This is a checkpoint that was used as a material when merging realistic checkpoints.
You can express your skills in a variety of styles based on the prompts, so show off your skills!
If you want to use more checkpoints or chat about checkpoints with me, please visit Patreon!
https://www.patreon.com/badbro
Description
This is the first version.
Realistic checkpoints are my main focus, so I didn't do a lot of research. Find hidden things
FAQ
Comments (5)
love the style of the checkpoint, but seems to have some issues with anatomy?
Nvm, after more testing, it seems to be more akin to an SDXL model than a Pony model. SDXL tags and LoRAs work better than Pony ones, and it does not have the nsfw capabilities of Pony models.
Great model, compatible with both SDXL and Pony LoRAs and tags. Model is somewhat a hybrid between SDXL and Pony, but Unet leans much more towards SDXL and does not have the NSFW comprehension of Pony models, kind of like Ponymagine and AstolfoMix. Has issues generating hands but love it regardless.
Weight Differences with other popular base models:
- 1.6% (Kohaku Epsilon Rev3, Rev3 is on HF)
- 1.8% (Animagine v3.1)
- 5.0% (AutismMix Confetti)
- 5.40% (AutismMix Pony6)
- 5.45% (Pony v6)
Oh, isn't that amazing? You seem very professional.
Yup, it's extremely rare for models to be compatible with both SDXL and Pony LoRAs and tags, i.e. "score_" and "source_" tags for Pony, "beautiful art, ultra-detailed, (8k), (HDR), (sharp focus), (intricate)" and the like for SDXL. Main shortcoming of your badbro_xl_pastel model is simply anatomy, whereas the good points are that it isn't overtrained on a certain style and it can be changed easily with pure prompting, as well as the LoRA compatibility with both SDXL and Pony LoRAs as mentioned above.
The UNet of Pony v6 is severely overtrained on the tags added by PurpleSmartAI and will result in pure noise being returned when certain keywords/phrases are used, some of which are commonly used in prompts for SDXL models, which is the reason behind LoRA incompatibility.
Not bashing on Pony v6 or anything to be clear, I use both Pony and SDXL models equally. SDXL models tend to perform better with scenery / backgrounds whilst Pony models tend to have better support with regards to Anime-styled content.
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.