CivArchive
    Boreal-FD: (Boring Reality Flux-Dev LoRA) - v2.0
    Preview undefined
    Preview undefined
    Preview undefined
    Preview undefined
    Preview undefined
    Preview undefined
    Preview undefined
    Preview undefined
    Preview undefined
    Preview undefined
    Preview undefined
    Preview undefined
    Preview undefined
    Preview undefined
    Preview undefined
    Preview undefined
    Preview undefined

    Update 09/05/24: Uploaded a new version of Boreal-FD.

    This version includes a fix on the latent shift dot issue, while still using the better trainer version that the older model used. This model has less creativity in areas due to its smaller dataset, but it behaves alright at much lower lora strength and does a decent job at flash photography.

    I am not very satisfied with how the v2 lora out came and am going to continue training a better version for it. I had been going back in forth over the past few weeks constantly tweaking training parameters across multiple trainers and datasets. I ended up using a smaller balanced dataset with single word captions on an older commit for the AI-Toolkit while update the late shift error.

    Notes on the older versions below here.

    Work in Progress This is a very early experimental lora for Flux-Dev. It uses the Boring Reality image datasets to work on shifting Flux towards more realistic images.

    Do not expect good image generations from them. They will be replaced by newer versions has better understanding on how to train Flux Models Emerges. The ideal workflow to use with them has not been fully understood yet either.

    Additional Notes These two flux-dev loras are not expected to create very good images. Many results may be overfitted, distorted, and have this slight faded dotted look for lesser known concepts.

    The 1000 step lora is more over-fitted with with distortion and lack of prompt understanding more likely to occur, but it may perform better on things like dynamic posing and skin texture.

    You will want to experiment between the two loras, tweaking the lora strengths between 0.5-2.0 and guidance between 3.0-5.0 along with testing many different seeds.

    As more understanding develops for Flux, better workflows for these current models will come along as well as newer Boreal-FD versions as the training improves.

    Description

    This Boring Reality LoRA version behaves a bit differently than the previous v0.4. It works better on night time photography and especially flash photography. It is a bit less creative in scene composition at higher strengths. As it is much more overtrained, the strength should be kept much lower. It performs better at making skin more realistic on low strength generations without changing the original Flux-Dev Layout.

    FAQ

    Comments (28)

    peterkickasspeterSep 5, 2024· 5 reactions
    CivitAI

    THE BOSS IS BACK

    kudzueye
    Author
    Sep 5, 2024· 2 reactions

    I owe you a lot of thanks for figuring out how to improve flux. I had not initially considered the longer prompt approaches as helping the model focus in on a better image. Previously in SDXL and what not, the additional tokens would cause it to blend into a generic image. Also the Flicker tags. I had thought that would have biased the images to more professional photos, but it seems to help.

    peterkickasspeterSep 5, 2024· 1 reaction

    Oh don't worry about it bro. You are always the king of boring pictures. Your sdxl lora is what got me inspired so I'm the one who should thank you

    peterkickasspeterSep 6, 2024· 1 reaction

    thank you for the tip. you are the best

    TopazStudioSep 6, 2024· 3 reactions
    CivitAI

    the samples are so amazingly ordinary i love it!!

    GodAlMightySep 6, 2024· 17 reactions
    CivitAI

    God loves boring things. He is so superior to everything that he often finds himself bored. He learned to love it. Thank you for allowing the world to create more boredom. Make it even more boring in the next version.

    GodAlMightySep 7, 2024· 3 reactions

    Second degree humour doesn't seem to be everyone's forte.

    cdrbondjamesbond587Sep 10, 2024· 2 reactions

    God's humour isn't lost on me. Love it!

    GodAlMightySep 10, 2024· 1 reaction

    @cdrbondjamesbond587 Finally someone who believes in God!

    Ash_LoveSep 12, 2024

    Oh.

    My.

    God

    makmuzyka469Sep 29, 2024

    wtf!!!!!!!!

    greentheorySep 6, 2024· 9 reactions
    CivitAI

    Lol epic prompt:

    "Flikr photo from 2008. A grandmother with a disappointed face is peeking into a bedroom from behind the door on the left side of the photo. On the right are two guys licking their greasing hands while watching a anime on their tv. The two guys are lounging on a disgusting couch. The room is a cluttered mess with food and trash everwhere. Jesus crucifiction on the wall and bunch of anime posters. flikr photo from 2009", "t5xxl": "Flikr photo from 2008. A grandmother with a disappointed face is peeking into a bedroom from behind the door on the left side of the photo. On the right are two guys licking their greasing hands while watching a anime on their tv. The two guys are lounging on a disgusting couch. The room is a cluttered mess with food and trash everwhere. Jesus crucifiction on the wall and bunch of anime posters. flikr photo from 2009

    shubasaur877Sep 6, 2024
    CivitAI

    really liking this one

    howellmannSep 6, 2024
    CivitAI

    Insane! Works great with FluxFusionDSQ6 and euler/beta

    UnpredictpofessionSep 9, 2024
    CivitAI

    OMG!

    Ash_LoveSep 12, 2024· 4 reactions
    CivitAI

    This is the most original thing I've seen in ai art.

    Even outside! Looks more real than the real itself!

    I'm impressed :)

    Norby123Sep 12, 2024· 1 reaction
    CivitAI

    holy hell, this looks amazing

    emmanuefidegnon1735Sep 28, 2024· 2 reactions
    CivitAI

    BEST !!

    aileverager555Sep 29, 2024

    more than amateur photography lora?

    750880Oct 15, 2024
    CivitAI

    why did you remove Boreal-Effect?

    hadengJan 7, 2025· 2 reactions
    CivitAI

    How does it work with character loras?
    Anybody experimented?

    AneleMelchiorreJan 31, 2025· 6 reactions
    CivitAI

    Any update?

    kudzueye
    Author
    Jan 31, 2025· 3 reactions

    I have been experimenting a lot more training on Hunyuan video models lately. Some of the early stuff is promising and may perform better on single images than Flux dev. I will probably release some LoRA related to that perhaps as soon as this Sunday.

    leonpwork17317Feb 4, 2025

    @kudzueye Would love to see!

    BlairBrooksBBMar 9, 2025· 2 reactions
    CivitAI

    This is the best of the realistic Flux based LORAs that I've used so far, as this one is giving me the most realistic images that I've generated to date. The other ones are good too, but not like this one. I've actually produced some of the most realistic images that I've ever produced in my life. I fooled myself with some of these. I'm not going back to the other ones.

    The only thing that's left is to start using the video version of this LORA and really fool people. LOL!

    jackhestarpurnamaMay 7, 2025· 1 reaction
    CivitAI

    nice and can be upgraded

    Enki_v_NibiruDec 23, 2025
    CivitAI

    By far the best reality lora I've ever used, and I've used quite a few. Excellent work. Also works well with character loras.

    idasjoeberg350Feb 4, 2026

    hey enki which settings did you use ? Kind regards