Conceptual Photography

V3-This one goes a lot harder than V2.-Suggested distilled cfg: 4.5-6 (you can use higher regular cfg and distilled cfg without deepfrying! Im not sure why this works), suggested regular cfg 1-2.Scheduler/Sampler: Euler/Simple or Huen/Beta for forge/comfy-Suggested strength–.75-1 for Forge and regular comfy workflowsNUNCHAKU USERS: Best lora strength is .5-.7, any higher and it starts breaking down. Best scheduler sampler is DEIS/DDIM uniform . Also you still need lots of steps for it to come through at all, around 45-50. Flux Guidance node can go all the way up to 8 without deepfrying/breaking down, from my tests.Tested on flux dev fp8 in forge ui and nunchakuThis lora is trained on REAL photos with tilted views, negative space, strange/surreal lighting and composition and multiple unique, real life subjects, etc. I used a dataset of 100 of my favorites from 2005-2012, all manually captioned with detailed captions (see some of the terms i used below).  The people generated with this lora will have that mid-aughts to early 10’s style hair/clothing and feel–its a mix of about 15 different people–mostly female– so it will have bias there, but it is able to generate males of course. Also there’s no NSFW in the dataset and i haven't tested it in that regard so ymmv. There are a few photos with multiple subjects in them so an extra person might pop up in generations when you didn’t ask. Not sure how it plays with other loras you'll have to experiment. Its pretty much the same dataset as v1 and v2 but i added a bunch of tilted/dutch angle/overhead view shots and more with interesting compositions. I trained with a HUGE dim and alpha–128/128, thinking I could scale it down with kohya with imperceptible loss--but alas, the scaled down version loses much of the unique angles/compositions that were the whole point of this lora so making it worse just defeats the point. So I’m posting the monster lora. I am sorry for the size.Another thing–i have almost NO bokeh in any of the trained photos, and most of them are full body shots–so this can be used at lower strengths to reduce unwanted extreme bokeh effects as well as to get more zoomed out/full body shots.At higher strengths it is more creative, but it may create extra limbs/fingers that need fixing w/inpainting. If you get deformities, lower the strength, and/or increase the steps, and/or increase the CFG a little(seems to work well up to 5-6 without getting that oversaturated flux look, you can even increase regular CFG which gives you access to the negative prompt) If all else fails you can always inpaint :) Terms i put in the captions, as with all things flux these MAY work SOME percentage of the time:<ul><li>ohwx(in all of them)</li><li>High angle / Dynamic angle</li><li>Dynamic perspective / Tilted perspective</li><li>Wide angle</li><li>Overhead view</li><li>Crooked framing</li><li>Top View / Looking down</li><li>Negative space in the top/bottom/left/right of the frame </li><li>Conceptual photo/picture</li><li>Y337 (i tagged all the very YELLOW pictures with this rare token, seems to have a slight influence)</li><li>Tilted view</li><li>Negative space (on the left/right)</li><li>Zoomed out</li><li>Exaggerated</li><li>Off-center/Asymmetrical</li><li>Satellite view</li><li>Dutch angle</li><li>Low angle</li><li>Worms eye view</li><li>Dramatic low angle</li><li>Top-down</li><li>Horizontal striped beams of light </li><li>Neon light trails</li><li>Decaying film stock / Aged film stock</li><li>Extreme closeup</li><li>Mirrored composition</li><li>Surreal</li><li>Colorful ohwx (the very colorful ones)</li></ul>This is a work in progress, download at your own risk!TRAINING INFO:100 images, detailed captionsEpoch: 141 out of 200 total (141 is clearly the best of the bunch, perfect mix of generalization vs learned traits)dim/alpha–128/128Steps: 7340 LR .00015Optimizer/LrScheduler: adamw8bit/cosineNum Repeats: 1Noise offset .05MinSnrGamma: 0Enable Bucket: true (almost none were cropped to square)Flip Augmentation: trueTrain batch size: 2Resolution: 1024

concept_photography_v3(neon_chaos).safetensors

concept_photography_v3(neon_darkness).safetensors

v3 (neon darkness)

v7 - (infrared)

v3.1 - (evil robot)

v6 - (ghost)

V2.1 - (angular v2)

v5.1 (junktopia v2)

v5 (junktopia)

v4 (monsterbeam)

V2 (angular)

V1 (love and wolves)

Files with SHA256: 19cff81b756265ccc850756002393a8c53ac2c91b08413819fac5b0c44d3c490