This lora is an attempt to recreate the Niji 6 personalization qkqmokt style on mj. The dataset is ~150 images with that style, with an emphasis on sharp line modern anime and often futuristic cybernetic subjects. This is my first lora, so sometimes it gets things wrong. The settings which have worked well for me, are as follows:Lora strength: 1.2Flux DevSampler: deis or unipcScheduler: beta25 stepsFlux guidance: 3.5My current favorite lora combination:flux_dev -&gt; FLUX meets Midjourney V6.1 lora (0.5) -&gt; MJanime_Flux_LoRa_v3_Final lora (0.5) -&gt; Niji qkqmokt (this lora) (1.0)IN NO EVENT SHALL civitai426 BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH USE OF THIS MODEL

Niji qkqmokt

Training: 1750 steps, no captions other than trigger words.Recommended lora strength: 1.2

Files with SHA256: 1be43d4f464edbed8cab52f643c0ee01870da7876de099d235e3594eba30cfcf