Wow, February 2024. A year and a half ago. This is the last time I registered here.
Tl; Dr: I have things I want to write, but they have nothing to do with AI photos. Perhaps a lot of readers are here because I was deep in stability and AI was sharing various techniques for generation. If you leave, this is understandable.
Here are some pictures I have shared since my last article:
What happened
After my last post, I was creating madness photos with SDXL overnight with several runs. I was greatly deepened in the indicators created using the selected Loras selected. I found many different styles from different styles and then I was going into the technique of mixing different Loras for interesting results.
And I made some pictures I really loved. A special SDXL Lora creator whom I have benefited Kapa_noro On Civitai: Kappa_neuro Creator Profile | Civitai – I especially loved her:
Many of Kapa_Noro’s Loras were terrific in making images in the SDXL, all of which together on different powers and other Loras. It was an amazing time for me.
If I made a picture I really like, I would take this specific gesture and then create a lot of pictures of a lot and then to see that it could be consistent with producing interesting results. Generally, yes.
Current
Floox giant Was released. It was an interruption. Oh, I am getting deep and flowing, with comfyui My machine and everything is running again. After that, Loras also started coming out for flow, so I went to Loras in Flux as I did with the SDXL.
Between SDXL and Flux, I easily produced 100,000 photos on my local machine, all had to be tried and trimmed them.
And the flow broke me. It broke me Nice.
The stories of unborn
When I started with SD1.0 and again (withPoint) SD1.5, I will see the pictures I created and what I liked looked like a true scene of the story … or a comedy or movie or the cover of an interesting book. I felt as if there were elements in this image that illuminated my mind and wrote a small story. Many of these photos did so in stable.
I came titled “Art of Innocent Tales”. I can even imagine creating a small story book to go with each picture.
But… hands.
The flow came along and, well, the hands were not so bad. Sometimes the flow also contains physical horror, but not so bad. And eyes! Very amazingly beautiful. The flow is focusing on making beautiful pose with amazing eyes and teeth.
And the moon that really looks like a moon. Some strange oval.
But… no stories. To me, the images I have made more and more look like Wapd Instagram Influxer Pose. There was any random emotions and surprises that brightened my mind.
And the non -loot brain is a terrible thing. The magic was gone. Floks showed me a world of good hands and beautiful eyes, but his recipe was dim.
I have collected some of the best indicators found on the seati and will repeatedly produce them with different types of luras. Technically: the photos were amazing for time. Structure: A type of blend, and no amount of indicating or lura or CFG can fix it. It was like settled for Floks Portrait Pose but not for the story of the story.
Between all of this, the model creator (To avoid financing and/or legalization) Adjust the training data very much. SDXL itself removed many authors and the flow itself did not produce, ah, naughty pictures. Model Hosting Site Civitai Buckled in pressure (Especially from 404 Media) And any model was considered inappropriate. The model has improved technical standards, but the models that can be more limited.
I know at the point that other models and variations are revealed, whether in SDXL or Flux or Kevin. But my brain has gone ahead. I see that many AI image creators have moved to small videos. I have no interest in small videos, so this is not a way for me.
While I just did a better part of the year to research Loros and techniques and create some cool things, at the end of the day, I was broken down and decided that it was time now that I am interested in other things that I am interested in. I can once again write about the AI image generation, perhaps even about the AI art vs. AI Slope, but now I’m going to move forward.
Take care