Latest page update: Updated recommended workflow again to a version 2, which is slightly even better.
My recommended local ComfyUI workflow for best results:
Trigger (must include in prompt): early 2010s snapshot photo captured with a phone and uploaded to facebook, featuring dynamic natural lighting, and a neutral white color balance with washed out colors, picturing
Please note that the online generator will NOT give you as good results as running the model locally in ComfyUI, because the online generator does not have the recommended sampler and scheduler and other customization options provided by my recommended local ComfyUI workflow!
Donations: If you want to support me financially, you can donate here: https://ko-fi.com/aicharacters
FINAL FINAL version notes:
I guess I lied and I am still updating this model after all
noticeably improved style likeness again, improved style consistency massively
new recommended inference settings
LoRa description:
This LoRa model was designed to generate images that look more "real" and less "artificially AI-generated by FLUX". It achieves this by making the natural and artificial lighting look more real and making the bokeh less strong. It also adds details to every part of the image, including skin, eyes, hair, and foliage. It also aims to reduce somewhat the common "FLUX chin and skin" and other such issues.
Training Workflow:
18 images at 1 repeat. ChatGPT generated captions. Trained on the default FLUX.1-dev model.
Disclaimer:
I do not claim to be the current owner, original creator of, person or legal representative of a person depicted in, any of the concepts that my models, including this one, aim to emulate nor the training data used to train this model. I do not claim that any of my models have received an endorsement by those respective individuals. I also do not claim that the emulation attempted by my models represents a 100% accurate depiction of the original concept in question nor do I claim that the quality of my emulation reaches the same quality of the original concept. All credit goes to the respective current owners, original creators, or people and I encourage you to support them in any way you can.
If you are the current owner, original creator of, person or legal representative of a person depicted in, any of the concepts that any of my models aims to emulate and would like that specific model to be removed, then please leave me a private message here or on Reddit with proof of authenticity of your identity and claim and I will remove it.
Additionally, I do not endorse my models being used to violate the law or enact still legal, but immoral acts such as the spread of misinformation using deep fakes.
Description
fixed oversaturation
improved skin further
much improved image coherency
slightly better prompt following
FAQ
Comments (15)
Had to leave a review - this is the best realism flux lora out right now. Never stop working on it, love it!
Thank you! Though I would argue some other models like Amateur Photography or UltraRealisticLoRaProject are situationally better/more realistic depending on the prompt.
Simply amazing...finally an image gen model that results in images that cross over to the real side of the Uncanny Valley
Anyway to run this with fp16? Or retrain?
Why? There is no functional difference between FP8 and FP16 in terms of model output.
@AI_Characters incompatible with some of workflows and sites I want to use to make the images they wont let me use the model at fp8
@topgai Ok for my future updates I'll upload float, fp16 and fp8 versions.
@topgai Nvm sry I cant. Its too much work with having to upload 3 separate versions each time.
@AI_Characters no worries bro. thank you for trying. anyway I can train it myself somehow?
fp16 would be nice, trying to use it on replicate and I can't change it to fp8
great work, i like it a lot. It works well with forge too, thanks.
Question, in your workflow, there are 3 ksampler nodes with individual steps.
When you talk about the recommended step counts. Should we change it for all of them?
For example, by default they're all at 20, if I wanted to use 50 steps, do I change all the ksampler nodes to 50 steps? Or just the "main"/first one
Uhm, no idea tbh. I think first one should be enough since the latter two are just higher res refining really.
Any chance of modifying the workflow to work with flux inpainting? I had a crack at it, but it didn't output anything useable :(
Nah sry. Not playing around with that model. I am also right now busy with testing HiDream and other things.
Details
Files
Available On (2 platforms)
Same model published on other platforms. May have additional downloads or version variants.