Credit to impossiblebearcl4060 on CivitAI
https://civitai.com/models/67427/star-trek-tng-uniformscaptains-variant-update
Author Comments:
There is already a good LORA for composite star trek uniforms, this one is created to get the exact look of TNG ones.
The brains of this LORA was ptdurnvcoedqegsdnw. It was his idea to do this, and he gave me very important hints on what to fix, so hats off to him.
Update V3:
-Isnt overcooked this time, should work much better with loras, and is able to generate images without highres fix
-Changed the trigger for later seasons to s3stngunf, because the word "suit" in the trigger phrase was sometimes causing problems. When using this trigger always try to add "(zipperline:2)" into the negative, as it keeps adding the zipperline from the first seasons version. Also remember, if you are not getting the unfiorm for any version, then add the strength to the trigger (like "(s3stngunf red uniform:1.3)") It should help you with more complex prompts.
-As you can notice from the example images, some characters are baked in to this version. The triggers are pckd for Pickard, wrkr for Riker,bcrs for Crusher and ltyr for Tasha Yar. Remember that this is only a side effect, and the main focus here are the uniforms.
UPDATE V2
-Added Captains Variant. The iconic version from Darmok. The trigger phrase is "cptvrn jacket". The jacket is red by default, but by adding a color before the phrase you can get for example a yellow jacket.
-Removed the color code words ("science" etc). Now for trekngs1 and s3sttngsuit you have to add a desired color before the trigger phrase.
-This version is trained in 768,768 which improved the quality of rankpips
-Should be much more flexible. You can now prompt for something like "on enterprise","on enterprise starship","sitting in a chair on eterprise bridge" etc, to get the enterprise background.
The trigger words are "trekngs1 uniform" for the version from seasons 1 and 2 and "s3stngunf uniform" for the look adopted in Season 3. You should add a color ("yellow","blue","red") before the trigger phrase or you will get a strange color.
I had best results with 0.7 and 0.8 weight
The LORA still sometimes generates colored pants, but this happens occasionaly, and can be fixed by prompting if needed (usually by adding strength to "black" word).
