aleksa-codes

/

flux-ghibsky-illustration

like 209

Hi there @ftopal ,

Thank you for your interest! The images I used for training were sourced from an Instagram page that focuses on AI-generated content. The owner of that page has kept their methods quite secretive, and I believe they may have used Leonardo AI to create those images. While I’d love to share more details, the page itself is a core part of my approach, so I can’t disclose its exact name.

The training was primarily an experiment to see if I could replicate the style of that page. In the future, I plan to fine-tune the model further by selectively retraining it on its best outputs, which I’ll choose manually.

Thanks again for reaching out, and feel free to ask if you have any more questions!

Hi @aleksa-codes

Thanks for your reply! I was wondering if you are planning to provide this SD 3.5 as well. That was my reason for asking. I do like image styles and would love to see it in other architectures as well.

Hi @ftopal ,

Yes, I'm considering it now that I see people having interest in this LoRA/style. I haven't tested SD 3.5 myself yet, so I’m curious about its performance compared to Flux. Have you had the chance to experiment with SD 3.5? I’d love to hear if you think it’s noticeably better.

Also, I noticed that a new LoRA trainer for SD 3.5 was recently released on Replicate, which could streamline this process: SD 3.5 fine-tuner by lucataco.

Really depends on the use case, i know some people are into exceptional realistic images and having better human anatomy and there FLUX seems to be better but for this image style, it's hard to say. That's why i wanted to like compare and see how both of them look.

Hi @aleksa-codes ,

Did you have person images in your dataset as well? I am asking because sometimes faces are not coming out that great. I was wondering if that has something to do with the lora itself.

Hi @ftopal ,

There were no person images in the dataset. I was getting decent results of person faces during my tests, the only problem I was noticing with people was that skin usually had wounds and was bruised up, which I think comes from the style of images it was trained on. Here are some of my results with people:

Yes, close up shots/portraits work fine for faces but natural environments ends up very less detailed faces and mostly eyes are not detectable. Let me share some results:

In most cases, when the people are bit away, their face quality shrink significantly. I thought maybe it's FLUX related but trying out other LoRas showed me it's not.
One example of what I mean used with the same prompt/same seed:

With this LoRa:

with another LoRa:

I think this is tied to having little to no human in the training dataset. This also affects the quality of hands/fingers. Again mostly for medium to long shots. I sort of think it would be improved if the dataset contains more human samples. Do you plan to have like v2 version of this by any chance? :)

Hi @ftopal , sorry for the late response and thanks for your interest in the model. Yeah, I think you are totally right about it being tied to having no human in the dataset.

There were already supposed to be 2 new versions of this model, but with both I was not satisfied and thought the results were worse than with this one, so I had to revert back both times to this version. One of them, my latest attempt, had more human and their faces in the dataset. But for some reason that latest one picked up the campfire from dataset too much and always tried to add campfires everywhere 😅

So, for example, those 2 attempts were just waste of money and time on my side, since I am using Replicate to do the LoRA trainings.

I am currently unemployed and can barely afford anything, so that is the only thing stopping me from experimenting further and making the model better. I would like to take this opportunity to ask you or anyone else who sees this comment, for a recommendation to someone for a job offer/career etc. Since I have been struggling for quite some time to make my skills, and knowledge into a career and actually live off it and afford things like LoRA training, domains, and living necessities. Coding is fun and tech stuff, but I see people living their lives nicely and actually affording things compared to me, maybe I chose the wrong path.

Other than financial issues, the technical problem would be getting the actual dataset with people and faces in it. One way I can think of is like in my last attempt to generate using current model and use "synthetic" data/images with various faces as dataset for the new model. The problem there is like you mentioned if the model fails to generate quality human faces for images that are not close up shots/portraits, then we don't have a good dataset. Maybe in that case close up shots/portraits would be good enough as dataset and the model would then generate good/not blurry faces that are far away as well (medium to long shots). That is my guess, I cannot know without trying.

So yeah, I guess once I have things sorted out, I might try again making v2. Hopefully soon.

Feel free to contact me on Discord for easier and faster communication: aleksa.codes or if you do not use Discord: [email protected]