This model is the result of a block weight merge between my realistic model TAME Pony v2.5 and DucHaiten-Pony-XL (noscore) v6.0. It is a very versatile model with a nice aesthetic. It can do a variety of styles but may need some extra emphasis in the prompt. I will likely add some custom training in a later version, but for now it is just a merge.
Settings:
IMPORTANT - Use SGM Uniform scheduler, NOT Karras!
Sampler: Euler a (25 steps) / DPM++ 2M SDE (20 steps) / DPM++ 3M SDE (25 steps) / DPM++ 2S a (14 steps)
CFG: 5
Hi.res fix: 1.5x, 7-12 steps (half of first run), 0.3, foolhardy_Remacri / Siax / Ultrasharp / NMKD Superscale / ESRGAN_4x
Note: all showcase images were generated using high.res fix but without any loras. Some of the prompts are original, but most were borrowed (usually with a few modifications) from other model showcases.
Score tags are recommended. While both of the models involved in the merge do not require score tags, I have noticed that score tags seem to improve results with this model, so I recommend to use them although they are not really necessary.
Please use this model with care and please do not post anything inappropriate in the gallery. Furthermore, I am not responsible for any misuse that may occur.
Finally, I would like to acknowledge and thank the creators of these other wonderful models, whose work I built upon.
First, the following model was a major part of this merge
DucHaiten-Pony-XL (no-score) by DucHaiten
Second, parts of these other models were already merged into TAME Pony v2.5
GODDESS of Realism by Oppkllll
CreaPrompt_Lightning_Hyper-SDXL by jice
Another Pony Realistic Merge by Error666
iCatcher Realistic by iCatcher
LEOSAM's HelloWorld XL by LEOSAM
Pony Diffusion V6 XL by PurpleSmartAI
Better ??? - Pony (Lora) by Topplok2
NightVisionXL by socalguitarist
One-Trick Pony XL by DarkDescent
Description
Version 1.0 is a block weight merge between my realistic model TAME v2.5 and DucHaiten-Pony-XL (no-score) v6.0. It took quite some testing to find the right weights, but now it has a very nice aesthetic and seems to be quite versatile. Unlike TAME, which only does realism, this one can do a variety of styles (you sometimes have to force it with extra emphasis though). I will likely do some custom training in a later version.
Settings:
IMPORTANT - Use SGM Uniform scheduler, NOT Karras!
Sampler: Euler a (25 steps) / DPM++ 2M SDE (20 steps) / DPM++ 3M SDE (25 steps) / DPM++ 2S a (14 steps)
CFG: 5
Hi.res fix: 1.5x, 7-12 steps (half of first run), 0.3, foolhardy_Remacri / Siax / Ultrasharp / NMKD Superscale / ESRGAN_4x
FAQ
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.







