Testing to merge HD LoRAs that should create better details and change the composition. Recommended for HD v1: 0.4-0.5 weight. Recommended for Eye v1: 0.4, 0.5 or 0.55 weight. HD v1 has changes to quality, composition. Higher weights can create a non-closeup composition. I don't know if it help hands. Eye V1 has better eyes and increased details, it works on closeups and medium view.
Basically, use this LoRA to control the zoom of your images generated with 768x512 quality with a sharpen effect. Don't try to add too much weight as it will add many sharpen effect.
HD V1 weight: 0.4~0.69 - More details, same composition From 0.7 to 0.9 - You can get different results 1 - More details, different composition
Eye V1 weight: 0.3 weight: less importance of LoRA 0.7: too much importance of Lora. Remember Eye V1 doesn't work in full body.
HD V1++ is another version of HD V1 that makes more changes and it also needs more weight to do so. I had good results with numbers between 0.9 and 1.3, but I don't recommend any specific number. Maybe 1, 1.1 or 1.2.
Important: This LoRA (HD v1) usually zooms OUT the image, which can be good if you want to make full body. If you want to see the results generate a image and add the same seed using my LoRA to generate the image again. Using Eye v1 you can closeups and better eyes and but don't use that much weight.
Now available on CivitAI for download https://civitai.com/models/234384/hd-all-in-one-lora
This is something important I noticed: the more weight you add, the more the composition changes and the more likely your image will become distant. On the other hand, with Eye v1 you can get a close-up.
Off-topic: If you like AI music I have an AI band. Please subscribe! https://www.youtube.com/@ForTheSkies
