2 and Swizz8-REAL)
Can create 99999999999999999999 different face variations in ComfyUI (random seed, no specific prompt). Fixes the problem that faces, hair colors are always the same.
Very good eyes, hairs, good skin texture and extreme beautiful faces (cutiepie, kawaii) possible or ordinary ppl. Hands are still a problem, but i have a idea.
+++ All Swizz8 models are (very) prompt-friendly.
One or two tokens (words) are enough to create a masterpiece. Most (if not all) prompts will give a good or a very good output.
And many other improvements. Soon more. Version 3 is more difficult.
Example 512 x 512 first generations (only a few SD 1.5 models offer this quality as first generation):
V2
<img src="https://void.cat/d/YLWqhXshRmhcvu9Afji2as">
---
V2
<img src="https://void.cat/d/FF5ZvVipvY7KH3C3QtMSqd">
---
|