Kirstie Alley's role as Saavik in Star Trek II: TheWrath of Khan was her film debut, and she went on to have a lengthy career in film and television
One of my fav characters from the Star Trek movies. It was sad she never returned as Saavik and the role was recast in Star Trek III.
Should work full strength on most realistic models (ex. CyberRealistic). I am no longer working on improving this Lora so decided to post it. None of the source pictures were sharp or digital. Most were grainy and some were screencaps. I was trying to refine it but after 2.5 versions I gave up. Newer versions do not appear to increase any likeness to her.
Keyword: Krloal
I have tested her 'wearing' a red star trek uniform, star trek dress, star trek leotard, etc. I did tag 'stoic expression' but it seems she has that look baked in anyway even w/o it prompted.
Description
Cleaned up dataset, removed some of the worst pictures, retrained. Might be better. Then again, maybe not.
Comments (2)
am i a bad guy if i say that she doesn't even closely look like the real one?
Good model. Thank you.
I give you a star rating.
Since you are new to creating models, I will give you some useful tips from my researches done on the Stable Difusion 1.5 since 2022.
When creating a model LORA or a T.I. do not mention face expressions. Let the AI solely train on learning the face type and details.
Never give your models the actual person name, ONLY name it to something that the AI does not recognize - a fictional name. You can also mix in numbers with letters for this purpose.
Because the original AI 1.5 program has around 1996 Known celebrity faces in its data/memory and can sometimes MIX those images with your trained model. This is not good if you did a person who is 22 years old , but the AI has info of that same person when they are 50 years old. It will start to MIX pictures AND you will not get a proper looking trained model. So, name your models to
some thing the AI will never recognize or understand who it is.
.
When you create an image for testing your model, there are some words which the AI will NOT recognize when you want a Face Expression from your model. One of those words is "Stoic Expression" , as you have noted.
There are many face expressions which your AI does recognize and can create on your model - without the use of any extra trained Lora for doing this.
All you have to do is write which "special trigger word" that your AI understands.
.
Let me give you some examples below....
From my many tests done on the SD-1.5 since 2022,
I would like to suggest to you to not use the words in your prompts...
"smug" or "smirk", as sometimes seen on some image prompts,
because in 98% of the time, your A.I. will not recognize these words to create a face expression.
.
If you use No words for face expressions,
you will not get the AI to create any expressions.
It will be just a simple closed mouth and a normal looking face.
The best words to use would be....
Smile
(Smile:1.1)
Seductive smile
Smiling
(Smiling:1.1)
(Smiling showing teeth:0.9)
(Smiling showing teeth:1.1)
(Smiling showing teeth:1.2)
(Smiling showing teeth:1.3)
... the higher number value of 1.3, 1.4, 1.5 will make the mouth smiling wider and
showing more teeth.
(Anger face expression)
(Contempt face expression)
(Disgust face expression)
(Fear face expression)
(Happy face expression)
(Happiness face expression)
(Sad face expression)
(Sadness face expression)
(Surprise face expression)
Happy
Sad
Fearful
Angry
Surprised
Disgusted
Happily Surprised
Happily Disgusted
Sadly Fearful
Sadly Angry
Sadly Surprised
Sadly Disgusted
Fearfully Angry
Fearfully Surprised
Fearfully Disgusted
Cry
(Crying face expression)
try some words like . . .
afraid,
angry,
concentrate,
confused,
contempt,
desire,
disgust,
excitement,
fear,
frown,
glare,
happy,
pain,
sad,
scream,
Looks like we don't have an active mirror for this file right now.
CivArchive is a community-maintained index — we catalog mirrors that volunteers upload to HuggingFace, torrents, and other public hosts. Looks like no one has uploaded a copy of this file yet.
Some files do get recovered over time through contributions. If you're looking for this one, feel free to ask in Discord, or help preserve it if you have a copy.
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.







