SPDAR-R

Hello everyone who comes here,
I wanted to try my hand at creating realistic control point formations. Unfortunately, I don’t really have the necessary processing power, so I had to improvise.
This is a simple model with the addition of numerous LoRa elements that I created myself. I put a lot of effort into it, and I probably can’t do any better at the moment. I’ll certainly improve it later.

Prompt: Simple explanation (see prompt image). There are many keywords that I will provide below.
Negative: Uses embeddings like bad_pictures, BadDream, UnrealisticDream, easynegative, or negative_hand-neg.
Steps: >20
CFG Scale: 7 (but experienced)
Sampler: DPM++ 2M SDE
Size: 512x640
Highres.fix: ScuNET, Noise Reduction: 0.7, and Embed steps at 3
CLIP Skip: 1 to 3

Useful Extensions !

After Detailer | ControlNet An AE can be used to reduce brightness, like difconsistencyRAWVAE_v10LOWSafestensor Thanks to future users of this model if you share your creations. I would like to have your feedback, both positive and negative. It’s only a 1.5 after all, I’ll understand if no one wants to invest time in it.

I used version 2.0 of SPDAR-R.

It performas better with or without integrations. Even a negative prompt isn’t necessarilly useful.

I’II try to make an SDXL version someday.