v2
Improved training data, more epochs, and enhanced tags (utilized GPT-4 captioner). This version excludes Yuna puppy images, resulting in a more consistent character output.
No need to specify details like 'black and tan fur, floppy ears,' etc. Just using keywords 'Yunadogv2, dachshund' is plenty.
Yuna can now be:
Generated from essentially any camera angle!
Contexts: (walkies, road trip)
Actions, such as 'eating, playing.'
Prompting basic emotions has also been very successful! (sad, happy, etc.)
However, if anyone has tips on how to include a second character Lora in the picture without SD merging the two characters into a horrific, distorted combination, hmu!
v1
I'd recommend inclusion of these uncommon tags
Postive:
dachshund, no humans, cute, brown and black fur, floppy ears
Negative:
(mutated paws), (mutated legs), (poorly drawn face), (poorly drawn paws), (distorted face), (extra heads), (sticky-up ears), (upright ears)
"Technically she's a a dorgi."
she's called Yuna and she's very sweet and has been learning to talk with communication buttons :)
Description
Better training data, more epochs, better tags (used GPT-IV captioner), omitting Yuna puppy images (for a more consistent character output)
Shouldn't require any mention of 'dachshund, black and brown fur, mutated paws, etc.', just saying 'Yunadogv2' should be enough.
angles and actions should also work, 'side profile, rear-view, eating, playing'
Try emotions should work too!
FAQ
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.