Type | |
Stats | 290 1,182 3.4k |
Reviews | (83) |
Published | Apr 30, 2025 |
Base Model | |
Training | Steps: 1,521 Epochs: 26 |
Usage Tips | Clip Skip: 1 Strength: 0.8 |
Trigger Words | articyborg, in the style of artics |
Hash | AutoV2 0765429E1A |
Artify´s Conceptdesigns for Flux
articyborg_V1.0
First of all a huge shoutout and thanks for @Chronoknight who helped me create this lora! His expertise in lora creation is invaluable!
Articyborg_V1.0, as the name suggests, is a concept style lora for cyborgs and robots. This lora works well with tag prompts and simple natural language, but probably the best results can be achieved with long descriptive prompts. Feel free to experiment! This Lora has been trained on a large number of high quality images I generated myself. This lora can produce quite complex cyborg concepts but is also capable to work as a world morph lora giving images a dark cybernetic look.
All provided sample images use only this lora and have the complete metadata included. They have not been enhanced (other than hiresfix) or been inpainted to purely showcase the capabilities of this lora.
Pros: pretty realistic and detailed results with humanoid (mechanical) and alien looking cyborgs and robots. Real nice futuristic weapon designs (swords, axes, spears).
Cons: This lora has not particularly been trained to look very human-like, so don´t expect great results with realistic looking skin and anatomy. This might be content of a future update.
Recommended (local) Generation Settings:
Trigger: articyborg, in the style of artics (but probably works well even without trigger)
Sampler: DEIS
Scheduler: Beta
Steps: 25-30
Distilled CFG: 3.0 – 4.0
Hiresfix:
upscaler: choose any you like
steps: 0
denoise: 0.7
upscale by: 1.5-2.0 x
Strength value: anything from 0.0 to 1.0. When used on it´s own a value of 0.8-1.0 is recommended. In combination with other loras you should go lower than that.
These are the values I used for most of the sample images, but you of course can experiment and use the ones you prefer.
As always, ... have much fun with experimenting and consider to leave a review/like, that is very much appreciated! 😊😊😊