Verified: 10 months ago
SafeTensor
The FLUX.1 [dev] Model is licensed by Black Forest Labs. Inc. under the FLUX.1 [dev] Non-Commercial License. Copyright Black Forest Labs. Inc.
IN NO EVENT SHALL BLACK FOREST LABS, INC. 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.
PLEASE READ ALL OF THE CONTENTS BELOW BEFORE USING
I am finding the best results with DEIS / BETA combo on the ksampler
Standard shift value works fine (1.15 / 0.5)
Depending on the style you are trying to achieve you will have to play with the LORA strength, with / without the trigger word (RU883R) and some other things I will list below:
If your image is looking too flat and cartooney (and this is not what you want) I've found adding common prompts that push realism (35mm analog photograph for example) and then just upping the strength of the lora yield much better results.
Similar to bullet point one, when prompting a specific character that the base model may already know really well (see my Shrek example in the images) I found experimenting with words like surreal, but still including the 35mm photo stuff really helps push a fun stylized cartoon look but also keeps some depth to the image
I've included all of the sample parameters I used to generate all of the images I've included. I urge you to look at those and see how I achieved the results I did.
THE TAKE AWAY HERE IS, THIS IS NOT A SET STRENGTH TO 1, PRESS GO, GET PERFECT IMAGES EVERY TIME. IT WILL TAKE SOME EXPERIMENTING ON YOUR END BUT I PROMISE YOU THE RESULTS CAN BE INCREDIBLE.