DIREHOWL ( PalWorld )

Are you seeking a breeding pal? Not a Lovander type? Hmm... You probably like others may prefer other species of the palworld universe. I assume you may be wishing if they was a way to make them anthropomorphized? Well, you are in luck! Let this LoRA of help bring your yiffing pal fantasies come to your vision!

• I am accepting LoRA training requests: Use Comments to suggest me the next pal to train that is not available on CIVITAI yet.

• Join Yiff_AI Discord: https://discord.gg/cgVbyatyKV

A good friend of mine recently started playing PalWorld, and he liked direhowl a lot, but there was not a single LoRA model available, so he asked me if I could train one. So I did, and my satisfaction with this LoRA would be 90%:

90% Satisfaction:

• Works perfectly well for anthropomorphism as well as for muscular bodies; depending on the LoRA models you are using.

• Perfectly nails the character's original design, especially the head.

• Keeps feet digitigrade.

• Can make humanoid hands with five fingers with claw nail.

• Can create detailed fur while maintain anatomical features.

10% Dissatisfaction:

• Sometimes, not always, you get color misalignment or total white or total black fur.

• White Mane is not consistent.

Recommended Float Value: 0.65 - 0.75

1.0:
It will perfectly get the design but image quality becomes terrible and perhaps blurry too

0.5:
It is too low as it loses the design and you end up with an OC anthro Husky.

0.65:
Based on experience with my personal Checkpoint Mix, which is an (A + B) mix of
Indigo XL Realism Beta and EchoAlpha V 5.5; I would say 0.65 is the most ideal value as long as you have the proper triggering words and tags.


Recommended prompt words:
(your prompts)......,
<this lora:0.64> ((direhow:1.6)), palworld, ((yellow eyes)), ((black fur:1.2). furry tail, white fluffy tail, husky fur pattern )), ((white fluffy mane)), ...
... (your prompts) <other lora:X><another different lora:X><again another lora:X>

Data set: I trained this lora with 5 images of direhowl I found on Google, and I assume the lack of image reference could be the reason why getting the mane in anthro form is difficult. Unfortunately, screenshots are terrible alternatives due to texts, symbols, HUD, and the scenery; therefore, every image must be either transparent, black, grey, or white.