3rd image shouldn't be you then if I understand correctly
3rd image shouldn't be you then if I understand correctly
Fyep
Fthis is just a man prompted
Dif you wanna investigate, I'd just use "man" and "ohwx man" as X on a grid, and checkpoints on Y. 4500 is a lot and maybe you are using stronger parameters than I, and it worked on the lower checkpoints? just a guess
Fcurrently i am testing on 2250 steps, 150 epoch
AHmmm, interesting. A question about the fine-tuning of checkpoints: can I “retrain” a checkpoint without losing the quality of existing people? Is everything re-evaluated when this is done? Thanks in advance!
AWith checkpoint finetuning, is it possible to have more than one "named" trained person alongside in a render image? if I do this within one lora, all people look the same^^
CI dont believe that is possible unless you are training the checkpoint on a new concept/class, if you want to train a new person/man/woman etc it will overpower existing data. The reg image experiments they are trying above is to attempt to stop the same person issue, but so far no one has been able to solve it and it would ultimately just be training new generic people not retaining existing people
ASo in this case I have to add all persons at once?
A(or objects, etc?)
CI have only seen people succeed with multiple trained subjects by either having them both appear in the training data images (as in both in the same picture) or by inpainting
AThanks, this explains a lot.
AIt is not really a problem, I need them in Blender, so my workflow is still ImagesToPlanes with png's^^
Cyou need them in blender?
AHave only an old 3080 mobile here, (~3070 desktop) but with 16GB VRam.
Chow are you running blender on that?
Csolving this is the holy grail problem for flux at the moment, I don't know if there is a case to be made for woman + man training being distinct enough classes for it to work, but the same class always seems to fail
Cflux isn't trainable, keep that in mind
Ci ran blender on a 3gb 1060 for years, wireframe or solid mode in viewport are fine enough
Ayep - I do a lot of technical presentations and I am using some of our staff to explain our systems 
Cwings3d might do just as good a job, and need a lot less resources than blender
AI'm using Blender now since 2018, dropped modo for it, no problems with 4k or 8k renders
Cwell i am not fussed on the definition of training in any strict sense, you can input data and then reliably output based on that data, whatever you want to call it it was 'trained' to deliver results. That is just for my purposes though, whether its good bad or terrible under the hood is largely irrelevant to me I just want some cool picture for some dopamine
Cyou'll get 'something' - however the reason you are running into errors and issues that you can't fix with training is becase flux is not trainable. it's rigid, inflexible, while you might succeed in creating something that'll paint your face on an image, that's not what Loras are actually for, or fine tuned check points. and so most of what will happen is that the training will fail
A
C i like blender too - but i wouldn't even consider running the current version of it on something less than a 4090AWhy? I'm using Blender 4.2 and 4.3 on a daily basis.
AAnd I have scenes with 80mio+ verts ....
AIt is about knowledge + tech. Like everywhere
C80mil verts hnnnngh
AYep. in such heavy scenes I'm using camera culling. Important if you have large landscapes with graswald objects + forestation trees + alpha trees in the regions far away.
Cwe use single sided image planes for that stuff
AYep, that's alpha trees + triplanar shaders.
Fwell normally you should be able do with minimal invasion. however with FLUX it is very easy to get broken. so not working atm
Fbut with SDXL if you properly train with reg images it should work decent
Fhowever quality of FLUX is many levels better
AThanks - it seems I might have misunderstood your posts then.
AI was the opinion Dreambooth is finetuning the model itself.
AWhat is the result then if I use dreambooth instead of the lora creation?
A(Or the AI toolkit...)
AI refer to your patreon post: Kohya FLUX Fine Tuning (Full Checkpoints) Training Full Tutorial For Local Windows and Cloud RunPod and Massed Compute
New
9 hours ago
New
9 hours ago
AI was the opinion I get a new checkpoint after the training
CYou will have a new checkpoint which you can use instead of the base model you trained on, or you can extract a lora from it to use in combination with the base model, but it will still have turned every 'man' into some version of you (if you were training yourself)
Fdreambooth / fine tuning way better than lora
Fcurrently we use dreambooth tab but we are doing fine tuning
Fsince we dont use any reg images
AWill try to get it running.
CI attempted using the 24gig config file for finetune on my 4090 but it estimated around 700 hours.. is this normal?
Fnope you must be using shared vram.
Fincrease number of blocks to swap or reduce your vram usage