PLEASE POST GENERATIONS OF MY MODELS!

I can't keep producing LoRAs if y'all don't help me get Buzz!

My T800 Terminator retrained with a fresh dataset for PonyXL.

This model was trained on a about 110 images of various sizes and angles and from various sources, but mostly 3d, movie stills, and high quality action figure photography.

I got a little train happy, testing different methods, and as a result, you get 4 versions all with their own seeminlgy strengths and weaknesses. The full Version trained at 20 epochs seems to reproduce T800's likeness most reliably, 19 epochs version seems as accurate but is more flexible, 13 epochs version produces a balanced likeness and flexibility but the likeness varies, and finally, the 10 epochs version is seemingly most flexible but the likeness extremely varies. All were tested at a strength of 1, so I encourage you to play around with each version and the strengths and see what suits your needs best

This LoRA was not trained on nsfw, neither I have I tested its nsfw capabilities. User discretion is advised.

Recommended keywords/tags:

In the event that T800 generates consistently dark and rusted - I highly recommend using the phrase ", he is made of metal, he is made of shiny chrome metal, new, clean"

Key triggers are going to be:

• T800_Endo

• Terminator robot

• a T800_Endo Terminator Robot with red eyes

• an UpgradedT800_Endo Terminator Robot with red eyes

• a T800_Endo Terminator Robot with glowing red eyes

• an UpgradedT800_Endo Terminator Robot glowing red eyes

• he is made of metal

• holding a futuristic gun

• holding two futuristic guns

Recommended helpful tags trained with the dataset

• black background

• white background

• standing

• standing in flames

• fight stance

• in a destroyed apocalyptic city

• dark skies

• dark cloudy skies

• at night

• Apocalyptic Landscape

• destroyed city

• with skulls covering the ground

• destroyed battlefield

• Aerial HK Terminator aircrafts flying in the skies (Note: Was not trained on HKs and only slightly produces their likeness)

• Fire

• Fire and smoke

• smoke and fog

• aiming at the viewer

• charred embers floating in the air

• dark smokey background

• sparks explosion in background

• energy projectile

Angles/views trained with the dataset:

• front view

• from the side

• slightly from the side

• from below

• from above

• close-up shot

• upper body shot

• three-quarters body shot

• full body shot

Source/quality tags trained with the dataset (include any combo of these in either the Positive or Negative prompt for the desired effect):

• source_3d

• source_ActionFigure

• source_DigitalArt

• source_IRL

• source_Movie

• source_Statue

• 3d

• statue

• toy

Highly recommended in negative prompt :

• human

• skin

I have had my best success when weighting them as such: ((human: 1.6, skin: 1.6))

I trained at 1024 res and highly recommend hires fix. Adetailer is a hit or miss with facial recognition, might be necessary to change up Adetailer's parameters for better Adetailer results.

If you enjoy my work, please Like/Review/Upload your generations and if you're especially interested in supporting my work you can tip me some Buzz or support me on Patreon/Discord, I accept commissions for both AI Assisted Art and LoRA model creation as well as donations to assist furthering my projects.